Conquests and Rents: A Political Economy of Dictatorship and Violence in Muslim Societies 1009367498, 9781009367493
Tragically, dictatorship and civil strife have led to less developed, less democratic, and more conflict-prone contempor
228 75 5MB
English Pages 404 [405] Year 2023
Recommend Papers
File loading please wait...
Citation preview
Conquests and Rents
Tragically, dictatorship and civil strife have led to less developed, less democratic, and more conflict-prone contemporary Muslim-majority societies. Ahmed argues, however, neither Islam nor aspects of Muslim culture are the cause. Grounded in a positive political economy approach, Conquests and Rents investigates why these societies are predisposed to political violence and low levels of development. Focusing on the role of political institutions and economic rents, Ahmed argues that territories where Islam spread via military conquest developed institutions and practices impervious to democracy and more prone to civil war, while societies in non-conquered territories developed governance structures more susceptible to democracy when rents decline. Conquests and Rents introduces a novel theoretical argument, with corroborative qualitative and statistical analysis, to examine the interplay of the historical legacy of institutions from the premodern period and contemporary rent streams in Muslim-majority societies. . studies political economy and international economics. His research is interdisciplinary and has appeared in journals, such as the American Political Science Review and The Review of Economics and Statistics. Ahmed is the author of The Perils of International Capital (Cambridge University Press, ).
Published online by Cambridge University Press
POLITICAL ECONOMY OF INSTITUTIONS AND DECISIONS
Series Editors Jeffry Frieden, Harvard University John Patty, Emory University Elizabeth Maggie Penn, Emory University Founding Editors James E. Alt, Harvard University Douglass C. North, Washington University of St. Louis
Other books in the series Alberto Alesina and Howard Rosenthal, Partisan Politics, Divided Government, and the Economy Lee J. Alston, Thrainn Eggertsson and Douglass C. North, eds., Empirical Studies in Institutional Change Lee J. Alston and Joseph P. Ferrie, Southern Paternalism and the Rise of the American Welfare State: Economics, Politics, and Institutions, – James E. Alt and Kenneth Shepsle, eds., Perspectives on Positive Political Economy Josephine T. Andrews, When Majorities Fail: The Russian Parliament, – Jeffrey S. Banks and Eric A. Hanushek, eds., Modern Political Economy: Old Topics, New Directions Yoram Barzel, Economic Analysis of Property Rights, nd edition Yoram Barzel, A Theory of the State: Economic Rights, Legal Rights, and the Scope of the State Robert Bates, Beyond the Miracle of the Market: The Political Economy of Agrarian Development in Kenya Jenna Bednar, The Robust Federation: Principles of Design Charles M. Cameron, Veto Bargaining: Presidents and the Politics of Negative Power Kelly H. Chang, Appointing Central Bankers: The Politics of Monetary Policy in the United States and the European Monetary Union Tom S. Clark The Supreme Court: An Analytical History of Constitutional Decision Making Peter Cowhey and Mathew McCubbins, eds., Structure and Policy in Japan and the United States: An Institutionalist Approach Gary W. Cox, The Efficient Secret: The Cabinet and the Development of Political Parties in Victorian England Gary W. Cox, Making Votes Count: Strategic Coordination in the World’s Electoral System Gary W. Cox, Marketing Sovereign Promises: Monopoly Brokerage and the Growth of the English State Gary W. Cox and Jonathan N. Katz, Elbridge Gerry’s Salamander: The Electoral Consequences of the Reapportionment Revolution Tine De Moore, The Dilemma of the Commoners: Understanding the Use of CommonPool Resources in Long-Term Perspective Adam Dean, From Conflict to Coalition: Profit-Sharing Institutions and the Political Economy of Trade
Published online by Cambridge University Press
Mark Dincecco, Political Transformations and Public Finances: Europe, – Mark Dincecco and Massimiliano Gaetano Onorato, From Warfare to Wealth: The Military Origins of Urban Prosperity in Europe Raymond M. Duch and Randolph T. Stevenson, The Economic Vote: How Political and Economic Institutions Condition Election Results Jean Ensminger, Making a Market: The Institutional Transformation of an African Society David Epstein and Sharyn O’Halloran, Delegating Powers: A Transaction Cost Politics Approach to Policy Making under Separate Powers Kathryn Firmin-Sellers, The Transformation of Property Rights in the Gold Coast: An Empirical Study Applying Rational Choice Theory Clark C. Gibson, Politicians and Poachers: The Political Economy of Wildlife Policy in Africa Daniel W. Gingerich, Political Institutions and Party-Directed Corruption in South America Avner Greif, Institutions and the Path to the Modern Economy: Lessons from Medieval Trade Jeffrey D. Grynaviski, Partisan Bonds: Political Reputations and Legislative Accountability Stephen Haber, Armando Razo, and Noel Maurer, The Politics of Property Rights: Political Instability, Credible Commitments, and Economic Growth in Mexico, – Ron Harris, Industrializing English Law: Entrepreneurship and Business Organization, – Anna L. Harvey, Votes Without Leverage: Women in American Electoral Politics, – Shigeo Hirano and James M. Snyder, Jr. Primary Elections in the United States Murray Horn, The Political Economy of Public Administration: Institutional Choice in the Public Sector John D. Huber, Rationalizing Parliament: Legislative Institutions and Party Politics in France Sean Ingham, Rule of Multiple Majorities: A New Theory of Popular Control John E. Jackson, Jacek Klich, Krystyna Poznanska, The Political Economy of Poland’s Transition: New Firms and Reform Governments Jack Knight, Institutions and Social Conflict Michael Laver and Kenneth Shepsle, eds., Cabinet Ministers and Parliamentary Government Michael Laver and Kenneth Shepsle, eds., Making and Breaking Governments: Cabinets and Legislatures in Parliamentary Democracies Margaret Levi, Consent, Dissent, and Patriotism Brian Levy and Pablo T. Spiller, eds., Regulations, Institutions, and Commitment: Comparative Studies of Telecommunications Leif Lewin, Ideology and Strategy: A Century of Swedish Politics (English Edition) Gary Libecap, Contracting for Property Rights John Londregan, Legislative Institutions and Ideology in Chile Arthur Lupia and Mathew D. McCubbins, The Democratic Dilemma: Can Citizens Learn What They Need to Know?
Published online by Cambridge University Press
C. Mantzavinos, Individuals, Institutions, and Markets Mathew D. McCubbins and Terry Sullivan, eds., Congress: Structure and Policy Gary J. Miller, Above Politics: Bureaucratic Discretion and Credible Commitment Gary J. Miller, Managerial Dilemmas: The Political Economy of Hierarchy Ilia Murtazashvili, The Political Economy of the American Frontier Douglass C. North, Institutions, Institutional Change, and Economic Performance Elinor Ostrom, Governing the Commons: The Evolution of Institutions for Collective Action Sonal S. Pandya, Trading Spaces: Foreign Direct Investment Regulation, – John W. Patty and Elizabeth Maggie Penn, Social Choice and Legitimacy Daniel N. Posner, Institutions and Ethnic Politics in Africa J. Mark Ramseyer, Odd Markets in Japanese History: Law and Economic Growth J. Mark Ramseyer and Frances Rosenbluth, The Politics of Oligarchy: Institutional Choice in Imperial Japan Stephanie J. Rickard Spending to Win: Political Institutions, Economic Geography, and Government Subsidies Meredith Rolfe, Voter Turnout: A Social Theory of Political Participation Jean-Laurent Rosenthal, The Fruits of Revolution: Property Rights, Litigation, and French Agriculture, – Michael L. Ross, Timber Booms and Institutional Breakdown in Southeast Asia Shanker Satyanath, Globalization, Politics, and Financial Turmoil: Asia’s Banking Crisis Alberto Simpser, Why Governments and Parties Manipulate Elections: Theory, Practice, and Implications Norman Schofield, Architects of Political Change: Constitutional Quandaries and Social Choice Theory Norman Schofield and Itai Sened, Multiparty Democracy: Elections and Legislative Politics Alastair Smith, Election Timing Pablo T. Spiller and Mariano Tommasi, The Institutional Foundations of Public Policy in Argentina: A Transactions Cost Approach David Stasavage, Public Debt and the Birth of the Democratic State: France and Great Britain, – Charles Stewart III, Budget Reform Politics: The Design of the Appropriations Process in the House of Representatives, – George Tsebelis and Jeannette Money, Bicameralism Nicolas van de Walle, African Economies and the Politics of Permanent Crisis, – Georg Vanberg, The Politics of Constitutional Review in Germany Stefanie Walter, Financial Crises and the Politics of Macroeconomic Adjustments John Waterbury, Exposed to Innumerable Delusions: Public Enterprise and State Power in Egypt, India, Mexico, and Turkey David L. Weimer, ed., The Political Economy of Property Rights: Institutional Change and Credibility in the Reform of Centrally Planned Economies
Published online by Cambridge University Press
Conquests and Rents A Political Economy of Dictatorship and Violence in Muslim Societies
FAISAL Z. AHMED Princeton University
Published online by Cambridge University Press
Shaftesbury Road, Cambridge , United Kingdom One Liberty Plaza, th Floor, New York, , USA Williamstown Road, Port Melbourne, , Australia –, rd Floor, Plot , Splendor Forum, Jasola District Centre, New Delhi – , India Penang Road, #–/, Visioncrest Commercial, Singapore Cambridge University Press is part of Cambridge University Press & Assessment, a department of the University of Cambridge. We share the University’s mission to contribute to society through the pursuit of education, learning and research at the highest international levels of excellence. www.cambridge.org Information on this title: www.cambridge.org/ : ./ © Faisal Z. Ahmed This publication is in copyright. Subject to statutory exception and to the provisions of relevant collective licensing agreements, no reproduction of any part may take place without the written permission of Cambridge University Press & Assessment. First published A catalogue record for this publication is available from the British Library Library of Congress Cataloging-in-Publication Data : Ahmed, Faisal, author. : Conquests and rents : a political economy of dictatorship and violence in Muslim societies / Faisal Ahmed, Princeton University, New Jersey. : Cambridge ; New York, NY : Cambridge University Press, . | Series: Political economy of institutions and decisions | Includes bibliographical references. : (print) | (ebook) | (hardback) | (paperback) | (epub) : : Islamic countries–Politics and government. | Political violence–Islamic countries. | Dictatorship–Islamic countries. : . . (print) | . (ebook) | ./–/eng/ LC record available at https://lccn.loc.gov/ LC ebook record available at https://lccn.loc.gov/ ---- Hardback ---- Paperback Cambridge University Press & Assessment has no responsibility for the persistence or accuracy of URLs for external or third-party internet websites referred to in this publication and does not guarantee that any content on such websites is, or will remain, accurate or appropriate.
Published online by Cambridge University Press
For Saba and our daughter, Ameena
Published online by Cambridge University Press
Published online by Cambridge University Press
Contents
List of Figures List of Tables Acknowledgments
page xi xiii xv
Political Violence
Analytical Framework
The Conquest Equilibrium
Conquest Fostered Autocratization The Autocratic Legacy of Muslim Conquest
Fueling Authoritarian Resilience Aiding Political Violence
Democratic Transitions in Non-Muslim Societies
Conclusion
References Index
ix
Published online by Cambridge University Press
Published online by Cambridge University Press
Figures
. Cohesive institutions (welfare spending) is associated with greater democracy . Research design . Percentage of a modern state’s territory conquered by Muslim armies . Muslim conquest and contemporary democracy . Persistent democratic deficit in Muslim societies . Cohesiveness in Muslim and non-Muslim societies . Democracy in Muslim oil and non-oil producers . Civil war and democracy in Muslim non-oil producers . Foreign aid and civil war in Muslim non-oil producers . Democracy and public goods . Democracies exert greater tax effort . Game tree . Political violence as a function of cohesive institutions (θ) and rents (R) A Democracy in non-Muslim countries A Democracy and civil war in Muslim oil producers A Civil war and foreign aid in non-Muslim non-oil producers A Civil war and oil prices in Muslim oil producers . From Muslim conquest to contemporary political institutions . The spread of Islam by . Change in state centralization during Muslim conquest . State centralization in the premodern period . State centralization, with a “typical” (treated) and synthetic conquered territory . Muslim rule and provincial first parliaments in Medieval Spain
page xi
Published online by Cambridge University Press
xii
List of Figures
. Muslim conquest reduces a country’s accumulated democratic experience . Religiosity and democratic experience . Religiosity and accumulated democracy, by Muslimmajority status . Religiosity and democratic experience in Muslim societies, by conquest status . Democracy over time in Muslim societies, by their exposure to mamluk institutions . Democracy in oil- and non-oil-producing Muslim societies exposed to mamluk institutions . Marginal effect of oil wealth on democracy as conquest varies . Democracy in “initially nondemocratic” petrostates over time, by conquest status . Average foreign aid receipts in Muslim and non-Muslim nonoil producers . Oil price and the aid windfall to Muslim countries . Political violence in Muslim aid recipients . Differential in foreign aid and civil war incidence between Muslim and non-Muslim recipients . Aiding authoritarianism in Muslim recipients . Commercial loans and oil prices in Latin American nonoil producers . Commercial loans and democracy in Latin American nonoil producers . Trade subsidies and oil prices in Eastern European Soviet satellites . Trade subsidies and democracy in Eastern European Soviet satellites . Timing since the Neolithic Revolution and democracy in
Published online by Cambridge University Press
Tables
. Comparison of democracy, civil war, and economic well-being across Muslim and non-Muslim societies . Summary of theoretical predictions . Conquest status, rents, and variation in political violence, with country examples . Incumbent’s optimal choice in political violence . Incumbent and opposition optimal choices . Territories “exposed” to Islam by . State centralization during Muslim expansion . Robustness of state centralization during Muslim expansion . Muslim conquest and autocratization: Matched counterfactuals and competing explanations . Examining the components of state centralization . Political survival during Muslim expansion . The importance of mamluk institutions during Muslim expansion . The marginal effect of foreign interventions on state centralization . Muslim rule and first parliaments in Spanish provinces . Muslim conquest harms a country’s contemporary democratic experience . Muslim conquest lowers institutional cohesiveness . Accounting for explanations from the premodern period . Accounting for the impact of European colonialism . Accounting for explanations from the contemporary (modern) period . Tracing the conquest equilibrium since
page xiii
Published online by Cambridge University Press
xiv
List of Tables
C The long-run effect of Muslim conquest on alternate measures of democracy C Muslim conquest and fiscal capacity C Muslim conquest (alternate measures) and democracy C SLS estimates with distance to Mecca as an instrumental variable . Political stability in nondemocratic petrostates . Oil wealth harms democracy . The causal effect of oil wealth on democracy . Oil wealth does not heighten civil war in conquest petrostates . Repression in conquest oil producers . The pacifying effect of US troop deployment in conquest petrostates . Incidence of civil war in Muslim and non-Muslim non-oil producers, by periods of low and high oil prices . The causal effect of oil-price-induced foreign aid on civil war . Alternate measures of foreign aid and accounting for unobserved heterogeneity . Addressing potential violations of the exclusion restriction . The heightened impact of aid on civil war in countries exposed to mamluk institutions . Trajectory of democracy in Muslim recipients after civil war D Sample of non-oil-producing foreign aid recipients . Correlates of cohesive institutions in Muslim and non-Muslim non-oil-producing societies . Foreign transfers are associated with less democracy . An exogenous decline in transfers and political violence . The Neolithic determinants of (modern) cohesive institutions . Political violence with changes in oil prices and exogenous institutions . Rents and politics in conquest and non-conquest states since . Developmental challenges associated with rents
Published online by Cambridge University Press
Acknowledgments
As part of my freshman writing seminar at Northwestern University, Professor Edward Gibson assigned Samuel Huntington’s The Clash of Civilizations and the Remaking of World Order. In this now infamous book, Huntington proffered a seductively simple explanation for patterns of democracy – and relatedly, economic development and violence – across the world: culture. In this account, Islam was a root cause for the lack of democracy and bouts of civil strife across many contemporary Muslim-majority societies. At the time, I found Huntington’s thesis unsatisfying as I had lived in and visited several Muslim countries that did feature democratic politics – albeit often with flaws that are characteristic of many developing countries – and whose national governments were often headed by elected female leaders. My personal experiences reinforced the notion that democracy and Islam are not inherently incompatible. The question of why Islam is seemingly associated with dictatorship and violence remained with me. This book attempts to provide a positive political economy perspective to understand why so many contemporary Muslim societies (but not all) tend to exhibit dictatorship and violence. This endeavor would have not been possible without the encouragement and feedback of my collaborators, colleagues, fellow researchers, and my family. I would like to take this opportunity to briefly thank them. I owe a special debt of gratitude to my collaborators, Daniel Schwab and Eric Werker. We have worked on several papers that helped me sharpen the argument and empirics in this book. In particular, the formal model in Chapter and parts of the empirical analysis in Chapters and draw from two of our articles, “The Political Transfer Problem: How Cross-Border Financial Windfalls Affect Democracy and Civil War” (published in the Journal of Comparative Economics) and “Aid and the Rise and Fall of xv
Published online by Cambridge University Press
xvi
Acknowledgments
Conflict in the Muslim World” (published in the Quarterly Journal of Political Science). I am also grateful for financial support from Princeton University and the Centre for Islamic Studies at Oxford University. Support from the politics department at Princeton allowed me to work on the published articles on which this book draws. My time as a visiting fellow at the Centre for Islamic Studies provided a truly interdisciplinary environment to discuss the book’s ideas. I particularly enjoyed and benefited from the grilling I received when I presented the book’s historical argument to a group of Islamic historians at the Centre’s research seminar! In addition to the discussions at the Centre, I am deeply indebted to many people who have offered feedback at various stages of this book project, particularly on prior published work. Conversations with many of my Princeton colleagues helped advance the book’s argument, including those with Gary Bass, Mark Beissinger, Carles Boix, Christina Davis, Joanne Gowa, Amaney Jamal, Melissa Lee, Helen Milner, Andrew Moravcsik, Grigore PopEleches, Kris Ramsay, Jake Shapiro, James Vreeland, Leonard Wantchekon, and Deborah Yashar. I am also extremely grateful for Amaney, Jim, and Leonard’s unwavering support and mentorship during my time at Princeton. I have also benefited from insightful feedback from scholars at other institutions. They include Scott Abramson, Eric Arias, Navin Bapat, Ida Bastiaens, Lisa Blaydes, Melani Cammett, Eric Chaney, Volha Charnysh, Adam Dean, Mark Dincecco, John Gerring, Timur Kuran, Ahmet Kuru, Bogdan Popescu, Joan Ricart-Hugeut, Paasha Madhavi, Tarek Masoud, Elizabeth Nugent, Nita Rudra, Hillel Soifer, David Stasavage, Yuhua Wang, and Andreas Wimmer. In addition to these scholars, numerous others asked helpful questions and offered constructive comments at various forums over the years. These venues include presentations at the meetings of the American Political Science Association, the International Political Economy Society, and the Association for Analytic Learning about Islam and Muslim Societies, and seminars at the University of Copenhagen, University College London, University of Essex, London School of Economics, Nuffield College, Princeton University, and the Centre for Islamic Studies at Oxford University. This book also greatly benefited from a “virtual” workshop. The generous feedback from Eric Chaney, Amaney Jamal, Tarek Masoud, Paasha Mahdavi, Elizabeth Nugent, David Stasavage, and James Vreeland helped sharpen the book’s argument, organization, and overall composition. Amaney and Jim deserve an additional thank you for shepherding the workshop’s discussions. I thank the Niehaus Center for Globalization and Governance, the Bobst Center for Peace and Justice, and the Department of Politics at Princeton for the financial support to make the workshop possible. Suggestions from two anonymous reviewers further improved the manuscript. In the final stretch, the series editors – Jeff Frieden, John Patty, and Maggie Penn – pushed me to clarify and improve the exposition of the book’s theory chapter. With their feedback,
Published online by Cambridge University Press
Acknowledgments
xvii
I hope I achieved this. I also thank Robert Dreesen and the team at Cambridge University Press for a smooth publication process. Finally, I thank my family for their love and support. My parents have been instrumental in all my academic endeavors. Growing up, my sister and I were very fortunate to visit and live in many Muslim-majority countries. My interactions with people in these societies and summers visiting my extended family in Bangladesh have greatly influenced my perspective in this book. My late Uncle Tony (Zahed Hossain) played an outsized role in my intellectual development, particularly in encouraging me to be “critical” of my (Western) education and embracing my South Asian and Muslim roots. I miss our weekly conversations about our faith, the state of American and British politics, and the performance of our family’s football team, Arsenal. A special thank you goes out to my daughter, Ameena. I started writing this book in earnest a couple of months after her birth. While she napped, I would write and when awake, we would play. Her energy and laughter then (and now) elevated my spirits. At the time of writing this, Ameena is three and maybe one day she will read this book and reconsider whether her Baba’s books are still “boring” (since they have no pictures). Finally, I thank my wife, Saba. Her steadfast support through all the uncertainties of academic life and our common “nerdiness” is a blessing. I devote this book to her and our daughter.
Published online by Cambridge University Press
Published online by Cambridge University Press
PART I
A POLITICAL ECONOMY APPROACH
https://doi.org/10.1017/9781009367509.001 Published online by Cambridge University Press
https://doi.org/10.1017/9781009367509.001 Published online by Cambridge University Press
Political Violence
During the Islamic Golden Age spanning the eighth through fourteenth centuries, Muslim societies were at their zenith, supplying their scientific, literary, and cultural ideas to Europe (Huff , Kennedy , Kuru ). Today, conditions in many Muslim majority (hereon, Muslim) societies look vastly different. Stories documenting the miserable socioeconomic conditions in civil war–stricken Afghanistan, Iraq, Somalia, and Syria are routinely splashed across media outlets. And tragically, like many countries in the Muslim world (e.g., Saudi Arabia, Uzbekistan), the publics in these societies tend to be ruled by dictatorships. The prevalence of civil war and dictatorship – which frequently characterize societies afflicted with political violence – seems to be particularly elevated in contemporary Muslim societies, which in turn and may be an important factor underlying their lower levels of human and economic development, relative to non-Muslim societies (Kuran ). Motivated by these observations, this book endeavors to explain why many contemporary Muslim societies are predisposed to political violence. My
For example, this includes Latin translations of the Greek Classics and of Arabic scholarship in astronomy, mathematics, agriculture, and medicine and the “exporting” of cultural innovations in the arts, alchemy, music, and pottery. Some Muslim countries feature democratic institutions but have experienced periods of “backsliding” to dictatorship since their independence (e.g., Bangladesh and Pakistan). According to a recent comprehensive study contrasting socioeconomic conditions in Muslim to non-Muslim countries (Kuran ), the typical citizen in a member state of the Organization of Islamic Cooperation (OIC) is about percent poorer than a denizen in a non-OIC country, significantly less likely to be literate, and has a shorter life expectancy (by about seven years). More generally, lower levels of economic and human development tend to characterize societies that exhibit broader patterns of pernicious political economy (e.g., rampant government corruption, lower public spending on welfare goods). In many Muslim societies this can manifest in “crony capitalism” (Diwan, Malik, and Atiyas ).
https://doi.org/10.1017/9781009367509.002 Published online by Cambridge University Press
A Political Economy Approach
. Comparison of democracy, civil war, and economic well-being across Muslim and non-Muslim societies
TABLE
All countries
Democracy Civil war
Panel A: Pairwise correlation with per capita income Democracy Civil war (a) (a) .*** −.***
(a) Muslim . (.) . (.)
Panel B: Comparison of group means (b) (b) Non-Muslim Difference . −. (.) (.)*** . . (.) (.)***
Notes: In both panels, democracy is the normalized POLITY score (on [,] scale), where a higher value corresponds to a greater quality (level) of democracy. Civil war is the incidence of civil war with a threshold of , battle deaths per annum. Panel A reports the pairwise correlation of a country’s annual level of democracy and civil war with its annual level of per capita GDP across a global panel of countries from onward. Panel B reports the average value of democracy and civil war across Muslim-majority and non-Muslim-majority societies. Column b reports the difference in group means. The corresponding standard errors for these statistics are reported in parentheses. In both panels, *** reports the level of statistical significance is below a p-value of ..
diagnosis begins with two empirical observations. First, looking across the globe, an individual’s economic welfare tends to be lower in countries with less democratic political institutions and those experiencing civil war. Second, Muslim societies tend to be more conflict prone and less democratic relative to non-Muslim societies. Table . provides a preliminary snapshot of these two observations. The table is separated in two panels and leverages widely used and publicly available data on economic development, political institutions (democracy), and civil war from the World Bank, POLITY IV project, and PRIO/Armed Conflict Database, respectively. While not the focus of this book, I begin by drawing a correlation between political violence and economic underdevelopment. Panel
As a growing body of scholarly research demonstrates, a robust correlate of underdevelopment is dictatorship and civil war (e.g., Besley and Persson a, b, Blattman and Miguel ). Among their many ails for economic development, dictatorships tend to under-supply public goods (investments in education, health care, etc.) and do not adequately protect property rights, while civil war depletes human capital (e.g., through battle deaths, substitution from economic production to fighting) and destroys a society’s productive assets. While dictatorship and civil war are associated with economic (under) development, this book does not focus on this economic dimension. These observations are further substantiated and analyzed in Chapter .
https://doi.org/10.1017/9781009367509.002 Published online by Cambridge University Press
Political Violence
A reports the pairwise correlation between a country’s per capita GDP (as a common, catchall measure of economic and human development) and its level of democracy and incidence of civil war across a global sample of Muslim and nonMuslim societies since . Consistent with existing findings, more democratic societies enjoy higher per capita income (as indicated with the positive correlation in column a) while standards of living are negatively correlated in societies experiencing civil war (column b). Honing in on the book’s focus, panel B probes whether the typical (average) Muslim society tends to be different from non-Muslim ones across its level of democracy and conflict propensity. For instance, the typical Muslim society is two times more likely to experience high intensity civil war with at least , battle deaths each year than a non-Muslim society (. percent versus . percent, respectively). Using a conservative cutoff of . for democracy, the typical Muslim society exhibits dictatorship, with a normalized POLITY score (=.). In contrast, the typical non-Muslim society tends to be democratic. Of particular importance are the statistically significant differences in the group means reported in column b. These differences suggest the typical Muslim society is significantly less democratic and more conflict prone compared to non-Muslim countries. Motivated by these patterns, this book examines why contemporary Muslim societies are more prone to political violence (and plausibly, lower levels of economic and human development) relative to non-Muslim societies. I am by no means the first person to embark on this endeavor. The prevalence of economic and political underdevelopment in contemporary Muslim societies has sparked a lively research agenda seeking to explain why. Some scholars identify various aspects of the Islamic faith, such as its organization (Platteau ) and its underlying principles (Huntington , Lewis ), while others point to the role of Islamic law (Kuran ) and an alliance between religious elites (ulema) and the state (Kuru , Rubin ); both of which, thwarted economic and political advancements in Muslim societies. Another
The measure of democracy ranges on a [,] scale where a higher value corresponds to a higher quality of democracy. It is based on normalizing the POLITY score. The raw POLITY score ranges from − to + where a score closer to + corresponds to a higher quality of democracy. Thus, a normalized POLITY score above . corresponds to a society with a raw POLITY score greater than zero (i.e., between and ). A number of studies theorize and empirically show the negative correlations between dictatorship and development, and civil war and development. See for example, Acemoglu and Robinson (, ), Besley and Persson (a, b), Blattman and Miguel (), Bueno de Mesquita et al. (). Civil war is conceptualized as the most extreme/intensified form of civil unrest between the government and opposition groups. While the patterns in Table . suggest a correlation between dictatorship/democracy, civil war, and per capita GDP, the correlations do not necessarily imply political violence causes underdevelopment. Existing scholarship demonstrates that these relationships tend to be bi-directional and endogenous.
https://doi.org/10.1017/9781009367509.002 Published online by Cambridge University Press
A Political Economy Approach
strand of scholarship argues that various characteristics of the Muslim heartland such as its Arab culture and the high concentration of hydrocarbons are to blame (e.g., Fish , Ross ). Yet, it is plausible that factors not intrinsically nor uniquely associated with Islam (nor Muslim/Arab culture) may be the underlying source of political violence in Muslim societies. Rather, I hone in on political institutions and nontax government income (rents) – particularly foreign sources of government revenue from oil production and foreign aid receipts – as important determinants of political violence in Muslim societies. This endeavor is grounded in a positive political economy approach that advances a formal theory that is tested in a historical and contemporary setting. I argue that Muslim societies tend to be less democratic and more conflict prone on average. However, there is considerable variation within the Muslim world depending on how Islam was spread. Territories where Islam spread via military conquest developed institutions and practices that led to political regimes that made them more impervious to democracy and, in response to declines in rents, more prone to civil war. In contrast, societies in nonconquered territories – including some Muslim societies, such as Indonesia and Malaysia – developed governance structures that made them more susceptible to political (democratic) development; and where declines in rents provided opportunities for transitions to democracy. A central takeaway from this book is that neither Islam nor aspects of Muslim culture are the root causes of political violence in many contemporary Muslim societies. Rather, it is due to the interplay of two independent variables: () the path dependent effects of political institutions (and associated governing
The list of scholars cited in the previous few sentences is not exhaustive of scholarship on this question. In the past decade alone, other works include those by Blaydes (), Blaydes and Chaney (), Brownlee, Masoud, and Reynolds (), Chaney (), Elbadawi and Makdisi (), Jamal (), and Mazaheri (). Rents comprise nontax revenue that accrues to the state, for example, from domestic oil production, foreign aid, and related forms of external support. Empirically, an important feature of the rents studied in this book are their foreign sources. For example, oil rents largely stem from the sale of oil abroad. Foreign aid is the receipt of financial transfers from foreign (donor) governments. Chapter provides a more sustained discussion of rents and their effects on political and economic outcomes. For example, this positive political economy approach is similar to those developed by Acemoglu and Robinson () and Besley and Persson (). Cammett et al. () advance a political economy perspective for countries in the Middle East and North Africa (MENA). This book focuses on Muslim societies in the MENA region and those in Africa, central and south Asia. Non-conquered Muslim territories (e.g., Indonesia, Malaysia) had a type of system that made them more susceptible to economic and political development. In fact, in many instances, nonconquered territories had a likelihood to develop similar to non-Muslim territories with the same economic and social conditions. In Chapter , I show that declines in rents linked to petrodollar investments and Soviet trade subsidies facilitated to democratic transitions in Latin America and Eastern Europe, respectively.
https://doi.org/10.1017/9781009367509.002 Published online by Cambridge University Press
Political Violence
coalitions and practices) from a trans-territorial process in the distant past and () increases and decreases in various types of rents (e.g., oil revenues, foreign aid) in those societies. By linking these two independent variables – conquest and rents – to patterns of political violence, the theory and empirics advanced contributes to rich literatures in institutional economics, historical legacies, international relations, and the resource curse. In the remainder of the chapter, I outline my argument, evidence, and contributions.
. This book places political institutions and rents at the heart of studying patterns of political violence in Muslim societies. I advance a general theory that identifies two parameters to explain patterns of democracy, dictatorship, and civil war: the quality of a country’s political institutions and its prevailing level of rents. I then test the model’s predictions across Muslim and non-Muslim societies. The empirical analysis explores how the institutional legacy associated with Muslim military conquest, coupled with variation in economic rents – especially those with a foreign source – explain patterns of dictatorship and civil war since the s (a) between Muslim and non-Muslim societies and (b) over time within these Muslim societies. A Theory In Chapter , I advance an original theory linking rents and political institutions to patterns of political violence: dictatorship/democracy and civil war. Existing theories predict a monotonic effect whereby greater amounts of rents in countries can entrench dictatorship and elevate the propensity for civil war (e.g., Besley and Persson a, Ross , Smith ). These theoretical formulations, however, are less specific to the effects from a reduction in rents and thus have difficultly reconciling empirical findings where declines in rents can also foster civil war (e.g., Ahmed and Werker , Nielsen et al. ).
The theory is developed in Chapter , with predictions generated from a formal model (developed in collobarative research with Daniel Schwab and Eric Werker). In a recent edited volume following the Arab Spring, Elbaawi and Makdisi () posit the Middle East’s history of intra and interstate conflict are robustly correlated with dictatorship. Their general argument is consistent with mine, but they do not provide a coherent account for why Muslim societies are conflict prone nor do they offer an explanation why certain societies might experience conflict at any point in time. The theory and empirics advanced in this book do. Moreover, the book’s empirical analysis also looks beyond the modern Middle East to Muslim countries in central, south and southeast Asia and Africa and draw comparisons to non-Muslim societies, particularly in South America and Eastern Europe. Relatedly, Benjamin Smith () articulates a theoretical narrative through which a decline in oil wealth may lead to a government’s removal from power depending on the strength of
https://doi.org/10.1017/9781009367509.002 Published online by Cambridge University Press
A Political Economy Approach
To reconcile this tension, I advance a formal model that predicts how high and low levels of rents have differential effects on dictatorship, democracy, or civil war in a unified framework. The model builds on a theory of state capacity and political violence developed by Timothy Besley and Torsten Persson (, , a, b) and extends it to allow for the possibility that the incumbent can deter attack by the opposition when government resources (e.g., nontax revenue) are sufficiently high. The model is essentially a one-shot sequential game, and when nontax revenue is moderate, an incumbent will respond to an insurgency with only moderate violence, which encourages an attack by the opposition. However, when nontax revenue dwarfs the costs of fielding a large army the incumbent would be willing to spend enough to squash an insurgency, and the insurgents are deterred from attacking. This setup allows for the possibility for a de-escalation of violence (i.e., political stability) even as nontax revenues increase. Crucially, the model’s comparative statics show that dictatorship can be sustained with buoyant levels of rents, while a reduction in rents may foster civil war or facilitate transitions to democracy. Whether rents foster dictatorship, democratic transitions, or civil war hinge on the quality of a society’s preexisting “cohesive institutions.” Employing the terminology from Timothy Besley and Torsten Persson (, , a b), institutional cohesion refers to the share of state revenues shared between supporters of the ruling government and those in opposition. In societies with more cohesive institutions, the ruler is more egalitarian in sharing state revenues with supporters and opposition groups. As Besley and Persson observe, various factors may be associated with institutional cohesion, such as a country’s history of international conflict and societal fragmentation. Crucially, they stress that societies with more cohesive
opposition forces (prior to the decline of oil wealth). His account focuses on regime change, rather than civil war. In Chapter ’s formal model institutions are fixed and exist prior (“exogenous”) to the reception of rents. I will refer to these as “ex ante” or “preexisting” cohesive institutions. The exogeneity of cohesive institutions in the model guides my use of a historical process (Muslim conquest) that precedes rents in the modern period. In particular, I use the institutional legacy of Muslim conquest as an important correlate (determinant) of less cohesive institutions in contemporary Muslim societies. Cohesive institutions are analytically distinct from other categorizations of political institutions, such as the protection of private property rights and/or formal constraints on the leader’s political authority (e.g., Acemoglu et al. , North ). That said, empirically, more cohesive institutions are correlated with political systems that protect private rights and place greater formal constraints on the executive. On the former, Besley and Persson among others, such as Charles Tilly () argue that in order to fight rival states, national governments developed fiscal and tax institutions. This process often required “sharing” revenues with the broader population. Thus, states with longer histories of fighting interstate wars often exhibit more cohesive institutions. On the latter, societies with more diverse ethnolinguistic and/or religious groups may be less prone to
https://doi.org/10.1017/9781009367509.002 Published online by Cambridge University Press
Political Violence
institutions tend to exhibit more democratic political institutions. Conversely, societies with less cohesive institutions tend to be dictatorships. This suggests that empirically, democracy should be correlated with greater institutional cohesion; and accordingly, I will use measures of democracy to proxy the quality of more cohesive institutions. Figure . provides some empirical corroboration: Governments in countries with more democratic institutions tend to allocate a higher share of their expenditures to welfare spending. The latter is measured with a government’s spending on subsidies and transfers to the masses. With this interpretation of cohesive institutions, the model in Chapter generates three testable predictions: . Higher levels of rents sustain dictatorship, particularly in societies where its preexisting institutions are less cohesive. . A decline in the level of rents can foster civil conflict in countries where its preexisting institutions are less cohesive. . A decline in the level rents can foster transitions to democracy in countries where preexisting institutions are more cohesive. These three predictions suggest that whether a society experiences dictatorship, democracy, or civil war depends on two parameters: its quality of preexisting cohesive institutions and prevailing level of rents. Table . summarizes how political violence varies along these two parameters.
“sharing” state resource with non-group members. Empirically, scholars have observed that countries with greater societal heterogeneity exhibit less cohesive institutions. The empirical analysis in Chapter evaluates the influence of societal heterogeneity and fiscal institutions on democracy in Muslim societies (relative to non-Muslim societies). As I discuss in Chapter , the concepts of cohesive institutions and democracy are not identical. Institutional cohesion captures how a government allocates resources between its supporters and opposition. Governments in more cohesive institutional settings share state resources more equitably between these two groups. Democracy, in contrast, typically refers to the process through which leaders are chosen (e.g., voting by the masses). In practice, the selection process of leaders in democracies will induce public policies that distribute state resources to both government supporters and its opposition. Thus, democracies tend to exhibit more cohesive institutions. Using democracy as a measure (proxy) for institutional cohesion is advantageous for several reasons. First, there is wide cross-country coverage of its prevalence and its various components (e.g., constraints on the executive, type of electoral institutions, and voting procedures). Second, measures of democracies span many decades and for some countries, centuries (e.g., Western Europe). Third, there are several measures of democracy from different scholars/institutes and that use varying coding criteria (e.g., POLITY IV Project, Varieties of Democracy, measures by Boix, Miller, and Rosato). Fourth, measures of government expenditures are less comprehensive and accurate, particularly for developing countries. As a result, this makes democracy a comparatively more attractive measure. I provide additional reasoning and evidence for this correlation in Chapter .
https://doi.org/10.1017/9781009367509.002 Published online by Cambridge University Press
A Political Economy Approach
TABLE
. Summary of theoretical predictions
High level of rents Low level of rents
Less cohesive institutions
More cohesive institutions
Dictatorship Civil war
No change Democracy
30 20 10 0
Subsidies and transfers (% Expenditures)
40
Notes: “No change” implies that higher levels of rents are unlikely to alter the societies underlying regime type. Dictatorships with high levels of rents will tend to remain dictatorships, while democracies with high levels of rents will tend to remain democratic. This follows from propositions and from the formal model in Chapter .
0
0.75
FIGURE . Cohesive institutions (welfare spending) is associated with greater democracy Notes: Based on author’s calculation. The figure reports the country (average) data from – from a global sample of countries.
B Research Design Table . is helpful in clarifying how variation in preexisting cohesive institutions and the level of rents can generate differential effects on dictatorship, democracy, and civil war. Using this table as a guide, my empirical analysis explores two phenomena in Muslim societies. The first examines why many contemporary Muslim societies exhibit a stock of less cohesive institutions in the contemporary period (i.e., from onward). The second explores how, in the presence of these less cohesive institutions, Muslim societies exhibit
https://doi.org/10.1017/9781009367509.002 Published online by Cambridge University Press
Political Violence
FIGURE
. Research design
dictatorship when the level of rents are high but are more conflict prone at lower levels of rents. Of course, in investigating these phenomena, a potential worry is endogeneity, particularly the relationship between rents and institutions. For example, dictatorships may be more predisposed to search for oil while donors may disburse greater amounts of foreign aid to “buy” policy concessions from autocratic regimes (e.g., Bueno de Mesquita and Smith , Menaldo ,). Thus, to mitigate these concerns and to better draw causal inferences, I exploit an event in the distant past (i.e., a so-called critical juncture) to “explain” variation in contemporary cohesive institutions that is “causally prior” (exogenous) to rents today. I then analyze how rents affect political violence in societies exposed to this causally prior historical event. Figure . diagrams the research design, with a brief overview to follow. Less Cohesive Institutions in Muslim Societies In answering why Muslim societies tend to have a stock of less cohesive institutions in the contemporary period, I build on recent scholarship arguing that the expansion of Islam via military conquest (from to ) set these “conquest societies” on a trajectory of authoritarian governance (e.g., Blaydes , Chaney ). In particular, as I outline in the next section (and with
While my empirical analysis focuses mainly on Muslim societies, in many instances I contrast the effects in relation to non-Muslim societies. For example, some of the analysis in Chapter examines the impact of rents in non-Muslim societies. In studying the lack of democracy in many Muslim countries, Rowley and Smith () speculate on the possible role in which Islam was introduced, particularly in Arab societies. In a subsequent series of articles, Eric Chaney and Lisa Blaydes argue that introduction of Islam via Muslim conquest introduced mamluk institutions. They postulate this affected the trajectory of governance after the first wave of Muslim conquest (which effectively ended by ). Their papers do not trace the political legacy of Muslim conquest (and mamlukism). The analysis in Chapters – do. In these chapters, I provide robust empirical evidence that Muslim conquest fostered authoritarian governance during the period of Muslim conquest (Chapter ) and trace its political trajectory from to the present (Chapter ). The statistical evidence supports the historical trajectory depicted in Figure ..
https://doi.org/10.1017/9781009367509.002 Published online by Cambridge University Press
A Political Economy Approach
more comprehensive analysis in Chapters –), I empirically document how Muslim conquest helped centralize authoritarian governance in conquered territories (by ) and then trace its trajectory through to the present. This does not necessarily imply that other (subsequent) historical processes, such as European colonialism (e.g., Hariri , Lawrence ) did not affect governance in Muslim societies. However, as I demonstrate in the empirical chapters, accounting for these alternate historical arguments does not dislodge the long-run effect of Muslim conquest on contemporary political institutions. Ultimately, I show that Muslim conquest exhibits a long-run association with less cohesive institutions in many contemporary Muslim societies. High and Low Levels of Rents In the contemporary era, this preexisting stock of less cohesive institutions has shaped whether prevailing levels of rents in conquest societies has sustained dictatorship or fostered civil war. Specifically, with buoyant amounts of rents, conquest societies are prone to dictatorship. However, when rents decline, these societies are prone to heightened political instability, particularly in the form of civil war. In contrast, non-conquest societies – which, in relation, have more cohesive institutions since they were not exposed to the institutional legacy of Muslim conquest – exhibit less entrenched dictatorship when rents are buoyant but distinctively, are more likely to experience peaceful transitions to democracy when rents decline. I evaluate these arguments by focusing on rents whose sources are foreign in nature, such as revenue derived from the sale of oil abroad and receipts of economic aid from foreign governments.
. I empirically evaluate these arguments with case vignettes and cross-national statistical analysis. The latter, recognizes the inherent challenges posed by endogeneity and tackles them with empirical strategies to identify causal effects, such as instrumental variables, difference-in-differences research designs, and quasi-natural experiments. Case vignettes are also peppered throughout the empirical chapters to add rich examples of the quantitative analysis. The historical vignettes include accounts of several Muslim societies during the period of Muslim conquest, Spain during its period of reconquest (from the tenth to thirteenth centuries), and comparisons of premodern (i.e., prior to ) Muslim societies with premodern societies in Europe and China. The vignettes from the contemporary era span several Muslim societies (e.g.,
This approach contrasts with other related works which tend to employ a more qualitative/ descriptive approach (e.g., Kuran , Kuru , Platteau ). The qualitative accounts in this book draw largely on secondary sources. An important avenue for future work is to complement these accounts with material that draws on historical archives (especially from the early Medieval period).
https://doi.org/10.1017/9781009367509.002 Published online by Cambridge University Press
Political Violence
Nigeria, Somalia, Saudi Arabia) and some non-Muslim ones (e.g., Venezuela, Peru, Chile, Romania, Hungary). The empirical evidence is structured in six chapters, split into two parts. Each part explores the key parameters from the book’s analytical framework: cohesive institutions and rents. Part II shows how societies exposed to Muslim conquest tend to exhibit less cohesive institutions in the modern era. Part III documents how variation in economic rents – mainly, oil revenues and foreign aid – explain ebbs and flows of dictatorship and civil war propensity in these conquest societies. Together, the findings in Parts II and III show how the interplay of the institutional legacy of Muslim conquest and temporal variation in rents explain patterns of dictatorship and civil war between Muslim and nonMuslim societies over time.
A Conquest Part II of the book – “Conquest” – studies the long-run effect of Muslim conquest on cohesive institutions in many contemporary Muslim societies. The analysis focuses on how Muslim conquest generated an equilibrium of centralized autocracy (less cohesive institutions) that has persisted through to the contemporary period. Chapter provides a historical narrative describing the process of Muslim conquest, the institutions it introduced in conquered territories and the persistence of centralized autocracy from through to the present. Following the death of Prophet Muhammad in CE, Islam spread rapidly through military conquest (Kennedy ). Figure . describes the geographic exposure of contemporary nation-states to Muslim conquest. By , Muslim rule stretched from Spain to the western boundaries of China and reached southward along the eastern coast of Africa (e.g., Tanzania). The percentage
Percentage [0.00,0.00] (0.00,0.00] (0.00,0.90] (0.90,1.00]
. Percentage of a modern state’s territory conquered by Muslim armies Notes: The figure above is based on digitizing data on the geographic distribution of Islam over time from Kennedy ().
FIGURE
https://doi.org/10.1017/9781009367509.002 Published online by Cambridge University Press
A Political Economy Approach
of territory (corresponding to modern-day national boundaries) under Muslim rule varied. For example, while the entire modern Middle East was fully conquered by Muslim armies by , only and percent of modernday Spain and Pakistan were. Some contemporary Muslim majority societies were not exposed to Muslim conquest and its subsequent institutional structures, such as Indonesia and Malaysia. The expansion of Islam via military conquest introduced two institutions that consolidated political authority in a dictator (caliph): the use of elite slave soldiers (mamluks) as the backbone of the regime’s repressive capacity and a non-hereditary form of compensation for their service (iqta). In conjunction with the Caliph and his agents of repression and control (mamluks), the ulema (clerics) helped legitimize this “conquest equilibrium” of centralized autocracy in conquered territories; for example through Islamic law that justified absolutist rule. After the end of the Muslim conquests, subsequent rulers did not actively undermine the conquest equilibrium. In particular, the Ottomans employed mamluks and adapted iqta (Lapidus , Tsugitaka ) while the use of indirect rule by European powers did not change the underlying institutional structures in colonized Muslim societies (Fieldhouse , Gerring et al. ). Thus, upon their independence, these autocratic structures remained in territories previously conquered by Muslim armies. This historical narrative suggests that Muslim conquest generated a trajectory that: () centralized political authority in absolutist rule in conquered territories during the initial phase of conquest with () autocratic structures that persisted. This trajectory is evaluated with statistical analysis in Chapters and . Chapter examines how the process of Muslim conquest moved conquered societies in the direction of more centralized autocracy. Drawing on data measuring political consolidation for a global sample of countries (corresponding to contemporary national boundaries) from the year to the
In Chapter , I describe how Islam spread to these non-conquest Muslim territories primarily via trade and not military conquest. While the ruler (state) did not necessarily control all land, in conquered territories the share of land controlled by the state was non-trivial and importantly generated sufficient revenue for the regime to maintain its political authority (Cahen ). The nuances of iqta are described in Chapter . More generally, Chapter provides an extended discussion of when and why mamluk institutions emerged during the (latter) period of the Muslim conquests. In several instances, European governments also purposefully thwarted democratic movements in their colonies (e.g., Lawrence , Thompson ). Relatedly, European governments may have influenced the political resource curse in several oil producers in the Persian Gulf (Girod and Walters ). More generally, the argument that colonial powers governed with existing institutional arrangements is not unique to Muslim countries. For example, in the Americas, European colonists were more likely to employ local, indigenous forces when they encountered an indigenous political administration that was already coercing labor (Arias and Girod ). The historical account in Chapter describes (and contrasts) the trajectory of institutions in Europe and China.
https://doi.org/10.1017/9781009367509.002 Published online by Cambridge University Press
Political Violence
present, I provide robust statistical evidence that countries that experienced greater “exposure” to Muslim conquest experienced a significant increase in state centralization (“autocratization”) during the period of Muslim conquest. Employing a differences-in-differences research design, I interpret these findings causally. Having shown that Muslim conquest helped “autocratize” conquered societies, Chapter traces the institutional legacy of this conquest equilibrium from through to the present. The analysis presents three main results that suggest that Muslim conquest is negatively associated with democracy. I first focus on the period around . Here, I compile original subnational data from medieval Spain to show how provinces under a longer duration of Muslim rule were slower to hold their “first parliament” during the Reconquista. I then expand the geographic scope of analysis and fast forward about years to examine how Muslim conquest affected the strategy of European colonial rule. The analysis reveals that Europeans were more likely to use indirect rule that preserved the governing structures and coalitions (e.g., an ulema-state alliance) in countries that were previously conquered by Muslim armies. Finally, I demonstrate the long-run effect of Muslim conquest on contemporary democracy: Countries whose modern-day territory was under greater control from Muslim armies (during the conquest period) are significantly less democratic today. Figure . provides an initial snapshot of this long-run association in the raw data. Taken together, the analysis in Chapters – strives to show that Muslim conquest is a robust determinant of less cohesive institutions in contemporary Muslim societies. In doing so, the analysis establishes an important pattern in the cross section of countries worldwide: Muslim conquest is associated with a “democratic deficit” (relative to non-Muslim societies) in the contemporary period.
B Rents Building on the findings in Part II, the empirical analysis in Part III of the book – “Rents” – focuses on the contemporary period (i.e., from onward) and shows how variation in rents in countries exposed to Muslim conquest (conquest states) helps explain patterns of dictatorship and civil war over time. Recognizing existing scholarship linking oil wealth to political violence in Muslim societies, I accordingly “split” my research design. Specifically,
As I explain in Chapter , in the premodern period the strengthening of political authority represented an expansion and consolidation of authoritarian governance. Thus, I interpret an increase in state centralization in the premodern period as indicative of state “autocratization.” A robust negative relationship exists in a plot of sample of countries that experienced Muslim conquest only. This sample of countries excludes countries where Muslim armies did not conquer any territory, such as Gambia (GMB) and North Korea (PRK).
https://doi.org/10.1017/9781009367509.002 Published online by Cambridge University Press
A Political Economy Approach
.2
Democracy in 2010 .4 .6
.8
1
GBR AUT AUS CAN CHL CRI CYP DNK FIN DEU GRC HUN IRL ITA JPN LTU MUS MNG NZL NLD NOR POL SVN SVK CHE SWE TWN TTO URY USA PRT ALB BGR COM HRV CZE EST FRA JAM MKD MDA MNE NIC PAN PER ROU ZAF IND ARG BEL BRA BWA DOM SLV GHA GTM IDN KOR LVA LSO MEX PRY PHL SRB SLB KEN BEN BOL COL HND SLE ZMB SEN BDI GUY GNB LBR MYS MWI NAM UKR NPL BGD COD ECU MOZ NGA PNG RUS THA BTN GAB LKA KHM DJI GIN ZWE BFA HTI CIV MDG CAF UGA TZA AGO SDN SGP TGO VENETH CMR COG FJI RWA GMB GNQ MMR KAZ BLR CUB LAO VNM CHN ERI
ESP
TUR
MLI PAK
NER
LBN GEO ISR ARM IRQ DZA
SOM TCD
MRT EGY JOR TUN BHR MAR AZE IRN KWT LBY SYR OMN ARE TKM UZB QAT SAU
0
SWZ PRK
TJK
0
.2 .4 .6 .8 1 Percentage of territory controlled by Muslim armies in 1100
FIGURE . Muslim conquest and contemporary democracy Notes: Democracy is a country’s normalized POLITY score in . A value closer to one implies a higher quality of democratic institutions and governance. Country names are indicated with three-letter country code (from the World Bank).
I probe the relationship between rents and political violence across countries that produce oil and those that do not. Chapter focuses on conquest states that produce and export a large share of oil abroad (e.g., Saudi Arabia, Libya, Kuwait, Iraq, Iran) which I refer to as conquest petrostates. The empirical analysis in this chapter demonstrates that oil wealth has entrenched dictatorship in conquest petrostates without significantly elevating the incidence of civil war. This analysis confirms the first prediction from my formal model: rents entrench dictatorship in societies where existing institutions are less cohesive. To identify causal effects, I leverage damage associated with natural disasters “far away” as an instrumental variable for oil wealth. Chapter focuses on Muslim societies – many of whom are conquest states – that do not produce oil but tend to be recipients of foreign aid, such as Jordan and Pakistan. I refer to these countries as Muslim recipients. Using plausibly exogenous variation in oil prices as an instrumental variable for largely untied foreign aid inflows, the analysis in this chapter shows that buoyant amounts of
This identification strategy is from Ramsay ().
https://doi.org/10.1017/9781009367509.002 Published online by Cambridge University Press
Political Violence . Conquest status, rents, and variation in political violence, with country examples
TABLE
Conquest societies
Non-conquest societies
Increase in rents
DICTATORSHIP (Saudi Arabia, Pakistan)
DICTATORSHIP (Nigeria, Peru, Poland)
Decline of rents
CIVIL WAR (Somalia)
DEMOCRACY (Nigeria, Peru, Poland)
Notes: Countries in italics are oil producers (petrostates). Non-italicized countries are non-oilproducing transfer recipients. Recall, conquest societies tend to exhibit preexisting political institutions that are less cohesive relative to institutions in non-conquest societies.
foreign aid foster political repression and dictatorship, but their subsequent decline can increase the incidence of civil war. These findings corroborate predictions and from the formal model. Finally, Chapter focuses on “negative cases” by examining the effect of rents in non-conquest states – and with relatively more cohesive societies – that receive foreign transfers. I will refer to these countries as non-Muslim recipients. Using case studies from Latin America and Eastern Europe coupled with cross-national statistical analysis from a global sample of transfer recipients (that also includes countries from Africa and Asia), I present two important sets of results. First, I show that buoyant amounts of foreign transfers can heighten political repression and entrench dictatorship in non-Muslim recipients. Second, I show that subsequent declines in transfers facilitated peaceful transitions to democracy in these countries. This latter finding corroborates the third prediction from my formal model that reductions in rents in societies where existing institutions are more cohesive can lead to peaceful transitions to democracy. Together, Parts II and III show how the interplay of the institutional legacy of Muslim conquest and two important rent streams – oil revenues and foreign aid receipts – explain patterns of political violence between Muslim and nonMuslim societies over time. Table . provides country examples that correspond to some of the evidence in Parts II and III. The columns correspond to Part II. They show that conquest societies tend to have less cohesive political institutions (i.e., proxied with a lower level of democracy) compared to nonconquest societies. The rows correspond to Part III and show increases or decreases in the level of rents affect patterns of dictatorship, democracy, and civil war in conquest and non-conquest societies. To foreshadow some of the evidence in Part III, the analysis in Chapters and reveals that buoyant amounts of rents help sustain dictatorship in conquest societies (e.g., oil in Saudi Arabia, foreign aid to Pakistan from to ). However, when rents decline, these societies are prone to
https://doi.org/10.1017/9781009367509.002 Published online by Cambridge University Press
A Political Economy Approach
heightened political instability, particularly in the form of civil war (e.g., Somalia from the late s onward). Higher levels of rents – e.g., from oil, petrodollar investments, trade subsidies – can also strengthen dictatorship in societies not exposed to Muslim conquest (e.g., Nigeria, Peru from –). However, a decline in rents in these societies does not necessarily incite violence (see Chapter ). Rather, non-conquest societies – which, relative to conquest ones, have more cohesive institutions since they were not exposed to the institutional legacy of Muslim conquest – are more likely to experience peaceful transitions to democracy when rents decline (e.g., Nigeria after , Peru after , Poland after ).
. A The Role of Islam My argument and evidence contrast with recent research that seeks to explain economic and political underdevelopment in many contemporary Muslim societies. An important strand of this literature also probes the historical origins associated with Islam, including its decentralized nature (Platteau ), Islamic law (Kuran ) and an alliance between religious clerics (ulema) and the ruler (Kuru , Rubin ). While these impressive works document a “long divergence” in economic (and political) development, primarily between the Middle East and Europe, they are limited in explaining within country variation in important contemporary political outcomes, particularly dictatorship/democracy and civil war. For example, while Islamic law can explain why certain Muslim societies can exhibit authoritarian political structures it is limited in explaining when and why that society might transition from dictatorship to civil war (e.g., Somalia in the mid-s). In short, Islamic law is a largely time-invariant variable, while explaining a country’s transition from dictatorship to civil war requires a variable that varies over time. My argument also contrasts with a related, second strand of scholarship that emphasizes the role of Islamic/Muslim “culture” in explaining authoritarianism. For example, in explaining patterns of democratization in the late twentieth century, Samuel Huntington () argues that Islamic culture has been an important obstacle to the adoption of democratic norms, practice, and
As discussed in Chapter , Nigeria is a Muslim majority country where Islam was adopted through commercial (trade) networks rather than via military conquest. In the contemporary period, except for a three-year period of democracy (–), Nigeria was a dictatorship from through . Chapter focuses on the effect of two types of foreign transfers that are like foreign aid in Latin America and Eastern Europe. Petrodollars in Latin America and trade subsidies in Eastern Europe helped sustain dictatorial rule in the s and early s. When these transfers started to decline in the mid to late s, these countries democratized.
https://doi.org/10.1017/9781009367509.002 Published online by Cambridge University Press
Political Violence
institutions. This book refutes this cultural explanation. The book’s empirical evidence clearly demonstrates that Muslim societies that were not exposed to the institutional trajectory of centralized autocracy introduced through Muslim military conquest can be democratic. Moreover, as I discuss in Chapter , several countries exposed to Muslim conquest – with more variable rent streams – have been democratic in the modern era (e.g., Lebanon, Pakistan). Democracy is compatible with Islam. B History and Institutions By tracing the origins of a society’s contemporary cohesive institutions to events in the past, this book contributes to burgeoning scholarship which recognizes that contemporary economic and political institutions have “deep” historical origins (see for example, Acemoglu et al. , North , Nunn , Rodrik, Subramanian, and Trebbi ). Two recent scholarly works exemplify this realization. In Why Nations Fail (Acemoglu and Robinson, ) and Pillars of Prosperity (Besley and Persson, a), political institutions are fundamental drivers of each book’s analysis of long-run economic development. While this book’s approach is inspired by this scholarship, it departs in two important ways. First, discussions of institutional development in Muslim societies receive scant attention in Why Nations Fail and Pillars of Prosperity. Rather, the authors focus much of their analysis explaining patterns of institutional development in the Americas and Western Europe. Second, my analysis of rents (i.e., nontax government revenue) differs from each book. Whereas rents play an important role in explaining temporal patterns of democracy, dictatorship, and civil war in this book, rents are not central per se in Acemoglu and Robinson’s account of extractive and inclusive institutions. Rents do feature in Besley and Persson’s analytical framework,
This list of cited works is far from exhaustive. Some additional studies on historical legacies, include Dell (), Engerman and Sokoloff (), Guardado (), Nunn and Wantchekon (), and Statasavage (). Regarding the theoretical underpinnings for development, Acemoglu and Robinson draw sharp distinctions between “extractive” and “inclusive” institutions, while Besley and Persson focus their attention on how “clusters” of institutions emerge (e.g., fiscal and legal capacity). Empirically, these scholars show that modern economic and political institutions have “deep” historical origins. For example, Acemoglu and Robinson discuss the role of colonial policies in shaping institutions in the Americas, while Besley and Persson describe the role of interstate war in developing strong fiscal and legal institutions. For example, Muslim and/or Islam appears only three times in the index of How Nations Fail and does not appear at all in Pillars of Prosperity. In a subsequent article, Acemoglu and Robinson () situate their book’s argument within Tullock’s () seminal work on rent-seeking. While related, my characterization of rents, which are predominantly nontax resources that accrue to the government, is not identical to Tullock’s notion of rent-seeking. Among economists, rent-seeking describes the distortionary
https://doi.org/10.1017/9781009367509.002 Published online by Cambridge University Press
A Political Economy Approach
although their theory predicts monotonic effects in which higher levels of rents can foster political violence: first, repression and then civil war. Unlike the model advanced in this book, Besley and Persson’s theoretical formulation does not allow for transitions to democracy or civil war when rents decline. Rather, their theory portends that declines in rents should induce reductions in political violence across all types of societies. As I empirically demonstrate in Part III of the book, Besley and Persson’s hypothesis may be conditional: Reductions in rents can facilitate transitions to democracy when preexisting institutions are more cohesive but incite civil conflict when these institutions are less cohesive.
C International Politics In contrast to existing studies that tend to focus on domestic political factors to explain underdevelopment in Muslim societies (e.g., an ulema-state alliance, sharia law in organizing state-society relations), international politics plays an important role in this book’s central explanatory variables: conquest and rents. The expansion of Islam via military conquest was a centrally orchestrated endeavor spanning successive Caliphs that introduced common governing institutions in geographically and culturally distinct regions, such as Spain, the Middle East, Sindh/Pakistan, Caucasus, and Central Asia. Understanding the aftereffects of this trans-territorial process therefore contributes to existing scholarship on how foreign rule (e.g., empire, colonialism, British and US hegemony) affected politics in conquered territories and its lingering effects (e.g., Arias and Girod , Gunitsky , Kohli , Lawrence , Olsson , and Thompson ). Turning to the second explanatory variable, this book “internationalizes” the concept and application of rents to explain political economy in Muslim and many non-Muslim societies, especially those not endowed with oil wealth. This includes the role of geopolitical rents (e.g., implicit security guarantee from a foreign power) in Muslim petrostates, foreign aid in Muslim non-oil producers, access to “cheap” sovereign borrowing (associated with petrodollars) in Latin America, and subsidies from the Soviet Union during the Cold War to its Satellite countries in Eastern Europe. In doing so, the internationalization of rent streams contributes to important literatures in international relations, including works that probe the relationship between foreign aid and regime type (e.g., Ahmed , Bermeo , Dunning ) and studies linking transnational capital to domestic politics (e.g., Ahmed , Boix , Freeman and Quinn , Lenin , Milner and Mukherjee , Pepinsky , Polanyi , Tilly , , ). effect associated with the “predation” of economic resources that is frequently controlled by the state/government (e.g., natural resource wealth, foreign aid). Some of the economic distortions associated with rent-seeking include corruption and lower economic growth.
https://doi.org/10.1017/9781009367509.002 Published online by Cambridge University Press
Political Violence
D Political Resource Curse While I am not the first scholar to emphasize the importance of rents in explaining patterns of political violence in contemporary Muslim societies, this book introduces two original contributions to research on the so-called political resource curse. The first contribution is theoretical. As I describe in Chapter , existing theoretical accounts predict monotonic effects between rents and political stability that are puzzling. On the one hand, greater amounts of rents stabilize dictatorship (e.g., Besley and Persson a, Morrison , Ross , Smith ). Yet on the other, greater amounts of rents can be politically destabilizing – particularly in dictatorships – by heightening the propensity of civil violence for control of the “state prize” (e.g., Besley and Persson a, Grossman , Hodler , Lane and Tornell , ). Together, these predictions beg the question: Under which conditions are rents politically stabilizing or destabilizing? Moreover, these existing theoretical formulations do not necessarily show how reductions in rents affect politics nor provide pathways to peaceful outcomes, such as democracy. Chapter ’s formal model addresses this theoretical puzzle by precisely articulating the conditions under which variation in the levels of rents affect dictatorship, democracy, and civil war in a unified framework. The book’s second contribution on the political economy of rents is empirical. Much of this extant literature – frequently with applications limited to a few Muslim societies in the Middle East and North Africa region – has focused on how oil wealth affects political stability (e.g., Dunning , Mahdavi , Ross , ). As a consequence, these studies are essentially limited in explaining politics in oil-producing Muslim states. This book broadens the empirical scope by also studying how other types of nontax income, such as foreign aid and additional types of foreign transfers (trade subsidies, foreign borrowing), explain patterns of political violence in non-oilproducing countries. As a result, this book provides at least two extensions to existing empirical findings. First, it provides a more comprehensive study of political violence in Muslim societies that extend beyond the MENA region, and particularly oil-producing countries. This is notable as many non-MENA countries are populated with the largest number of Muslims worldwide (e.g., Bangladesh, Indonesia). Second, the book’s empirics show that various forms of foreign transfers can generate a political resource curse akin to the pernicious
This scholarship has identified various channels through which oil wealth affects political stability (Morrison , Ross , ,), specific features of the state’s ownership of these resources (Luong and Weinthal , Mahdavi ), and prevailing socioeconomic conditions that mediate these relationships, such as income inequality and human capital (Brooks and Kurtz , Dunning ).
https://doi.org/10.1017/9781009367509.002 Published online by Cambridge University Press
A Political Economy Approach
effect of oil, particularly in nondemocratic regimes. This implies that inflows of certain types of foreign capital can induce a political resource curse. E Analytical Political Economy Finally, the book’s methodological approach pushes the frontier for studying political economy and violence in Muslim societies. First, the book’s theory is parsimonious: It identifies two key independent variables – rents and cohesive institutions – to explain variation in political violence. The formal model is original in showing how increases and decreases in rents affect political outcomes across societies with different institutional settings. As such, the theory’s predictions are applicable to both Muslim and non-Muslim societies. Second, the empirical strategy compiles data across a global sample of countries from the year to the present and conducts a variety of statistical analyses. This strategy distinguishes it from several prominent studies that take a largely qualitative historical approach (e.g., Kuru , Platteau ) and/ or often develop arguments for a single or few Muslim societies (e.g., Blaydes , , Jamal , Masoud ). Cognizant of the challenges posed by endogeneity, my statistical analysis strives to gauge causal relationships using cutting edge techniques. For example, in Chapter , I leverage a difference-in-differences research design to evaluate how Muslim conquest consolidated authoritarian rule and subsequently affected the trajectory of authoritarianism in conquered societies relative to non-conquered societies. My analysis of rents in the contemporary period in Part III utilizes several instrumental variables (for rent streams that vary over time) to identify causal effects between rents, civil war and dictatorship/democracy. In the process, the use of various econometric methods allows me to place objective bounds on whether the observed relationships are ‘statistically significant’ and not just random associations in the data. It should be noted, that while my empirical analysis relies extensively on quantitative analysis, the book’s qualitative accounts – which are primarily based on case vignettes from secondary sources – could be further strengthened by drawing on primary source material (e.g., archives from the medieval era) to more fully process trace the book’s arguments.
Future work could explore the model’s predictions in non-Muslim societies. Chapter provides some preliminary analysis in that direction. Amaney Jamal () focuses on how US foreign policy affects authoritarianism politics in Jordan, Kuwait, Morocco, Palestine, and Saudi Arabia. Other leading works carefully study strategies of authoritarianism (e.g., repression, electoral manipulation) in single Muslim societies. Blaydes (), for example, examines strategies of repression employed in Saddam Hussein’s Iraq, while Blaydes () and Masoud () focus on electoral politics in Egypt’s hybrid authoritarian system. In Chapter , I am careful to show that the underlying assumptions are met to use difference-indifferences approach, such as satisfying the parallel trends assumption.
https://doi.org/10.1017/9781009367509.002 Published online by Cambridge University Press
Political Violence
. With these contributions in mind, I proceed as follows. The next chapter presents the book’s analytical framework. It begins with several figures to motivate the analysis at the heart of the book. These figures demonstrate the cross-sectional and temporal variation in dictatorship/democracy and civil war between Muslim and non-Muslim societies and across Muslim societies with different rent structures (i.e., oil versus non-oil producers). With this initial ‘dynamic’ illustration of dictatorship and civil war, I then develop a theory linking variation in rents and cohesive institutions to dictatorship, democracy, and civil war. Consistent with existing formal models, I confirm that higher levels of rents can entrench autocracy, but unlike these formulations, I show how declines in rents can lead to democracy or civil war in the same model. The specific path hinges on the quality of preexisting cohesive institutions. A peaceful transition to democracy is possible when these institutions are more cohesive. In contrast, a decline in rents can foster civil war when preexisting institutions are less cohesive. In short, Chapter identifies two parameters that affect patterns of dictatorship, democracy, and civil war: societal differences in cohesive institutions and variation in the level of rents. I then turn to empirical evaluation of these two parameters separately in Parts II and III of the book. Structured over three chapters (–), the aim of Part II is to show how the institutional legacy of Muslim conquest is a robust determinant of less cohesive institutions in contemporary Muslim societies. In doing so, Part II serves to explain the cross-sectional variation in dictatorship in Muslim societies relative to non-Muslim societies today. Building on these insights, Part III turns to showing how variation in the level of economic rents – oil revenues and foreign aid – explain temporal variation in dictatorship/democracy and civil war across and within contemporary Muslim societies. Organized in three chapters (–), the analysis in Part III demonstrates that buoyant levels of rents entrench dictatorship, while their decline heightens the prospect for civil war (democracy) in societies whose institutions are less (more) cohesive. Chapter concludes. After briefly summarizing the book’s main argument, evidence, and contributions to existing scholarship, I offer extensions from the theory and empirical findings. On the former, I situate the book’s theoretical conjectures in a broader context of the political economy of rents and economic development (e.g., Dutch disease) and evaluate the role of several indirect types of rent streams, such as remittances and trade protectionism (e.g., Ahmed , Eibl and Malik , Monroe ). On the latter, I discuss the book’s empirical implications for democratization in Muslim societies, particularly in the aftermath of the Arab Spring.
https://doi.org/10.1017/9781009367509.002 Published online by Cambridge University Press
Analytical Framework
In Chapter , I identified two tragic characteristics of many contemporary Muslim societies: the pervasiveness of dictatorship and an elevated propensity of civil war. Among political economists, the prevalence of these outcomes – dictatorship and civil conflict – captures the central features of societies afflicted with political violence. Accordingly, in this chapter, I adopt this terminology and develop an analytical framework to better understand why many Muslim societies are inflicted with varying and heightened levels of political violence. I begin with several figures to motivate the analysis at the heart of this book. These figures demonstrate the cross-sectional and temporal variation in political violence between Muslim and non-Muslim societies and across Muslim societies with different rent structures (i.e., oil versus non-oil producers). These figures suggest a relationship between foreign-based rents – notably, oil revenues and foreign aid – and patterns of dictatorship and civil war within Muslim societies. With this empirical motivation, I then advance a theory linking variation in preexisting cohesive institutions and rents to explain patterns of dictatorship, democracy, and civil war and transitions between these outcomes. This discussion proceeds in a stepwise fashion.
For example, in a series of theoretical formulations, Besley and Persson (, , a,b) model a hierarchy of political violence that ranges from peace, repression (one-sided violence), and civil war (two-sided violence). Relatedly, Svolik (, ) describes how dictators govern in societies in which “violence is an ever-present and ultimate arbiter of conflicts in authoritarian politics.” Rents refer to nontax government revenue. This includes revenues from the extraction and sale of natural resources (e.g., oil, natural gas) that can accrue to national governments and the receipt of foreign economic and military aid. Rents can also include sovereign borrowing and subnational transfers (Morrison ). In some instances, rents can be non-economic, for example geopolitical rents in the form of security guarantees (e.g., foreign troop deployments) that can augment a
Published online by Cambridge University Press
Analytical Framework
I first distill the rich scholarship on rents and political stability, identifying an important tension in existing theoretical formulations: whether a reduction in rents contributes to violence or not. This discussion in turn provides the rationale for advancing a formal theory (model) to help reconcile this tension. Whereas existing models predict a monotonic effect linking greater rents to political violence, my model allows for non-monotonic effects. Consistent with these existing theories, I show that higher levels of rents can sustain autocracy, but unlike these models, I show how declines in rents can generate divergent political trajectories. Specifically, I articulate the conditions under which a decline in rents can generate a peaceful transition to democracy or engender civil war. The model highlights the important role of preexisting cohesive institutions in shaping how a decline in rents can lead to these divergent outcomes: democracy or civil war. To the best of my knowledge, the model advanced in this chapter is the first to generate these differential trajectories in a unified analytical framework. The chapter’s unified framework (model) is an important contribution to existing scholarship. As this chapter describes, existing scholarship has tended to study whether rents and institutions affect patterns of democracy/dictatorship or civil war as distinct and separate outcomes (e.g., Hodler , Menaldo , Smith ). This is surprising since regime type and civil war tend to covary: dictatorships are more prone to civil war (e.g., Blattman and Miguel ). I end the chapter by delineating the model’s empirical implications and preview the research designs to evaluate their capacity to explain patterns of dictatorship and civil war in Muslim and non-Muslim societies (which is the focus of Parts II and III of the book).
. : In this section, I present several figures describing patterns of political violence in Muslim and non-Muslim societies that motivate the core questions (“puzzles”) that this book seeks to explain. The first figure plots the average
government’s capacity to maintain political control. For example, the stationing of US troops in various countries in the Persian Gulf (e.g., Saudi Arabia, Kuwait) can serve to safeguard stable world oil supplies and also serve to deter domestic insurgencies (that could disrupt these oil supplies). While I refer to this model with “my” and “I” in this chapter, it was developed in collaborative work with Daniel Schwab and Eric Werker. See Ahmed, Schwab, and Werker (). Societies with more cohesive institutions are those that share more state resources with groups out of power (e.g., supporters of opposition political parties, different ethnic or religious groups). I discuss cohesive institutions in Section ..
Published online by Cambridge University Press
A Political Economy Approach
level of democracy in Muslim- and non-Muslim-majority societies since . To measure a country’s level (quality) of democracy, I use its composite POLITY score (Marshall and Jaggers ). This index and its sub-indices are used widely in empirical research on political regimes, institutional design, variation, and durability. The POLITY score combines information about the competitiveness of and constraints on executive recruitment with information about the extent of political competition. The measure ranges from − (full dictatorship) to + (consolidated democracy). I normalize each country’s annual score on a [,] scale where a higher value corresponds to greater democracy. Countries where percent or more of their population identify with Islam are classified as Muslim.
A Democratic Deficit in Muslim Societies Figure . reports the average normalized POLITY score across Muslim and non-Muslim countries on an annual basis since . The figure shows that Muslim societies have, on average, been less democratic than non-Muslim societies. For example, using conventional classification schemes in economics and political science, the typical Muslim country has always been a dictatorship: its normalized POLITY has not exceeded .. In contrast, non-Muslim societies have exhibited greater temporal variation in their normalized POLITY score. While the typical non-Muslim country started out as a democracy in , it became less democratic over the subsequent decade. In the mid-s, non-Muslim societies gradually became more democratic on average, a pattern consistent with the so-called Third Wave of democratization (Huntington ). This upward trend accelerated after the end of the Cold War in and has stabilized since . Despite some improvements in democratic governance in Muslim societies after , Figure . notably reveals a persistent “democratic deficit” in Muslim countries (relative to their non-Muslim counterparts).
See for example, Acemoglu and Robinson (), Ahmed (), Besley and Persson (a), Dunning (), and Morrison ().Bueno de Mesquita et al. () use sub-indices of the composite POLITY score to construct their measure of selectorate and coalition size. Alternate measures of democracy (e.g., Boix, Miller, and Rosato ) are contrasted and evaluated in relation to POLITY. Note that my use of “typical” refers to the statistical average across each group of countries (Muslim, non-Muslim). Scholars often use cutoffs in the POLITY index to differentiate dictatorships from democracies. Economists frequently classify countries with a normalized POLITY score greater than or equal to . as a democracy and those below . as a dictatorship. In contrast, political scientists employ a higher cutoff for democracy: a normalized POLITY score equal to or exceeding . (which corresponds to a POLITY score equal to or above +).
Published online by Cambridge University Press
0
.2
POLITY (normalized) .4 .6
.8
1
Analytical Framework
1960
1965
1970
1975
1980
1985 Year
Non-Muslim FIGURE
1990
1995
2000
2005
2010
Muslim
. Persistent democratic deficit in Muslim societies
B Cohesiveness in Muslim Societies The democratic deficit identified in Figure . is associated with less cohesive institutions in Muslim societies. Figure . provides some suggestive evidence using welfare spending as a measure of cohesive institutions (like Figure .). The figure shows that governments in Muslim societies, on average, allocate a persistently lower percentage of their expenditure on welfare spending relative to non-Muslim societies. Over the entire period, the difference in shares (of about percentage points) is statistically different from zero (with a p-value < .). This implies that Muslim societies tend to be significantly less cohesive relative to non-Muslim societies. It is worth emphasizing that reliable crossnational data on welfare spending is available from and is therefore less extensive than measures of democracy, which are available from (and for many countries for decades prior to ). As democracy and cohesiveness tend to be positively associated (see Figure .), my analysis going forward will
Recall that cohesiveness refers to the proportion of state revenue (expenditure) shared between supporters of the ruling government and those in opposition. In societies with less cohesive institutions, the ruler is less egalitarian in sharing state revenue with supporters and opposition groups. That is, the ruler tends to spend a lower share of expenditure on broad-based subsidies and transfers to the masses.
Published online by Cambridge University Press
0
Subsidies and transfers (% Expenditures) 10 20 30 40
A Political Economy Approach
1990
1995
2000 Year Non-Muslim
FIGURE
2005
2010
Muslim
. Cohesiveness in Muslim and non-Muslim societies
use democracy as a proxy of cohesiveness, primarily since there is more extensive cross-national and temporal data coverage for this measure.
C Political (In)Stability in Muslim Societies The relative stability of dictatorship in Muslim societies suggested in Figure ., however, seems to mask differences in their underlying rent structure. To see this, Figure . plots the average normalized POLITY score across oil and nonoil-producing Muslim societies (I use British Petroleum’s designation of oilproducing countries). This figure reveals several interesting patterns regarding the relative stability of dictatorship across oil and non-oil-producing Muslim societies. First, while oil and non-oil-producing Muslim societies exhibited, on average, similar levels of democracy in , oil producers became increasingly less democratic in the s. Second, despite some improvements in democracy in the s, non-oil-producing Muslim societies reversed this trend in the s. And by the early s, Muslim non-oil producers were less democratic than their oil-producing countries. Third, Muslim non-oil producers experienced another reversal: in the mid-s they started to become slightly more
As I discuss in Chapter this trend may be due to the nationalization of oil companies in the s and the subsequent surge in oil income to national governments (Mahdavi ).
Published online by Cambridge University Press
0
POLITY (normalized) .2 .4 .6
.8
Analytical Framework
1960
1965
1970
1975
1980
Oil FIGURE
1985 Year
1990
1995
2000
2005
2010
Non-oil
. Democracy in Muslim oil and non-oil producers
democratic and by the early s passed the threshold (barely) for democracy. Fourth, Muslim oil producers did not experience these improvements in democracy. By , these societies tended to be less democratic than they were in . The greater variation in democracy among Muslim non-oil producers prompts greater scrutiny. Figures . and . provide some suggestive explanations. Figure . plots the average incidence of civil war in Muslim non-oil producers (left axis) against their average level of democracy (right scale). It shows that Muslim non-oil producers were relatively peaceful in the s and s (when they also became less democratic), but became increasingly conflict-prone in the mid-s until the late s. During this period, these
Similar patterns are not observed in non-Muslim countries. Since the s, both oil and nonoil-producing countries, on average, have become more democratic over time. Many of these oil producers have been consolidated democracies since the s (e.g., Norway, United States) and only recently discovered significant quantities of oil (e.g., the United Kingdom in the North Sea in the s). See Figure A in this chapter’s appendix. Data on civil war is from the Armed Conflict Database (Harbom, Melander, and Wallensteen ). Following convention, conflicts exceeding , battle deaths per annum comprise civil war. In contrast, battle deaths below the , threshold comprise civil conflict. Similar patterns are not observed in Muslim oil producers (see Figure A in the chapter’s Appendix). These countries have remained relatively stable politically compared to Muslim non-oil producers. In Muslim oil producers, the average incidence of civil war has hovered
Published online by Cambridge University Press
0
.2
.05
Civil war .1
.15
.3 .4 .5 POLITY (normalized)
.2
.6
A Political Economy Approach
1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 Year Civil war FIGURE
POLITY (normalized)
. Civil war and democracy in Muslim non-oil producers
societies were twice as likely to experience civil war relative to the prior two decades. Concomitantly, these societies started to become more democratic as the average normalized POLITY score started to rise. With these improvements in democracy, by the early s civil strife in these countries subsided significantly. A potential source for the patterns in Figure . may be attributable to variation in a specific type of unearned (nontax) government income received in many non-oil-producing Muslim societies: foreign aid. As Neumayer () and Werker, Ahmed, and Cohen () document, these countries have been the recipients of significant amounts of foreign aid from Western and Arab donors. Figure . shows that movements in foreign aid per capita are correlated with political stability in Muslim non-oil producers. The figure suggests
between . and . since and has been driven primarily by sustained violence in Algeria and Sudan. Despite some fluctuation – both increases and decreases – over time, the average level of democracy in Muslim oil producers has not changed since . These patterns are not present in non-Muslim non-oil producers (see Figure A in the chapter’s appendix). First, the average incidence of civil war in non-Muslim countries has been significantly lower (e.g., . at its maximum in ). Second, a decline in aid is not associated with an uptick in civil war. As I describe and formalize later in this chapter, this null association is theoretically plausible because non-Muslim countries could have (relatively) more cohesive institutions. The analysis in Chapter provides empirical support.
Published online by Cambridge University Press
0
0
.05
50
Civil war .1
100 Aid per capita
.15
150
.2
200
Analytical Framework
1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 Year Civil war
Aid per capita
FIGURE . Foreign aid and civil war in Muslim non-oil producers Note: Aid per capita reports the annual average value among Muslim non-oil producers in in US dollars. These figures are based on the author’s calculations using data from the World Bank.
buoyant amounts of foreign aid may “buy” political stability, while their decline may engender conflict. Moreover, notice how a significant decline in foreign aid in the mid-s coincides with the uptick in civil war in Muslim non-oil producers. Figures .–. generate several questions at the heart of this book. For instance, why is there a persistent democratic deficit in Muslim societies relative to non-Muslim societies? What explains its temporal variation, particularly its widening since the s? For example, why did many Muslim societies miss the Third Wave of democratization? Looking across Muslim societies, why have oil producers remained more politically stable compared to non-oil-producing Muslim societies? Why might a country’s rent structure matter? Moreover, why have non-oil-producing Muslim countries experienced greater
In contrast, a reduction in oil prices (and associated oil rents) seems to be unassociated with, and notably has not triggered, heightened violence in Muslim oil producers (see Figure A in the chapter’s appendix). As I argue in Chapter , this may stem from the provision of geopolitical rents (e.g., implied US military support for the government) in Muslim oil producers that has effectively deterred opposition groups from engaging in combat against the government.
Published online by Cambridge University Press
A Political Economy Approach
political upheaval in the form of civil war but remained relatively nondemocratic? To answer these questions we need to explain () the persistence of nondemocratic political structures in Muslim societies relative to non-Muslim societies and () the possible role of rents in explaining variations in political violence – dictatorship and civil war – across oil and non-oil-producing Muslim societies. In Part II of this book, I provide an answer to the first question by demonstrating how the institutional legacy of Muslim conquest – a process that occurred over a millennia ago – helps explain the persistence of autocracy in many contemporary Muslim societies relative to non-Muslim societies (i.e., the democratic deficit in the cross section of countries). In the remainder of this chapter, I focus on the second question by explaining how rents can affect variations in political violence both across countries and over time, and how this may be related to the quality of preexisting cohesive institutions.
. There is an expansive literature on the causes and consequences of dictatorship and civil war. Rather than summarizing this rich scholarship, my aim in this section is to distill the analytical features that are relevant to understanding how rents can explain variation in political violence. This discussion highlights some of the prominent theoretical and supporting evidence and showcases some of the mixed findings. In doing so, I identify several tensions in this existing scholarship, which serves as a motivation to formulate my own theoretical formulation (model) to reconcile these tensions. Before delving into this, I first define key terms. A Political Violence Both as a cause and consequence, dictatorship is an important feature of political violence. At its core, dictatorship comprises rule by the few over the many, with many scholars conceptualizing it as a “residual” to
As such, Part II provides an account for the cross-sectional variation in contemporary levels of democracy between Muslim and non-Muslim societies. See Blattman and Miguel (), Haggard and Kaufman (), Kalyvas (), and Svolik () for excellent reviews of this scholarship. Throughout this book, I will use the terms dictatorship, nondemocracy, autocracy/ authoritarianism interchangeably. I am not unique in investigating some of these tensions. In evaluating the relationship between oil and democracy, Benjamin Smith and David Waldner () identify the “scholarly dissensus,” highlighting conceptual, theoretical, and empirical challenges in existing scholarship. While civil wars occur most frequently in dictatorships, this is not universal. Democracies too may experience civil conflict, such as the American Civil War. Thus, I argue that dictatorship is a strong “correlate” (rather than source) of civil war.
Published online by Cambridge University Press
Analytical Framework
democracy. In this tradition, dictatorships and democracies are contrasted on their procedural approach to the selection of leaders (Dahl , Schumpeter ), often by clearly articulating the minimal electoral requirements for democracy (Alvarez et al. , Boix ). Empirically, this generates a dichotomy between dictatorships and democracies (see e.g., Boix, Miller, and Rosato ). In contrast, other formulations identify and seek to categorize the broader institutional variations that feature across dictatorships, such as differences in the constraints faced by the leader, the competitiveness of leader recruitment/selection, and more generally who can participate in the political process. These variations have been measured on a continuum or with discrete categories. The POLITYindex (Marshall and Jaggers ) is an example of the former, while categorizations, such as “competitive” (Levitsky and Way , ), “electoral” (Schedler ), or “semi-authoritarianism” (Ottaway ) characterize the latter. Turning to their operation, as Milan Svolik () cogently distills, all dictatorships must grapple with two central problems: those of authoritarian control and power-sharing. Each problem relates to a distinct group in a polity: those who rule (“the elite”) and those who are ruled (“the masses”). The challenge of balancing against the threats from the masses (majority) excluded from power constitutes the problem of authoritarian control. In practice, of course, dictators rarely control enough resources by themselves to thwart these challenges and therefore rule with a (small) number of allies, such as military officers, prominent party members, or clerics. This can generate a second, distinct political problem when dictators confront challenges from those with whom they share power. This is the problem of authoritarian power-sharing. A dictator who fails to address these problems faces a heightened risk of losing power through elite defection (coups) and violence from opposition groups (civil wars) and less frequently via social revolution. Moreover, these dual challenges are magnified by the dismal setting of authoritarian politics. First, a dictator lacks an independent authority that can enforce agreements with his governing allies and repressive agents, thus
Gandhi (, chapter ) provides a superb discussion of the historical and contemporary use of dictatorship. Many prominent scholars conceptualize dictatorship as a residual category of democracy. For example, Boix (), Cheibub, Gandhi, and Vreeland (), Przeworski et al. (), and Svolik () define dictatorship as an independent country that fails to satisfy at least one of the following two criteria for democracy: () free and competitive legislative elections and () an executive who is elected either directly in free and competitive presidential elections or indirectly by a legislature in parliamentary systems. A related formulation characterizes who rules in dictatorships, such as the military, a single party, or a personalistic ruler. See, for example, Geddes () and Geddes, Wright, and Frantz (). For example, over the period –, only percent of dictators exited office due to popular uprisings (Svolik , ).
Published online by Cambridge University Press
A Political Economy Approach
elevating the risk of elite defection. Second, violence is ever-present and ultimate arbiter of conflicts. Thus, Svolik (, ) proclaims the “key features of authoritarianism – including institutions, policies, as well as the survival of leaders and regimes – are shaped by the twin problems of power-sharing and control against the backdrop of the dismal conditions under which authoritarian politics takes place.” The dismal setting that dictators face raise their societies prospects for experiencing civil war; a finding, well documented in the empirical civil war literature. Conceptually, civil war entails two-sided violence that pits the incumbent (government) in armed combat against one (or more) opposition groups. In explaining their determinants, many scholars differentiate between “grievances” and “opportunities” as important sources of civil war (e.g., Blattman and Miguel , Collier and Hoeffler , Kalyvas ). On the former, since dictatorships frequently exclude certain groups from power, grievances can be an important driver for conflict. For example, regimes that restrict political participation frequently pursue redistributive policies that perpetuate inequalities and perceptions of deprivation that can mobilize rebellion (Besley and Persson , Cederman, Weidmann, and Gleditsch , Gurr ). Of course, the mere existence of grievances does not imply civil war. As Stathis Kalyvas (, ) observes: “Obviously, the problems with grievances as a determinant of civil war onset is that they seem to be much more prevalent than civil war and that they are very hard to measure directly. Ultimately, it is difficult to escape the conjecture that one has to look for combinations of demand for, and supply of, rebellion (or intention and opportunity).” Greed-based explanations posit that economic conditions comprise an important driver of civil war (Collier and Hoeffler ). One channel posits that economic decline, for example through depressed wages, can both lower the opportunity cost for rebels to mount an insurgency and weaken an incumbent’s capacity to fight back (Miguel, Satyanath, and Sergenti , Powell ). Another channel identifies the motive for the incumbent and opposition to control state resources (“the state prize”), such as foreign aid and a variety of natural resources (Besley and Persson a, b, Le Billon ). For instance, revenues derived from the state’s control of natural resources, such as oil, natural gas, diamonds, and timber, can incite violence from all sides. Revenues from natural resources can finance an incumbent’s repressive
The literature on civil war identifies different types of groups based on their: ethnic, linguistic, and religious identification/fragmentation, geographic location (e.g., urban versus rural, near natural resources), social/economic class and modes of production (i.e., levels of vertical and horizontal inequality in societies), and ties to foreign powers. Differences in the composition, size, and/or intensity of grievances among individuals in these groups can motivate armed violence.
Published online by Cambridge University Press
Analytical Framework
capacity, such as the size of its army. Similarly, opposition groups may seek to control the natural resources to finance their rebellion. Empirically, several studies document a higher likelihood, intensity, and duration of civil war in countries with natural resources (e.g., Le Billon , Lei and Michaels , Synder ). Greed-based explanations, of course, do not exclude the importance of political grievances. In practice, the presence of both motives – grievances and greed – heighten the incidence of civil war. Consequently, dictatorships with valuable state prizes (e.g., oil revenues, foreign aid receipts) often face an elevated risk of civil war. The preceding overview of political violence highlight two important points. First, dictatorships entail the rule of the few over the many, in which the regime confronts two problems – of control and power-sharing – in a dismal setting of ever-present violence. Second, dictatorships especially those with valuable state prizes, also face an increased risk of two-sided violence in the form of civil war. B Rents and Political Regimes Rents and Governance Given the political challenges described above, how can incumbents better ensure the longevity of their power? Scholars across multiple disciplines have identified the essential role of buoyant public finances in helping governments (rulers) remain in power. In describing the primacy of revenues for all governments, Margaret Levi (, ) describes how they can benefit a ruler personally and/or his subjects: Rulers may use the funds to line their own pockets or to promote their personal power. They may use the funds to support social or personal ends. They have ideological ends they wish to promote. They may be altruistic. Randomness of ends characterizes rational choices in general, and rulers operate with a wide range of them. It’s by the means of the state and its revenues that rulers achieve their personal and social ends.
As Levi subsequently elucidates, the importance of revenues thus incentivizes rulers to be “predatory” and extract as much revenue as possible. In doing so, however, rulers face constraints. For example, maximizing state revenue involves the cost of extraction or a trade-off between extraction costs (e.g., deadweight loss associated with taxation) and economic development (e.g., growth). Minimizing the transaction costs associated with taxation, therefore, can compel rulers to derive their revenue from alternate sources. Scholars frequently refer to these alternate sources of nontax government revenue as unearned income or rents (e.g., Ahmed , , Morrison , Smith ). Whereas tax revenue entails the extraction of income from
The incumbent could also transfer funds to regime supporters and/or “buy off” rebel groups. Since this sub-section builds on Levi’s () seminal work, I adopt her use of the term “rulers” to refer to governments/regimes.
Published online by Cambridge University Press
A Political Economy Approach
economic agents (e.g., wage earners, firms) and associated transactions (e.g., value-added taxes, trade tariffs), rents may entail revenues derived through the state’s exploitation of natural resources (e.g., oil, minerals) and external transfers (e.g., economic and military assistance from foreign governments). Rents can include revenues from oil because it can be generated through the operation of state-owned oil enterprises rather than through taxes. Rents also include income from abroad (e.g., sovereign borrowing, foreign aid) and intergovernmental grants at the subnational level. For many governments, rents are nontrivial and account for a large share of their expenditures, such as in Nicaragua ( percent), Bolivia ( percent), Republic of Congo ( percent), Burundi ( percent), Romania ( percent), Syria ( percent), Iran ( percent), Nepal ( percent), Pakistan ( percent), and Japan ( percent). Rents and Democracy The generation and composition of government finances from tax and nontax sources has political implications. For instance, in premodern Europe, the need to finance interstate wars compelled European sovereigns to develop tax collection systems. According to Charles Tilly (), taxation required bargaining between revenue-maximizing leaders and their citizens through which rulers relinquished some influence over state policies in exchange for taxes. With more warfare over time and thus a greater reliance on tax revenue, rulers tended to become more accountable to their populations. This bargaining mechanism thus generates a pathway from taxation to more democratic governance; and by extension, governments that generate their revenue from nontax sources are less likely to feature democratic characteristics. This framework linking taxation to democracy has been applied to explain modern political development in various regions, including Africa (Guyer ), Asia (Bernstein and Lu ), the Middle East (Crystal ), Europe and North America (Scheve and Stasavage ), and the former Soviet Union (Gehlbach ).
To be clear, tax revenue entails the extraction of income from individuals (e.g., wages, salaries), firms, land and capital owners, and associated economic transactions via consumption (e.g., value-added taxes) and international trade (e.g., tariffs), among others. Note, my definition of rents excludes remittances (which accrue directly to individuals). However, through an expenditure switching strategy, governments may be able to incorporate remittance income into their revenue base, thus comprising a type of nontax income (Ahmed , ). State-owned enterprises control percent of the world’s oil production and percent of its reserves (Ivanhoe , Tordo et al. ). Examples include Saudi Aramco, China National Petroleum Corporation, Nigerian National Petroleum Corporation, and Petroleos de Venezuela. Based on calculations from Morrison (, ) using data averaged over the period, –. Bates and Lien () formalize this bargaining process and identify two important considerations. First, greater asset mobility may weaken the sovereign’s ability to tax. If the tax rate is too high, citizens may shift their assets abroad (an argument further explored by Boix ). Second, the relationship between taxation and representation may be conditional on the amount of spending that citizens (expect to) receive in return.
Published online by Cambridge University Press
Analytical Framework
While taxation tends to be positively associated with democracy, it frequently generates political contestation within societies. Indeed, redistributive conflicts – typically regarding the level of taxation and associated transfers from the “elites” to the “masses” – are critical elements in prominent theories of political instability and regime transitions (e.g., Acemoglu and Robinson , Boix , Moore ). According to Kevin Morrison (), two tensions underlie this redistributive conflict: the need for governments to spend while minimizing their tax burden. The availability of nontax revenue (rents) can assuage these tensions: “Nontax revenues enable governments to reduce their taxation levels while increasing government spending, thereby increasing the stability of democracies, dictatorships, and their leaders” (Morrison , ). To understand the logic, consider how rents can strengthen democracy in Daron Acemoglu and James Robinson’s () theory of democratic transitions. They posit that elites democratize as means to credibly commit to future redistribution rather than suffer revolution in which the masses expropriate the assets held by the elites. In this context, the availability of rents can help lower a regime’s tax burden, thus relieving the costs of redistribution for elites and facilitating democracy. Thad Dunning () applies this logic in articulating how oil rents can foster “crude democracy”: By reducing the extent to which democratic majorities redistribute private income or wealth away from the rich elites through taxation, land reform, or other measures, resource rents can reduce the economic cost of democracy to elites. . . . This occurs because both resource rents may reduce the current redistributive costs of democracy and because resources may endogenously lower the extent to which future democratic regimes may redistribute private wealth from elites when there is not a coup threat; resource rents thus shape elite expectations about redistribution under future democratic regimes.
Rents and dictatorship. While rents may lower the costs of democratization in theory, there is a large body of research documenting how rents tend to entrench dictatorship. One strand of this literature identifies the role rents can play in introducing authoritarianism in existing democracies. For instance, rents can engender a strong “incumbent advantage” that can reduce political competition and introduce creeping authoritarianism (Wantchekon ). In clientelistic settings, one pathway is via targeted redistribution: The incumbent can transfer rents directly to their supporters (e.g., co-ethnics) and exclude other groups from the direct benefits of rents (Jablonski ). Consequently,
This logic also applies to Boix’s () theory of democratization. Relatedly, Menaldo () argues whether natural resources enhance democracy is conditional on preexisting institutional quality. Rents in countries with good/democratic institutions do not harm democratic governance (e.g., Norway). In support of this conjecture, Dunning shows that natural resource rents, particularly from oil, supported “crude democracy” in several Latin American countries, most notably Venezuela.
Published online by Cambridge University Press
A Political Economy Approach
the recipients of rents may be more likely to turn out and vote for the incumbent. Another strand of this literature discusses the channels through which rents strengthen dictatorship. Much of this scholarship originates from scholars of the Middle East investigating whether the prevalence of rents fostered authoritarian governance in that region (Beblawi , Chaudhry , Madhavy ). The central argument in this “rentier state” literature posits that governments funded by external rents are relieved from the need to raise taxes, which makes them less accountable to their citizens. As a result, these governments are less likely to deploy the rents in ways that promote economic development. Focusing on the political ramifications, scholars have refined the theoretical channels through which rents can hinder democracy. Michael Ross (, –) identifies and tests three broad “effects” through which rents can foster dictatorship: a “rentier effect,” through which rents “relieve the social pressures that might lead to demands for greater accountability” from the broader population; a “repression effect,” through which rents may empower “governments to spend more on internal security and so block the population’s democratic aspirations”; and a diminution of a “modernization effect,” through which “resource-led growth does not lead to higher education levels and greater occupational specialization. . . [and thus] should fail to bring about democracy.” As Ross concludes, these effects are not mutually exclusive as “all three explanations, or any combination of them, may be simultaneously valid.” In delineating these effects, Ross elucidates how rents can help dictators mitigate their dual challenges of authoritarian control and power-sharing, for instance, through his decomposition of the rentier effect into three distinct mechanisms: a taxation, spending, and group formation effect. Consistent with Tilly () and my earlier discussion, the taxation effect argues that rents reduce a government’s need to tax their population for revenues and consequently leads their populations to demand less political accountability. In
Mahdavy (, ) first advanced the contemporary concept of the rentier state as a state that receives substantial rents from foreign individuals, concerns, or governments. Beblawi (, ) developed a more precise conceptualization of a rentier state. In such a state, () the rents are paid by foreign actors, () these rents accrue directly to the state, () only a few domestic actors are engaged in the generation of this rent (wealth), and () the majority are only involved in the distribution or utilization of the wealth. Chaudhry’s () seminal work broadens the theoretical and empirical scope, arguing how oil wealth affects governance both in oil producers and labor-exporting countries in the Middle East. For the latter, Chaudhry argues migration to oil producers from nonproducers and associated remittance inflows can affect governance in laborexporting countries (Ahmed provides causal evidence in support of this argument). For more recent treatments on the relationship between rents – mainly oil – and democracy in the Middle East, see Cammett et al. () and Smith and Waldner (, chapter ). While Ross discusses these effects with regards to oil wealth, the insights can be generalized to other rents.
Published online by Cambridge University Press
Analytical Framework
combination this can enhance the dictator’s control over society. The spending effect posits that rents allow governments to spend more on patronage, which in turn dampens latent pressures for democratization. These transfers may reward a small and critical group of regime supporters (e.g., military commanders, clerics, co-ethnic leaders) and/or fund social services that benefit the broader population. The former, which supplies targeted private benefits, can help dictators reduce the problem of authoritarian power-sharing by retaining the loyalty of key regime supporters. In contrast, if a government uses its rents to fund generous social benefits, it may be an effective instrument of authoritarian control. For example, Entelis () describes how the Saudi Arabian government uses its oil wealth to fund a generous welfare state that reduces pressures for democratization. Relatedly, the group formation effect argues that when rents are buoyant, the government can use its largesse to prevent the formation of social groups that are independent from the state and thus, be inclined to demand greater political representation. Heterodox Views Some scholars point to the mixed evidence between rents – particularly, oil wealth – and democracy/dictatorship to suggest that rents may have no independent (causal) effect on democracy or dictatorship. For example, examining data over two centuries, Stephen Haber and Victor Menaldo () find a tenuous relationship between oil wealth and dictatorship and argue that there is no “oil curse,” at least over the long run; and in some instances, oil rents may comprise a political “blessing.” In a similar vein, Benjamin Smith and David Waldner (, ) highlight a “scholarly dissensus” in existing scholarship linking oil to dictatorship. They maintain:
Moreover, the utility of this strategy is magnified in increasingly smaller power-sharing coalitions. This is because the per capita transfers can be higher to coalition members. Smith (, ) summarizes this logic: When coalition size is small, leaders can greatly reward their essential supporters by supplying private goods [targeted benefits]. However, as coalition size becomes large, private goods become more thinly spread, and leaders can more efficiently reward their supporters through the provision of public goods.” Bueno de Mesquita et al. () and Lake and Baum () provide empirical evidence of this trade-off. Ahmed (, chapter ) provides a numerical example. As discussed shortly, providing social benefits for the masses (e.g., public goods) may not be optimal for dictators. A more politically beneficial strategy for a dictator is to fund repression and targeted transfers (e.g., Smith ). A variant of the group formation effect builds on Putnam’s () work on social capital. Here rents can stifle the formation of civic groups that lie above the family but below the state. These civic groups are often important drivers for greater democratic accountability. Andersen and Ross () rebut this conclusion, finding that oil wealth may be a blessing in (some instances) prior to . However, after there is stronger evidence pointing to an oil curse. Smith and Waldner () reach this conclusion by pointing to several conceptual (measurement), empirical, and theoretical issues in extant scholarship.
Published online by Cambridge University Press
A Political Economy Approach
That when one looks closely at specific claims made in the literature, and the specific methodological warrants of those claims, the conclusion that a political resource curse exists appears less unassailable than would be the case when all the diverse findings are pooled together without sufficiently discriminating appraisal.
Smith and Waldner’s synopsis engages with the empirical evidence linking oil to politics across time and space (regions). Their analysis reveals that a political resource curse exists primarily in the Middle East and North Africa, which they attribute to factors related to institutions and coalitional politics. To the extent that the institutional legacy of Muslim conquest may explain the configuration of institutions and coalitions (e.g., relationship between the ruler, the military, and religious clergy) in many contemporary Muslim societies (as explored in Part II of this book), Smith and Waldner’s inferences may be compatible with those advanced in this book. C Conditional Theories of Rents and Political (In)stability While rents can help dictators mitigate their dual challenges of authoritarian control and power-sharing, the specific policies they permit (e.g., kleptocracy, repression) may incite political instability, such as revolutions and civil war. As noted earlier, dictatorship and its accompanying politics, often creates a climate conducive for violence. With this environment in mind, existing scholarship identifies several pathways linking rents and regime type to political instability. Variation in Institutions One pathway evaluates the effects of rents on political stability as institutions vary. In the “perils of unearned income,” Alastair Smith () formalizes these dynamics within the selectorate model of politics (Bueno de Mesquita et al. ). A critical parameter in the selectorate model is the size of the “winning coalition” (W) required for a leader to remain in power. Small W systems characterize dictatorships, where the leader’s optimal strategy is to
For the latter, they study the relationship between oil and dictatorship (and the prospects of democratic transitions) in four regions – Latin America, the Middle East, Africa, and Southeast Asia. Here I focus on civil war, but rents may also incite war between states. See for example, Caselli, Morelli, and Rohner (), and Colgan (). A related explanation focuses on the strength of opposition forces and the timing of oil wealth. Focusing on oil producers, Benjamin Smith () posits that variation in the strength of opposition forces (to the government) in late industrializers can affect how rulers manage their oil wealth with divergent political implications when oil wealth declines. Where opposition forces are strong in late industrializers (e.g., Indonesia), the ruler undertakes policies (e.g., more effective taxation of non-oil income) that reduce its probability of losing power when oil income declines. In contrast, where opposition forces are weak in late industrializers (e.g., Iran), the ruler is prone to adopting policies (e.g., patronage to selected/narrow groups) that increase its probability of losing power when oil income declines.
Published online by Cambridge University Press
Analytical Framework
provide targeted benefits to coalition members and to reduce the provision of public goods (e.g., welfare transfers, rule of law). One of Smith’s key findings identifies a non-monotonic relationship, where the effect of rents on revolutionary threats varies with the size of W. Referring to rents as “free resources,” Smith’s () comparative statics reveal that “large coalition systems are relatively more efficient than small coalitions at converting free resources into societal benefits and economic activity because free resources fund public goods rather than corruption, graft, and kleptocracy. From the perspective of citizens outside of the winning coalition, if the level of free resources rises, then the desirability of a large coalition increases relative to a small coalition system . . . This encourages the citizens to support revolution” (). Of course, if the costs of rebellion are prohibitively high, the availability of rents may not necessarily induce revolution. As Smith (, ) observes, these costs can be large in small W systems: The citizens’ decision to rebel is affected by two factors: the difference between their welfare under the current institutions relative to the welfare they expect under postrevolutionary institutions and the probability of the revolution succeeding. When coalition size is small, leaders provide few public goods. This leaves the citizens extremely dissatisfied with their lot relative to what they could obtain under postrevolutionary institutions. This gives them a great desire to rebel. However, the citizens also factor the probability of success into their decision to rebel. When the coalition is small, leaders provide few of the public goods which help revolutionaries coordinate and organize. As a result, when W is small, citizens choose not to rebel; not because they don’t want change, but they are unlikely to succeed.
The confluence of these comparative statics generates an inverted U-shaped relationship between rents, coalition size, and revolutionary threats. In small W and large W systems, rents are less likely to induce revolution. However, in the intermediate range of coalition size, greater rents incite revolution. Of course, as Smith () also shows, the availability of free resources can empower governments to finance their repressive capacity to crush rebellion.43 In conjunction with the U-shaped relationship, this policy response implies that relatively robust dictatorships (e.g., monarchies or personalist dictatorships with smaller W) can effectively deter revolutionary threats induced by rents. In short, rents can monotonically increase political stability in dictatorships. In subsequent research, Bruce Bueno de Mesquita and Alastair Smith () provide empirical corroboration for this prediction.
Figure in Smith () demonstrates this. In Smith’s model, governments can also contract their supply of public goods which can make it more difficult for individuals and groups to collectively organize. Note, the non-monotonic effect formalized by Smith () and empirically supported by Bueno de Mesquita and Smith () varies by coalition size. It does not vary by changes in the level of rents. This distinction is important as the model I introduce in Section . generates nonmonotonic effects that vary by rents and coalition size. I conceptualize the latter as cohesive institutions.
Published online by Cambridge University Press
A Political Economy Approach
Variation in Rents Another pathway evaluates the effect of institutions on political stability as rents vary. Here, the existing literature identifies two divergent effects associated with changing levels of rents on civil war. The first is the converse of the politically stabilizing role of rents discussed in the previous section. Here a reduction in rents weakens the government’s capacity to thwart insurgency and/or funds to buy off regime supporters and rivals. Relatedly, rents can mitigate civil conflict by providing a “cushion” to government spending during slumps in the “private” economy. One such example is a well-timed foreign aid package (or opportune increase in commodity prices) that can provide governments with the necessary financial windfall to make rebellion less attractive for opposition groups (Savun and Tirone ). Of course, if these rents disappear the prospect for civil war may be elevated. Citing the commitment problem articulated by Robert Powell (), Richard Nielsen et al. () argue that a severe drop in government revenues (e.g., stemming from a rapid decline in foreign aid) can weaken the government’s capacity to effectively buy off rebel groups and contribute to “inefficient conflict.” They leverage matching techniques to provide cross-national corroboration, documenting how negative aid shocks accelerate the likelihood of civil war. The other path links rents to heightened conflict. This account posits that rents increase the size of the government “pie,” and if there are multiple groups dividing the pie, rent-seeking can contribute to increased fighting over it (Grossman ). Several studies formalize this central idea, identifying the critical role of preexisting political institutions in shaping conflictual behavior. For example, Aaron Tornell and Philip Lane () formalize a “voracity effect” by which an increase in a government’s fiscal resources (e.g., due to a natural resource boom, foreign aid windfall) incites groups within societies with “weak” institutions to compete violently for and squander the windfall and reduce economic growth. Relatedly, Jacob Svensson () develops a repeated game with stochastic shocks in unearned income, where the increase in rent-seeking arises due to coordination failure across interest groups. This coordination failure is often heightened in ethnically diverse societies. For example, Roland Hodler () constructs a model where a surge in rents fosters Cournot-type competition in the form of group fighting in ethnically fragmented societies with a corresponding elevated risk of conflict and lower growth. In these models, political institutions are unable to get incumbent governments to credibly commit to allocating state resources to stave off violence. As I discuss below, this commitment problem is characteristic of societies with less cohesive political institutions. More generally, the conditional effects of rents on political stability (associated with the models above) fall under a broader theory of an “institutions curse” in which rents received in societies with less cohesive institutions are
Published online by Cambridge University Press
Analytical Framework
prone to pernicious political economy (Menaldo ). As discussed above, in some of these accounts a reduction in rents leads to civil war, while in others higher rents do. Moreover, the likelihood of civil war in the presence of an increase or decrease in rents often depends on the quality of “cohesive political institutions.”
D Cohesive Institutions But what are cohesive political institutions? Cohesive institutions describe how an incumbent government “shares” state resources within its territory (Besley and Persson a). In general, societies with more cohesive institutions allocate a more equitable share of state resources to groups that are out of power (i.e., opposition groups). As Besley and Persson (b) note, real-world examples of societies with more cohesive institutions are those with “more minority protection through a system of constitutional checks and balances, through a parliamentary form of government, or through a proportional electoral system” (). This categorization clearly situates democracies as being more cohesive and dictatorships as being less cohesive. In formal models, the degree of cohesive institutions can be captured with a bounded (continuous) parameter; for example, θ 2 ½0, 1 where a higher value of θ corresponds to more cohesive institutions. Thus, if θ ¼ 1=2, then state resources are transferred equally between supporters of the incumbent and the opposition. In contrast, if θ ¼ 0 the incumbent allocates all the state’s resources to its supporters and none to supporters of the opposition. Empirically, scholars have documented important determinants/correlates of cohesive institutions. For instance, governments in more ethnically, religious, and/or linguistically diverse societies tend to supply higher amounts (shares) of public goods to individuals that share the same group affiliation/identity as the incumbent (Alesina, Baqir, and Easterly , Baldwin and Huber , Habyarimana et al. ). Thus, many “fractionalized” societies exhibit less cohesive political institutions.
Although he does not use the terminology of cohesive institutions, Victor Menaldo () provides an excellent account of an institutions curse associated with natural resource rents. Faisal Z. Ahmed () develops a parallel account of an institutions curse with various forms of international capital (e.g., remittances, foreign aid, and investment). Relatedly, correlates of domestic political institutions can also generate a broader “conditional resource curse.” This may arise from other features of a country’s domestic political economy, including the strength of opposition forces (Smith ), the level of income inequality (Dunning ), ownership structure of natural resources (Luong and Weinthal ), the quality of electoral institutions (Mahdavi ), and degree of economic integration (Kurtz and Brooks ). An alternative categorization of cohesive institutions can situate it along Acemoglu and Robinson’s () concept of “inclusive” institutions. As these scholars document, societies with more inclusive institutions share power and state resources (public finances) across various societal groups.
Published online by Cambridge University Press
A Political Economy Approach
Subsidies and transfers (% Expenditures) 20 40 60 80
SPN GMY CZR SWZ NTH UKR VEN MLD
BLR
0
ITA CAN POL SLV SER SWD CHL BUL KZKRUS MON ROK USA URU ALB SLO BEL BRA AZE MAC CRO UGA HUN EST GRG UKG BOS FIN RUM MAL IND NOR ALG SAF LAT EQG GRC AUS FRN IRE TUN ARM INS TAJ NEW JPN COLTUR AUL BOT KUW BOL IRN TRIISR ETI BNG NIC MAS BUI PAN POR MOR PER DEN PAR KYR SRI NIG PNG CYP COS LES GUALEB SEN GHA ZIM FJI CEN SIN SYR CAM EGYANG QAT SWA JOR TOG LBR ZAM THI PHI SAL HON PAK RWA NAM GAM SIEBEN MAG KEN MLI BAH LAO BFO OMA BHU AFGIVO ZAI JAM
0
.2
.4 .6 POLITY (normalized)
.8
1
FIGURE . Democracy and public goods Note: Country average based on author’s calculations. Data on subsidies and transfers is from the World Bank ().
As noted above, another important correlate of cohesive institutions is a society’s quality of democracy: Societies with more cohesive institutions tend to exhibit a higher quality of democratic governance and institutions. This association is unsurprising as many theories predict that governments in democracies will allocate a greater share of their resources to public goods, such as expenditures on welfare payments and subsidies to the population (e.g., Acemoglu and Robinson , Bueno de Mesquita et al. , Lake and Baum , Olson ). Given this incentive, democratic governments correspondingly tend to exert greater “tax effort”: They derive greater share of their revenues from taxation (Besley and Persson a). Using data from the World Bank, Figures . and . corroborate these conjectures for a global cross section of countries. Looking first at government spending, Figure . shows that some of the least democratic countries (e.g., Bhutan, Zaire) allocate less than percent of their expenditures on social transfers. In sharp contrast, the most democratic countries, on average, spend
Greater tax effort is also exerted in resource-abundant democracies, such as Norway and the United States.
Published online by Cambridge University Press
Analytical Framework
60
AUL JPN NEW INS
USA CAN
SAF
Tax income (% revenue) 20 40
QAT ANG
TRI
MAL
ZIM
PNG
TUR
UKG DEN BEL IRE VEN MEX LBR FJI NAM ISR ZAM ITA SYR SPN KEN JAM SIN VIE ROK BOT SLO NTH FIN SWA KZK THI BRA IND AUS GUA CHN NIG MON COL LAO CYP NOR MORALG HON AZE SAL POR POL GHA EQG TOG RUM FRN SEN IVO LES PANPERCHL ETI BEN KYR HUN ZAI MYA SIE EGY RWA CZR GMY TUN BUL ARM GRC SWD BFO PAK NIC EST SWZ MAS BUIUGA SLV BNG LEB SRI COS UKR MLI MAC ARG LAT IRN OMA CEN MAG NEP SER JOR BLR GAM PAR CRO GRG KUW BHU URU ALBCAM AFG RUS BOL BAH TAJ MLD BOS
0
PHI
0
.2
.4 .6 POLITY (normalized)
.8
1
FIGURE . Democracies exert greater tax effort Note: Country average based on author’s calculations. Data on tax income is from the World Bank ().
over percent of their budgets on social transfers. This finding is not driven by greater per capita wealth in more democratic societies. The positive relationship between democracy and social transfers remains statistically robust in a regression that controls for log GDP per capita, as well as with other common measures of democracy, such as constraints on the executive (Acemoglu and Johnson , Gledistch and Ward ). Relatedly, Figure . shows that governments in more democratic polities exert greater tax effort. In this figure, the positive association between the share of government revenue collected from income taxes and a higher normalized POLITY score is consistent with predictions from several prominent political economy models of public finance based on the median voter theorem (Meltzer and Richards ), electoral incentives (Lizzeri and Persico ), credible commitments to redistribution (Acemoglu and Robinson ), and selectorate theory (Bueno de Mesquita et al. ). Taken together, the purpose of the Figures . and . is to simply illustrate how variation in democracy captures the core feature of cohesive institutions in the contemporary era: the extent to which state resources are equitably
Calculated percentage based on average across countries with a normalized polity score greater than or equal to ..
Published online by Cambridge University Press
A Political Economy Approach
distributed across individuals (groups) that support and oppose the incumbent. That said, it is worth noting that cohesive institutions may also be correlated with (and endogenous to) many historical and contemporary conditions within a given society, such as its climate/geography (), precolonial state capacity (Hariri ), “accumulated democratic experience” (Gerring et al. ), factor endowments (Acemoglu and Robinson , Engerman and Sokoloff , Menaldo ), colonial legacy (Olsson ), and structure of the international system (Boix ). And, as the cited scholars demonstrate, these conditions are in turn highly correlated with democracy in the contemporary period. The covariance of these conditions implies that isolating their separate effects poses a critical empirical challenge; a task that Part II of this book addresses in greater detail. As a preview, this is achieved through the research designs (e.g., using panel data, instrumental variables) and choice of control variables. For example, since many correlates of cohesive institutions are country-specific and timeinvariant (e.g., precolonial state capacity and history, colonial legacy, climate, and geography), in econometric analysis with panel data, they can be accounted with country fixed effects. Similarly, controlling for year (or period) fixed effects can capture the effects of important temporal factors that affect all societies, such as the structure of the international system.
E Theoretical Puzzle The discussion in the preceding sections generates an important theoretical puzzle stemming from three core quandaries: () Rents can be a source of political stability, especially as a means for dictatorships to increase their repressive capacity and provide transfers to their key supporters. () Some theories, however, predict that higher levels of rents heighten the propensity for civil war, while others predict the opposite: A drop in rents makes civil war more likely. () These politically destabilizing effects tend to be elevated in societies with less cohesive societies. Surprisingly, these divergent paths have yet to be reconciled in a unified analytical framework. Thus, I seek to answer the following theoretical puzzle: Under what conditions should an increase or decrease in rents lead to political violence? In collaborative work with Daniel Schwab and Eric Werker, we advance a formal model to reconcile this tension in existing theoretical work and also
For example, the use of instrumental variables can help mitigate concerns with omitted variables (Angrist and Pischke ).
Published online by Cambridge University Press
Analytical Framework
provide a peaceful pathway for democratization (Ahmed, Schwab, and Werker ). In particular, the model derives plausible equilibrium conditions under which variation in rents and the quality of cohesive institutions can lead to differential trajectories of democracy (peace), dictatorship (repression), and/or civil war in a unified theoretical framework. Building on the now canonical model of state capacity and political violence developed by Timothy Besley and Torsten Persson (, , a,b), the model is driven by competition for political power between two groups within a country – an incumbent and an opposition – where the winner distributes state resources disproportionately to its own members depending on the cohesiveness of institutions. Consistent with the discussion in the previous section, in the model more cohesive institutions correspond to a more equitable distribution of state resources across society by the incumbent. The model assumes that it is not possible to credibly commit in advance to a more equitable redistribution, so both groups have an incentive to use violence to increase their likelihood of gaining or holding power. In doing so, this assumption captures an important dimension of politics in democracies and dictatorships (Svolik ). Like Besley and Persson’s formulation, my model generates three outcomes: peace/ democracy (no violence), repression (one-sided violence inflicted by the incumbent), and civil war (two-sided violence involving the opposition fighting the incumbent). While the model identifies nontax income (rents) as an important source of these political outcomes, I am careful to capture other important features that can affect politics, such as per capita income (through wages). The model extends the framework from Besley and Persson to allow for the possibility that the incumbent can deter attacks by the opposition when government resources are sufficiently high. The model is essentially a one-shot sequential game, and when rents are moderate, an incumbent will respond to an insurgency with only moderate violence, which encourages an attack by the insurgents. When rents dwarf the costs of fielding a large army, the incumbent would be willing to spend enough to squash an insurgency, and the insurgents are deterred from attacking. To preview my predictions, whereas existing models predict a hierarchy of violence in which higher levels of rents monotonically increase the likelihood of political stability (e.g., Smith ) or conflict in dictatorships (e.g., Besley and Persson b), I provide conditions for differential and non-monotonic
In the analysis to follow, references to “my model” (or using “I”) reflect collaborative work with Daniel Schwab and Eric Werker. In extensions to the model (e.g., where players may incorporate “common knowledge” acquired via repeated interactions), there may be Pareto dominant equilibria. The degree of cohesion can be shaped by several factors, including historical legacy (e.g., colonial history, premodern state capacity), factor endowments (e.g., land inequality, natural resources), and ethnolinguistic and religious fractionalization. The empirical analysis in subsequent chapters will discuss and control for these factors.
Published online by Cambridge University Press
A Political Economy Approach
effects. My model generates two key predictions. First, consistent with many existing accounts I show that higher levels of rents raise an autocrat’s repressive capacity and ability to maintain authoritarian rule. Second, upon the reduction of these rents (to medium or low levels), these regimes are susceptible to challenges from opposition groups. In particular, the flexibility of the model allows for two paths from a reduction in the level of rents: a heightened incidence of civil war or peaceful transitions to democracy. The particular outcome hinges on the preexisting cohesion of political institutions. Violence is more likely in societies where preexisting institutions are less cohesive, while a peaceful transition to democracy is more likely in societies whose preexisting institutions are more cohesive. To the best of my knowledge, this differential and non-monotonic effect is novel in the theoretical literature on rents and political stability. For interested readers, I present this model in the next section. Section . distills the model’s main empirical implications and previews the research designs to evaluate them.
. In this section, I advance a parsimonious game theoretical model to illustrate how () higher levels of rents can entrench dictatorship and () the conditions under which their decline can engender civil conflict or democracy. For the latter, I show that countries with less cohesive institutions are prone to conflict when rents decline, while those with more cohesive institutions can transition peacefully to democracy. A Basic Setup As already noted, my model closely follows a series of theoretical formulations of state capacity and political violence developed by Besley and Persson (, , a,b). In any society, there are two groups ðI, PÞ, each with a population normalized to . Each member of the population earns a wage ðwÞ which is set to . The incumbent (denoted, I) starts the period in power, while the other group ðPÞ is in the opposition. Ending the period in power allows the incumbent to favor its own group when distributing state resources ðRÞ, which is an exogenous parameter. State resources ðRÞ can be thought of as nontax government income (rents) that may come from domestic sources (e.g., income from natural resources,
For instance, Bueno de Mesquita and Smith (), Dunning (), Menaldo (), Morrison (), and Ross (, ) do not generate these predictions. Since the model’s comparative statics will focus on analyzing variation in cohesive institutions and rents setting wages to simplifies the subsequent mathematical derivations. That said, allowing wages to vary would not affect the comparative statics analysis of institutions and rents.
Published online by Cambridge University Press
Analytical Framework
such as oil) and from abroad (e.g., foreign aid, sovereign borrowing). In certain settings, state resources could also include external security guarantees – “geopolitical rents” – available to the incumbent, such as the presence of foreign troops in their territory and diplomatic cover to pursue repressive domestic policies (e.g., imprison political opponents). Unlike material resources such as revenues from natural resources or from foreign aid, geopolitical rents may be tied exclusively to safeguarding the incumbent and are not shared with the opposition. For example, the presence of US troops in certain Persian Gulf countries may serve to better ensure US foreign policy preferences (e.g., a stable supply of oil, peaceful relations with Israel) and tacitly help support the incumbent regime’s continued rule. If that incumbent loses power to the opposition, the US may withdraw military support; in which case, geopolitical rents would be unavailable to the opposition. Both the incumbent and the opposition have access to a violence technology: an army of size A 2 ð0, 1Þ that costs wA to raise. I assume that there are two sizes of armies available for the incumbent, L and H (for low and high), and one for the opposition, which is the same as the incumbent’s low level. Limiting army sizes to discrete options is primarily for tractability, but it does also have some real-world applicability; for example, a government may decide between allowing rebels to hold zero or one city, or they may choose whether or not to use its air force. Allowing the incumbent access to violence levels , L, and H means that there is a possibility that an incumbent’s response to violence is dependent on the level of rents. I assume that the opposition first chooses whether to attack or not, which the incumbent observes, and then the incumbent chooses its own level of violence. If the investment needed for the incumbent to field a large army is sufficiently close to its investment for a small army, then the opposition will never attack. The reason is that the incumbent would employ a large army against an attack because the expenditure would be small relative to the gain
I abstract from the important question of intragroup inequality and assume that all members of a group receive an equal share of the group’s resources. I provide evidence consistent with this conjecture in Chapter . Egypt is an example in a nonoil-producing exporter. During the Arab Spring, the US government effectively withdrew support from the democratically elected Morsi government after the demise of the US-backed Mubarak regime. Upon the resumption of military dictatorship, the USA resumed its support of the regime. This is an important modification to Besley and Persson (): In that paper, the incumbent can choose to attack or not, but has no flexibility on army size if it does attack. In that paper and in this chapter, assumptions on the parameters and the conflict function ensure that in equilibrium the incumbent always responds violently to an insurgency. In Besley and Persson (a,b), the choice of violence is continuous, so the incumbent’s choice of violence can depend on unearned income. However, other assumptions in that paper rule out the possibility that opposition violence increases as unearned income decreases. This timing is the same as Besley and Persson () but differs from Besley and Persson (b) where the two groups attack simultaneously.
Published online by Cambridge University Press
A Political Economy Approach
from retaining power, and so the opposition would be deterred. As I show in the model’s proofs (in the chapter’s Appendix), the relevant threshold is that there is never two-sided violence when H < 3L. Since opposition groups do sometimes mount an insurgency, I assume H > 3L. This ensures that two-sided violence does occur for some values of R. Power changes hands stochastically according to the conflict function γ AP , AI 2 ½0, 1, where either side can improve their chances by investing in violence. The following specifies the functional form as: 8 I P > >0 A > A P I < I P γ A ,A ¼ 1 A < A > > : 1 AI ¼ AP 2 This functional form means the group choosing the higher level of violence wins power with certainty, and if they choose the same level, then power changes hands with probability 12. At the end of the period, the level of rents R is divided between the two groups, and utility is linear in consumption, meaning I can treat the agents as risk neutral. The group that ends the period out of power gets θR, and the inpower group gets ð1−θÞR, where 0 θ 12 represents equality of institutions. Importantly, a higher value of θ implies institutions are more cohesive. Note that θ is assumed to be “exogenous” in the model, which is unlikely to be the case empirically. Cognizant of this concern, Part II of this book strives to establish a plausibly exogenous determinant of cohesive institutions that is unrelated (“causally prior”) to contemporary levels of rents. An important distinction between the two groups is thatthe group that started in opposition pays for their entire military expenditure AP , while payment for the incumbent group comes out of the public purse before resources are distributed. Then the game ends and payoffs are realized.
It would be straightforward to introduce incumbent advantage by assuming that power changes hands with probability α when both groups show the same level of violence for some α < 12, but the results do not meaningfully change, so I retain the simpler formulation. Specifically, I use the percentage of territory conquered by Islamic armies (corresponding to contemporary country boundaries) during the period – CE to “explain” contemporary levels of democracy. This percentage of conquered territory is clearly uncorrelated with timevarying rent streams in the modern era. The results would be unchanged if financing for the incumbent’s army came from a combination of public and private funding. It is worth noting that the equilibrium outcomes I identify in this single-shot game are also subgame perfect Nash equilibrium (SPNE) in a repeated game where each generation cares about other generations (to some degree). It can be shown that in a repeated game, the equilibrium outcomes identified in the single-shot setting comprise the “worst case” equilibrium. Future work could investigate the existence of Pareto dominant SPNE in the repeated game. For example, if one were to take the standard approach to repeated games and presume that the history of play is common knowledge between the players throughout the game, then one could
Published online by Cambridge University Press
Analytical Framework Opposition Peace
Attack
Incumbent Peace
Payoffs:
]
FIGURE
. Game tree
, ]
Incumbent Repression
[q(R–L), (1–q)(R–L)]
Attack with L
[
(
) (
,
)
Attack with H
]
[q(R–H)–L, (1–q)(R–H)]
The two groups’ respective payoffs are the sum of their wage, their expected receipts of nontax income, and for the opposition, the cost of financing an army. Since the wage is always one by assumption, it is omitted for clarity. The payoff for the incumbent is: V I AP , AI ¼ 1 γ AP , AI ð1 θÞ R AI þ γ AP , AI θ R AI and the opposition’s payoff is: V P AP , AI ¼ 1 γ AP , AI θ R AI þ γ AP , AI ð1−θÞ R AI −AP :
B Optimal Investment in Violence In this section I solve for the optimal choice of violence for each group. Since this is a one period game with two actions, I start at the end of the game and work backward. Figure . illustrates the corresponding game tree, where the opposition moves first. The last row in the figure states the payoffs for each series of actions, with the opposition’s payoff listed first in parentheses. I first consider the case where the opposition does not attempt an insurgency. The incumbent then chooses between peace and using one-sided violence, which is interpreted as repression. A peaceful incumbent realizes a payoff of: 1 1 R V I ð0, 0Þ ¼ ð1−θÞR þ θR ¼ 2 2 2
use the Folk Theorem to establish the existence of Pareto superior SPNE. I leave this interesting line of inquiry to future work.
Published online by Cambridge University Press
A Political Economy Approach
TABLE
. Incumbent’s optimal choice in political violence
R
Response to peace
Response to violence
L 1−2θ
ð1−θÞ R 2L1−2θ
L
L
L
L
H
0R L 1−2θ
2Lð1−θÞ 1−2θ
1−θÞ−L R 2H ð1−2θ
1−θÞ−L R > 2Hð1−2θ
Note: The first column shows a range of values for R. The next two columns show the incumbent’s optimal response to peaceful or aggressive opposition. Recall, L is a low deployment of armed forces. H is a high deployment of armed forces.
while a repressive incumbent is guaranteed to retain power, generating a payoff of: V I ð0, LÞ ¼ ð1 θÞðR LÞ: Note that the conflict function means the incumbent would never use a high level of violence against a peaceful incumbent, because high or low violence both guarantee success. The main aim of the model is to clarify the impact of rents on violence choices, so I allow R to vary and hold all other parameters fixed. The incumbent will be peaceful in response to peaceful opposition when: V I ð0, 0Þ V I ð0, LÞ which simplifies to:
ð1−θÞ Peace is the incumbent’s optimal response to peace if 0 R 2L1−2θ .
ð1−θÞ Low violence is the incumbent’s optimal response to peace if R > 2L1−2θ .
I assume that an indifferent group will break ties by choosing the lower level of violence. Similarly, if the opposition is violent, then the incumbent chooses between peace, low violence, and high violence. In the appendix to this chapter, I show Table . presents the incumbent’s optimal response to a peaceful or violent opposition: I now solve for the opposition’s optimal choice for every value of R. If L , then the opposition knows the incumbent will be peaceful, 0 R 1−2θ regardless if it attempts an insurgency. Thus, the opposition chooses between V P ð0, 0Þ ¼ 12 R and V P ðL, 0Þ ¼ ð1 θÞðR LÞ. Over the relevant range of R, the former value is larger; meaning that when L both the opposition and incumbent will be peaceful. This process can 0 R 1−2θ be continued for every value of R, by solving the incumbent’s response to peace and violence, and then determining which choice maximizes the opposition payoff. Table . presents the payoffs (see the chapter’s Appendix for the proofs).
Published online by Cambridge University Press
Analytical Framework TABLE
. Incumbent and opposition optimal choices
Interval
R
Inc. response to
Inc. response to L
Opp.
Inc.
L 0 R 1−2θ
L 1−2θ
L
2Lð1−θÞ 1−2θ
L
L
L
2Lð1−θÞþL 1−2θ
L
L
L
L
L
H
H
ð1−θÞ R 2L1−2θ
R>
1−θÞþL < R 2Lð1−2θ 1−θÞ−L < R 2Hð1−2θ
2H ð1−θÞ−L 1−2θ
Note: The first column is a label for the range of R (unearned resources) being described. The second column is that range. The next two columns are the incumbent’s response to a peaceful or violent opposition. The final two columns are the actions that the opposition and the incumbent would actually take. The first two rows are both labeled as Interval because the outcome (both groups peaceful) is the same in both.
As R ranges from to infinity, there are four relevant intervals. In the first interval, both groups are peaceful because the prize for winning power is too small to justify fighting. There are two rows in Table . that make up Interval because the incumbent would be peaceful in response to rebellion in the first row and violent in response to rebellion in the second row. But in both rows the actual outcome would be peace on both sides. The incumbent will be tempted to use violence at a lower threshold of R than the opposition because the incumbent funds its army using public funds, and in Interval the opposition is peaceful while the incumbent uses the smaller army size against peace or violence, so the result is repression. In Interval the opposition would still always use a small army but now R is sufficiently large that the opposition is also willing to use, and both groups use low violence (civil conflict). In Interval R has increased to the point where the incumbent would employ a high level of violence in response to rebellion, but low violence in response to peace. It is disastrous for the opposition if the incumbent uses a high level of violence, because the incumbent is guaranteed to hold power and uses up state resources as well. This discourages the opposition from attacking, so the result is low-level repression again, the same as in Interval . These dynamics may arise in natural resource dependent countries where an incumbent and opposition fight over the rents (e.g., Angola, South Sudan, Colombia). This result – that violence can fall as R increases – is driven by two factors. First, since the incumbent has two choices of army size, as R rises to Interval the incumbent’s response to an attack can increase from a small to a large
As I demonstrate in Chapter , higher resource dependence on oil revenue is not a robust determinant of high intensity civil war in the Persian Gulf.
Published online by Cambridge University Press
A Political Economy Approach
army size, which can deter the opposition. Second, the result stems from the sequential nature of the groups’ decisions, because the incumbent cannot respond to the opposition if they act simultaneously.
C Transitions to Democracy or Civil War The aim of Table . is to depict how the incumbent’s and opposition’s optimal choices change as R varies, holding the size of armies (L, H) and cohesive institutions ðθÞ fixed. Figure . also shows optimal choices but allows θ to vary. Specifically, I plot the four intervals depicted in Table ., while θ ranges from . and . and L and H remain fixed at . and . respectively. Going forward, I use the term “transition” to describe how changes in the model’s exogenous parameters (R or θ) leads to changes in violence (e.g., movements from repression to civil war, repression to peace). Figure . demonstrates how a country can transition between peace, repression, and conflict as rents and cohesive institutions vary. In general, higher levels of rents tend to increase the intensity of political violence: Holding fixed the level of θ, “moving up” the y-axis makes civil conflict more likely. This monotonic effect is consistent with the predictions of Besley and Persson (, , a,b). That said, Figure . demonstrates the possibility of nonmonotonic effects whereby a reduction in rents can make peace or civil conflict more likely. Whether a country experiences peace or conflict depends critically on the preexisting level of institutional cohesion. Transition to Democracy To observe a transition to democracy associated with a decline in rents, consider a country initially in Interval (low repression). In this instance, a reduction in rents can shift the country to Interval in which the incumbent decommissions its army (i.e., AI changes from L to ) and peace is achieved: a condition conducive for democracy. Moreover, observe that this peaceful transition to democracy is more likely to occur at higher levels θ. Since θ is “exogenous” in this model, this implies that a reduction in rents in societies with a higher preexisting (exogeneous) level of cohesive institutions can experience a peaceful democracy.
In Besley and Persson (, b), it is not possible for a decrease in rents to lead to an increase in violence. The incumbent in Besley and Persson () has only one possible army size so the deterrent effect observed in my model cannot occur. The two sides choose army size simultaneously in Besley and Persson (a), so there cannot be an exactly analogous deterrence effect, and the authors’ assumptions limiting the strategic complementarities and substitutabilities in the conflict function ensure that both groups’ army choices are weakly decreasing as R decreases. These numerical values are chosen for illustrative purposes, and other examples give similar results.
Published online by Cambridge University Press
2.5
Analytical Framework
R
1.5
2
Repression (Interval 4)
1
Conflict (Interval 3)
0.5
Repression (Interval 2)
0
Peace (Interval 1)
0.2
0.25
0.3 q
0.35
0.4
FIGURE . Political violence as a function of cohesive institutions ðθÞ and rents (R) Notes: This figure simulates the violence outcomes as a function of θ (cohesiveness of institutions) and R (rents). The values of L and H (fraction of workforce deployed for low and high army size) are set to . and ., respectively. In Interval , both groups are peaceful. In Interval , the incumbent uses low violence in response to peace, and would have used low violence in response to violence from the opposition. Both groups use low violence in Interval . In Interval , the incumbent uses high violence in response to an attack, so the opposition is deterred and the result is peace from the opposition and low violence from the incumbent (repression). The values of R range from to infinity.
Transition to Civil Conflict Of course, Figure . also illustrates that a reduction in rents can also make civil conflict more likely. To see this, consider a country initially in Interval of “low repression.” In this case, a reduction in R can shift a country to Interval in which both the incumbent and opposition deploy small armies and engage in low intensity warfare. Moreover, being in Interval in the first place is more likely to occur at lower levels of θ. And since θ is assumed to be exogenous in the model, this implies that a reduction in rents in societies with a lower preexisting level of cohesive institutions may heighten the propensity for civil war. To the best of
In Chapter , I argue these features capture important dynamics potentially explaining why many Muslim foreign aid recipients experienced heightened internal political strife – characterized at the extreme with high intensity civil war – in the s and s and remained nondemocratic.
Published online by Cambridge University Press
A Political Economy Approach
my knowledge, these differential and non-monotonic effects associated with a decline in rents – peaceful transition to democracy or heightened propensity for civil war – is novel in the theoretical literature on rents and political stability. Preserving Peace (Democracy) Finally, Figure . also shows that more cohesive institutions can preserve peace at higher levels of rents: a finding consistent with existing models (e.g., Besley and Persson b). When θ ¼ 0:2 there will be peace only when R 0:2667, but when θ ¼ 0:4, an incumbent will not be tempted to repress the opposition as long as R 0:6. This behavior stems from the expectation that the incumbent will be treated fairly even if it loses power, an expectation that underlies the peaceful transition of power that occurs in democracies. Moreover, this means that if two countries are initially both in Interval (repression) with the same level of R, the country with a lower θ will require a larger decline in R to shift to Interval (peace/democracy). In other words, more cohesive institutions can provide some protections against rent-driven civil conflict; a finding consistent with existing models of rent-seeking and political (in)stability (e.g., Lane and Tornell , Svensson ).
. A Propositions The discussion in the preceding section is summarized in the following three empirical implications and two corollaries: Proposition : Rents and Dictatorship Higher levels of rents can foster dictatorship (and other correlates, such as political repression), particularly in societies where preexisting institutions are less cohesive. This corresponds to moving from Interval to Interval or . Proposition : Reduced Rents and Political Stability Reduced levels of rents can shift a country from conflict or repression into democracy (i.e., movement from Intervals – to Interval ), but cannot move a country out of democracy. Proposition : Reduced Rents and Political Transitions as Cohesive Institutions Vary The threshold R below which democracy exists is greater where institutions are more cohesive (i.e., θ is greater). Empirically, this can imply two divergent outcomes that hinge on the cohesion of preexisting institutions: Corollary A: A reduced level of rents is more likely to foster civil conflict in societies where preexisting institutions are less cohesive. This corresponds to a movement from Interval to Interval .
Published online by Cambridge University Press
Analytical Framework
Corollary B: While the same reduction in rents is more likely to lead to democratization in societies where preexisting institutions are more cohesive. This corresponds to a movement from Interval or to Interval . B Empirical Evaluation: A Road Map I evaluate these propositions in Parts II (“Conquest”) and III (“Rents”) of this book to show why so many Muslim majority societies are prone to dictatorship and civil conflict. In evaluating these propositions, I am cognizant that both rents and institutions are endogenous. Accordingly, each part will articulate the challenge posed by endogeneity and present identification strategies to better establish causal relationships. I briefly preview the research designs below. Conquest In Part II (“Conquest”), I establish a plausibly exogenous source of variation in cohesive institutions for contemporary Muslim societies. This is relevant as the level (quality) of cohesive institutions is assumed to be exogenous in my formal model. In Chapters –, I use qualitative historical evidence and cross-national statistical analysis to document the institutional legacy of Muslim conquest. The empirics strive to show how the initial period of Muslim military conquest following the death of Prophet Muhammad (– CE) established authoritarian political structures and then trace its subsequent trajectory of centralized autocracy in “conquered” territories through to the present. Since the process of military conquest and consolidation of centralized autocracy predated the discovery and distribution of rents (e.g., oil, receipts of foreign aid), the historical legacy of these institutions can serve as a source of exogenous variation in cohesive institutions in many contemporary Muslim societies. Thus, the institutional legacy of Muslim conquest may help explain the crosssectional difference in democracy between Muslim and non-Muslim societies; that is, the persistent democratic deficit depicted in Figure .. I interpret this finding to imply that the legacy of Muslim conquest fostered less cohesive
For example, the extraction of natural resource and subsequent generation of rents can be affected by domestic institutions. Menaldo () documents the higher levels of foreign investment and subsequent resource extraction in oil-producing democracies. Thus, the direction of this endogeneity suggests the effect of oil wealth on autocracy may be downward biased. Chapter discusses the emergence and persistence of an “Islamic equilibrium.” The chapter also presents cross-national causal evidence using a novel differences-in-differences research design. In Chapter , I examine the process of reconquest in medieval Spain and adoption of Islam in Indonesia and Malaysia (through trade) to document how the process of military conquest and subsequent rule in conquered territories mattered for the introduction and persistence of the classical Islamic equilibrium.
Published online by Cambridge University Press
A Political Economy Approach
institutions in Muslim societies (exposed to conquest) in the contemporary period. Rents This exogenous source of variation in preexisting cohesive institutions in the modern period allows me to evaluate how variation in rents affects variation in dictatorship and civil war in contemporary Muslim and non-Muslim societies over time (as depicted in Figures . and .). Thus, I can directly test propositions – and corollaries A and B. In Part III of this book (comprised of Chapters –), I demonstrate how temporal variation in rents – oil revenues in oil producers and foreign aid in non-oil producers – explains patterns of dictatorship and civil conflict in Muslim “conquest societies” that are consistent with this chapter’s motivating trends in Figures .–. (and test propositions and A). To test proposition B, I evaluate the effect of rents in nonMuslim societies and demonstrate how their higher levels of preexisting cohesive institutions facilitated their peaceful transitions to democracy when their rents declined. Chapter demonstrates how oil revenues have sustained dictatorship in oilproducing societies that were exposed to Muslim conquest without elevating the propensity for civil conflict. This latter finding can be attributed in part to the presence of geopolitical rents from the United States that have strengthened the repressive capacity of governments in oil-producing Muslim countries to thwart domestic insurgencies. To the best of my knowledge, introducing geopolitical rents is new to the oil curse literature. Endogeneity, of course, looms large in identifying causal relationships. To mitigate these concerns, I follow Kris Ramsay () and use out-of-region disaster damage as an instrumental variable for oil wealth to estimate a causal relationship between oil and political violence, particularly in conquered Muslim territories. This chapter also provide evidence that oil producers with more cohesive institutions (i.e., those not exposed to Muslim conquest) have been able to transition to greater democracy. A finding consistent with proposition . In Chapters and , I probe the effect of rents in non-oil-producing developing countries. In these chapters I focus on rents that emanate from abroad – foreign aid in Muslim societies, “petrodollar” investments in Latin America, and trade subsidies in Eastern Europe – that also co-vary with movements in world oil prices. This second ‘fact’ is attractive in drawing causal inferences. Since oil prices are plausibly exogenous to economic and political conditions in non-oil-producing developing countries, I exploit their variation to predict movements in rents to explain variation in political violence in non-oil
I use data on US troop deployments and qualitative accounts of the US government’s strategic engagement in the Persian Gulf to justify this argument. Limiting the sample to non-oil-producing countries purges the direct effect of oil revenues on politics.
Published online by Cambridge University Press
Analytical Framework
producing countries. I do so by employing difference-in-differences and instrumental variables research designs. In Chapter , I leverage a quasi-natural experiment of oil price driven aid disbursements to demonstrate how higher amounts of foreign aid entrenched autocracy in non-oil-producing Muslim countries (many of whom were conquered by Muslim armies), but their subsequent decline led to civil war. These dynamics are consistent with proposition A (and Figure .). Chapter evaluates proposition B. Using case studies – from Latin America and former Soviet Eastern Europe – and cross-national statistical analysis, the chapter argues that societies with preexisting institutions that are more cohesive transitioned to democracy when their rents declined. The chapter’s empirical analysis looks beyond Muslim societies – thus, providing a broader evaluation of the book’s theoretical arguments to a global sample of countries.
. This chapter presented the book’s analytical framework. It began by describing several empirical patterns that motivate the empirical and theoretical puzzles. These can be distilled to: () What explains the persistent democratic deficit between Muslim and non-Muslim societies? () Across Muslim societies, why do differences in a country’s rent structure (i.e., oil versus non-oil production) and its temporal variation explain movements in dictatorship and civil conflict? Motivated by the second question, the rest of the chapter advanced a theory linking variation in rents and exogenous (preexisting) cohesive institutions to varying intensities of political violence (dictatorship and civil war). Using a formal model, I generated several propositions, three of which may “answer” the two puzzles described above: () Higher levels of rents entrench dictatorship in societies whose preexisting institutions are less cohesive. () A decline in rents elevates the prospect of civil war in societies whose preexisting institutions are less cohesive. () A decline in rents can foster a peaceful transition to democracy in societies whose preexisting institutions are more cohesive. I ended the chapter by previewing the research designs to evaluate these propositions and empirical puzzles. My aim in Part II of the book is to demonstrate how the spread of Islam through military conquest (between
Chapter also introduces a new source of exogenous variation to measure preexisting cohesive institutions in non-Muslim societies.
Published online by Cambridge University Press
A Political Economy Approach
and ) introduced an institutional trajectory that created a ‘sticky’ set of less cohesive institutions in contemporary Muslim societies. Thus, I leverage the historical legacy of Muslim conquest as a plausibly exogenous source of preexisting cohesive institutions to evaluates propositions and corollaries A and B in the contemporary period. Part III of the book does just that by examining how variation in rents explains the persistence of a democratic deficit between Muslim and non-Muslim countries and variation in political violence across oil and non-oil-producing Muslim societies.
Published online by Cambridge University Press
Appendix A
0
POLITY (normalized) 0.2 0.4 0.6
0.8
Additional Figures
1970
1975
1980
1985
1990 Year
Oil FIGURE A
1995
2000
2005
2010
Non-oil
Democracy in non-Muslim countries
Published online by Cambridge University Press
0.35 0.3 0.25 0.2 0.15
0
0.1
0.05
POLITY (normalized)
Civil war 0.1 0.15
0.2
0.25
A Political Economy Approach
1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 Year Civil war
Democracy and civil war in Muslim oil producers
100
0.04 Civil war
0
50
0.02 0 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 Year Civil war FIGURE A
Aid per capita
Civil war and foreign aid in non-Muslim non-oil producers
Published online by Cambridge University Press
Aid per capita
0.06
150
FIGURE A
POLITY (normalized)
0
0
20
0.05
Civil war 0.1 0.15
40 60 80 Oil price (2015 US$)
0.2
100
0.25
Analytical Framework
1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 Year Civil war FIGURE A
Oil price (2015 US$)
Civil war and oil prices in Muslim oil producers
Published online by Cambridge University Press
Appendix B Proofs for the Formal Model
In this appendix, I present the proofs for Section ..
. First, it is important to confirm the ranges of R presented in Table . cover the entire (non-negative) real line without overlap for every valid parameter value of θ, L, and H. That is, I need to show that 2Lð1−θÞ 2Lð1−θÞþL 1−θÞ−L L < 2H ð1−2θ . 1−2θ < 1−2θ < 1−2θ The first inequality holds because 0 < θ < 12 by assumption, so 2ð1−θÞ > 1. The second inequality is true because L > 0. To verify the third inequality, note that 2H ð1−θÞ > 6Lð1−θÞ, because of the assumption that H > 3L. This means 2H ð1−θÞ−L ð1−θÞ 1−θÞ−L 1−θÞþL > 2L1−2θ þ 4Lð1−2θ > 2Lð1−2θ , and implies that Table . covers the 1−2θ non-negative real line without overlap. The assumption that H > 3L guarantees the lower bound for R in Interval of Table . is in fact lower than the upper bound, which ensures that there are values of R which generate twosided violence. I next verify that the stated strategies for the incumbent are in fact optimal for the incumbent. In Section ., I showed that:
ð1−θÞ Peace is the incumbent’s optimal response to peace if 0 R 2L1−2θ .
ð1−θÞ Low violence is the incumbent’s optimal response to peace if R > 2L1−2θ .
On the other hand, if the opposition is violent, then the incumbent chooses between peace, low violence, and high violence. The payoffs are:
V I ðL, 0Þ ¼ θR These proofs are from Ahmed, Schwab, and Werker ().
Published online by Cambridge University Press
Analytical Framework
V I ðL, LÞ ¼ 12 ð1−θÞðR−LÞ þ 12 θðR−LÞ ¼ 12 ðR−LÞ V I ðL, H Þ ¼ ð1−θÞðR−H Þ.
It is straightforward to show that the first value is highest when 1−θÞ−L L L 0 R 1−2θ , the second highest when 1−2θ < R 2Hð1−2θ , and the last value
1−θÞ−L , which confirms the values presented in Table .. is highest when R > 2Hð1−2θ
. Table . shows the optimal choices of both groups as a function of R. The incumbent’s choices in each region follows from Table ., so the only remaining task it so verify the opposition’s choices, conditional on how the incumbent would respond to peace and violence. The opposition’s payoffs for each of the possible outcomes is:
V P ð0, 0Þ ¼ 12 R V P ð0, LÞ ¼ θðR−LÞ V P ð0, H Þ ¼ θðR−H Þ V P ðL, 0Þ ¼ ð1−θÞR−L V P ðL, LÞ ¼ 12 ð1−θÞðR−LÞ þ 12 θðR−LÞ−L ¼ 12 ðR−LÞ−L V P ðL, H Þ ¼ θðR−H Þ−L.
L Consider the first row of Table . where 0 R 1−2θ . The incumbent will be peaceful regardless of whether the opposition attacks, so the opposition 2L , chooses between 12 R andð1−θÞR−L. The former is higher whenever R 1−2θ which includes the entire row of Table .. Next consider the range where the incumbent is peaceful in response to peace but uses low violence in response to violence. The opposition chooses between V P ð0, 0Þ and V P ðL, LÞ. The former is always higher, meaning that in the second row of Table ., the opposition will be peaceful. Next, consider the range of R where the incumbent will use low violence regardless of the opposition’s choice, which spans the third and fourth rows of Table .. A peaceful opposition group will receive V P ð0, LÞ, while an aggressive opposition group will receive V P ðL, LÞ. The former is higher when 1−θÞþL R 2Lð1−2θ , which confirms that in row of Table . the opposition will choose peace and in row it will fight. Finally, consider the region where the incumbent will use low violence in response to peace and high violence in response to violence. The opposition is guaranteed to lose power, so it prefers to remain peaceful, which confirms the final row of Table .
Published online by Cambridge University Press
Published online by Cambridge University Press
Analytical Framework
In Chapter , I identified two tragic characteristics of many contemporary Muslim societies: the pervasiveness of dictatorship and an elevated propensity of civil war. Among political economists, the prevalence of these outcomes – dictatorship and civil conflict – captures the central features of societies afflicted with political violence. Accordingly, in this chapter, I adopt this terminology and develop an analytical framework to better understand why many Muslim societies are inflicted with varying and heightened levels of political violence. I begin with several figures to motivate the analysis at the heart of this book. These figures demonstrate the cross-sectional and temporal variation in political violence between Muslim and non-Muslim societies and across Muslim societies with different rent structures (i.e., oil versus non-oil producers). These figures suggest a relationship between foreign-based rents – notably, oil revenues and foreign aid – and patterns of dictatorship and civil war within Muslim societies. With this empirical motivation, I then advance a theory linking variation in preexisting cohesive institutions and rents to explain patterns of dictatorship, democracy, and civil war and transitions between these outcomes. This discussion proceeds in a stepwise fashion.
For example, in a series of theoretical formulations, Besley and Persson (, , a,b) model a hierarchy of political violence that ranges from peace, repression (one-sided violence), and civil war (two-sided violence). Relatedly, Svolik (, ) describes how dictators govern in societies in which “violence is an ever-present and ultimate arbiter of conflicts in authoritarian politics.” Rents refer to nontax government revenue. This includes revenues from the extraction and sale of natural resources (e.g., oil, natural gas) that can accrue to national governments and the receipt of foreign economic and military aid. Rents can also include sovereign borrowing and subnational transfers (Morrison ). In some instances, rents can be non-economic, for example geopolitical rents in the form of security guarantees (e.g., foreign troop deployments) that can augment a
Published online by Cambridge University Press
Analytical Framework
I first distill the rich scholarship on rents and political stability, identifying an important tension in existing theoretical formulations: whether a reduction in rents contributes to violence or not. This discussion in turn provides the rationale for advancing a formal theory (model) to help reconcile this tension. Whereas existing models predict a monotonic effect linking greater rents to political violence, my model allows for non-monotonic effects. Consistent with these existing theories, I show that higher levels of rents can sustain autocracy, but unlike these models, I show how declines in rents can generate divergent political trajectories. Specifically, I articulate the conditions under which a decline in rents can generate a peaceful transition to democracy or engender civil war. The model highlights the important role of preexisting cohesive institutions in shaping how a decline in rents can lead to these divergent outcomes: democracy or civil war. To the best of my knowledge, the model advanced in this chapter is the first to generate these differential trajectories in a unified analytical framework. The chapter’s unified framework (model) is an important contribution to existing scholarship. As this chapter describes, existing scholarship has tended to study whether rents and institutions affect patterns of democracy/dictatorship or civil war as distinct and separate outcomes (e.g., Hodler , Menaldo , Smith ). This is surprising since regime type and civil war tend to covary: dictatorships are more prone to civil war (e.g., Blattman and Miguel ). I end the chapter by delineating the model’s empirical implications and preview the research designs to evaluate their capacity to explain patterns of dictatorship and civil war in Muslim and non-Muslim societies (which is the focus of Parts II and III of the book).
. : In this section, I present several figures describing patterns of political violence in Muslim and non-Muslim societies that motivate the core questions (“puzzles”) that this book seeks to explain. The first figure plots the average
government’s capacity to maintain political control. For example, the stationing of US troops in various countries in the Persian Gulf (e.g., Saudi Arabia, Kuwait) can serve to safeguard stable world oil supplies and also serve to deter domestic insurgencies (that could disrupt these oil supplies). While I refer to this model with “my” and “I” in this chapter, it was developed in collaborative work with Daniel Schwab and Eric Werker. See Ahmed, Schwab, and Werker (). Societies with more cohesive institutions are those that share more state resources with groups out of power (e.g., supporters of opposition political parties, different ethnic or religious groups). I discuss cohesive institutions in Section ..
Published online by Cambridge University Press
A Political Economy Approach
level of democracy in Muslim- and non-Muslim-majority societies since . To measure a country’s level (quality) of democracy, I use its composite POLITY score (Marshall and Jaggers ). This index and its sub-indices are used widely in empirical research on political regimes, institutional design, variation, and durability. The POLITY score combines information about the competitiveness of and constraints on executive recruitment with information about the extent of political competition. The measure ranges from − (full dictatorship) to + (consolidated democracy). I normalize each country’s annual score on a [,] scale where a higher value corresponds to greater democracy. Countries where percent or more of their population identify with Islam are classified as Muslim.
A Democratic Deficit in Muslim Societies Figure . reports the average normalized POLITY score across Muslim and non-Muslim countries on an annual basis since . The figure shows that Muslim societies have, on average, been less democratic than non-Muslim societies. For example, using conventional classification schemes in economics and political science, the typical Muslim country has always been a dictatorship: its normalized POLITY has not exceeded .. In contrast, non-Muslim societies have exhibited greater temporal variation in their normalized POLITY score. While the typical non-Muslim country started out as a democracy in , it became less democratic over the subsequent decade. In the mid-s, non-Muslim societies gradually became more democratic on average, a pattern consistent with the so-called Third Wave of democratization (Huntington ). This upward trend accelerated after the end of the Cold War in and has stabilized since . Despite some improvements in democratic governance in Muslim societies after , Figure . notably reveals a persistent “democratic deficit” in Muslim countries (relative to their non-Muslim counterparts).
See for example, Acemoglu and Robinson (), Ahmed (), Besley and Persson (a), Dunning (), and Morrison ().Bueno de Mesquita et al. () use sub-indices of the composite POLITY score to construct their measure of selectorate and coalition size. Alternate measures of democracy (e.g., Boix, Miller, and Rosato ) are contrasted and evaluated in relation to POLITY. Note that my use of “typical” refers to the statistical average across each group of countries (Muslim, non-Muslim). Scholars often use cutoffs in the POLITY index to differentiate dictatorships from democracies. Economists frequently classify countries with a normalized POLITY score greater than or equal to . as a democracy and those below . as a dictatorship. In contrast, political scientists employ a higher cutoff for democracy: a normalized POLITY score equal to or exceeding . (which corresponds to a POLITY score equal to or above +).
Published online by Cambridge University Press
0
.2
POLITY (normalized) .4 .6
.8
1
Analytical Framework
1960
1965
1970
1975
1980
1985 Year
Non-Muslim FIGURE
1990
1995
2000
2005
2010
Muslim
. Persistent democratic deficit in Muslim societies
B Cohesiveness in Muslim Societies The democratic deficit identified in Figure . is associated with less cohesive institutions in Muslim societies. Figure . provides some suggestive evidence using welfare spending as a measure of cohesive institutions (like Figure .). The figure shows that governments in Muslim societies, on average, allocate a persistently lower percentage of their expenditure on welfare spending relative to non-Muslim societies. Over the entire period, the difference in shares (of about percentage points) is statistically different from zero (with a p-value < .). This implies that Muslim societies tend to be significantly less cohesive relative to non-Muslim societies. It is worth emphasizing that reliable crossnational data on welfare spending is available from and is therefore less extensive than measures of democracy, which are available from (and for many countries for decades prior to ). As democracy and cohesiveness tend to be positively associated (see Figure .), my analysis going forward will
Recall that cohesiveness refers to the proportion of state revenue (expenditure) shared between supporters of the ruling government and those in opposition. In societies with less cohesive institutions, the ruler is less egalitarian in sharing state revenue with supporters and opposition groups. That is, the ruler tends to spend a lower share of expenditure on broad-based subsidies and transfers to the masses.
Published online by Cambridge University Press
0
Subsidies and transfers (% Expenditures) 10 20 30 40
A Political Economy Approach
1990
1995
2000 Year Non-Muslim
FIGURE
2005
2010
Muslim
. Cohesiveness in Muslim and non-Muslim societies
use democracy as a proxy of cohesiveness, primarily since there is more extensive cross-national and temporal data coverage for this measure.
C Political (In)Stability in Muslim Societies The relative stability of dictatorship in Muslim societies suggested in Figure ., however, seems to mask differences in their underlying rent structure. To see this, Figure . plots the average normalized POLITY score across oil and nonoil-producing Muslim societies (I use British Petroleum’s designation of oilproducing countries). This figure reveals several interesting patterns regarding the relative stability of dictatorship across oil and non-oil-producing Muslim societies. First, while oil and non-oil-producing Muslim societies exhibited, on average, similar levels of democracy in , oil producers became increasingly less democratic in the s. Second, despite some improvements in democracy in the s, non-oil-producing Muslim societies reversed this trend in the s. And by the early s, Muslim non-oil producers were less democratic than their oil-producing countries. Third, Muslim non-oil producers experienced another reversal: in the mid-s they started to become slightly more
As I discuss in Chapter this trend may be due to the nationalization of oil companies in the s and the subsequent surge in oil income to national governments (Mahdavi ).
Published online by Cambridge University Press
0
POLITY (normalized) .2 .4 .6
.8
Analytical Framework
1960
1965
1970
1975
1980
Oil FIGURE
1985 Year
1990
1995
2000
2005
2010
Non-oil
. Democracy in Muslim oil and non-oil producers
democratic and by the early s passed the threshold (barely) for democracy. Fourth, Muslim oil producers did not experience these improvements in democracy. By , these societies tended to be less democratic than they were in . The greater variation in democracy among Muslim non-oil producers prompts greater scrutiny. Figures . and . provide some suggestive explanations. Figure . plots the average incidence of civil war in Muslim non-oil producers (left axis) against their average level of democracy (right scale). It shows that Muslim non-oil producers were relatively peaceful in the s and s (when they also became less democratic), but became increasingly conflict-prone in the mid-s until the late s. During this period, these
Similar patterns are not observed in non-Muslim countries. Since the s, both oil and nonoil-producing countries, on average, have become more democratic over time. Many of these oil producers have been consolidated democracies since the s (e.g., Norway, United States) and only recently discovered significant quantities of oil (e.g., the United Kingdom in the North Sea in the s). See Figure A in this chapter’s appendix. Data on civil war is from the Armed Conflict Database (Harbom, Melander, and Wallensteen ). Following convention, conflicts exceeding , battle deaths per annum comprise civil war. In contrast, battle deaths below the , threshold comprise civil conflict. Similar patterns are not observed in Muslim oil producers (see Figure A in the chapter’s Appendix). These countries have remained relatively stable politically compared to Muslim non-oil producers. In Muslim oil producers, the average incidence of civil war has hovered
Published online by Cambridge University Press
0
.2
.05
Civil war .1
.15
.3 .4 .5 POLITY (normalized)
.2
.6
A Political Economy Approach
1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 Year Civil war FIGURE
POLITY (normalized)
. Civil war and democracy in Muslim non-oil producers
societies were twice as likely to experience civil war relative to the prior two decades. Concomitantly, these societies started to become more democratic as the average normalized POLITY score started to rise. With these improvements in democracy, by the early s civil strife in these countries subsided significantly. A potential source for the patterns in Figure . may be attributable to variation in a specific type of unearned (nontax) government income received in many non-oil-producing Muslim societies: foreign aid. As Neumayer () and Werker, Ahmed, and Cohen () document, these countries have been the recipients of significant amounts of foreign aid from Western and Arab donors. Figure . shows that movements in foreign aid per capita are correlated with political stability in Muslim non-oil producers. The figure suggests
between . and . since and has been driven primarily by sustained violence in Algeria and Sudan. Despite some fluctuation – both increases and decreases – over time, the average level of democracy in Muslim oil producers has not changed since . These patterns are not present in non-Muslim non-oil producers (see Figure A in the chapter’s appendix). First, the average incidence of civil war in non-Muslim countries has been significantly lower (e.g., . at its maximum in ). Second, a decline in aid is not associated with an uptick in civil war. As I describe and formalize later in this chapter, this null association is theoretically plausible because non-Muslim countries could have (relatively) more cohesive institutions. The analysis in Chapter provides empirical support.
Published online by Cambridge University Press
0
0
.05
50
Civil war .1
100 Aid per capita
.15
150
.2
200
Analytical Framework
1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 Year Civil war
Aid per capita
FIGURE . Foreign aid and civil war in Muslim non-oil producers Note: Aid per capita reports the annual average value among Muslim non-oil producers in in US dollars. These figures are based on the author’s calculations using data from the World Bank.
buoyant amounts of foreign aid may “buy” political stability, while their decline may engender conflict. Moreover, notice how a significant decline in foreign aid in the mid-s coincides with the uptick in civil war in Muslim non-oil producers. Figures .–. generate several questions at the heart of this book. For instance, why is there a persistent democratic deficit in Muslim societies relative to non-Muslim societies? What explains its temporal variation, particularly its widening since the s? For example, why did many Muslim societies miss the Third Wave of democratization? Looking across Muslim societies, why have oil producers remained more politically stable compared to non-oil-producing Muslim societies? Why might a country’s rent structure matter? Moreover, why have non-oil-producing Muslim countries experienced greater
In contrast, a reduction in oil prices (and associated oil rents) seems to be unassociated with, and notably has not triggered, heightened violence in Muslim oil producers (see Figure A in the chapter’s appendix). As I argue in Chapter , this may stem from the provision of geopolitical rents (e.g., implied US military support for the government) in Muslim oil producers that has effectively deterred opposition groups from engaging in combat against the government.
Published online by Cambridge University Press
A Political Economy Approach
political upheaval in the form of civil war but remained relatively nondemocratic? To answer these questions we need to explain () the persistence of nondemocratic political structures in Muslim societies relative to non-Muslim societies and () the possible role of rents in explaining variations in political violence – dictatorship and civil war – across oil and non-oil-producing Muslim societies. In Part II of this book, I provide an answer to the first question by demonstrating how the institutional legacy of Muslim conquest – a process that occurred over a millennia ago – helps explain the persistence of autocracy in many contemporary Muslim societies relative to non-Muslim societies (i.e., the democratic deficit in the cross section of countries). In the remainder of this chapter, I focus on the second question by explaining how rents can affect variations in political violence both across countries and over time, and how this may be related to the quality of preexisting cohesive institutions.
. There is an expansive literature on the causes and consequences of dictatorship and civil war. Rather than summarizing this rich scholarship, my aim in this section is to distill the analytical features that are relevant to understanding how rents can explain variation in political violence. This discussion highlights some of the prominent theoretical and supporting evidence and showcases some of the mixed findings. In doing so, I identify several tensions in this existing scholarship, which serves as a motivation to formulate my own theoretical formulation (model) to reconcile these tensions. Before delving into this, I first define key terms. A Political Violence Both as a cause and consequence, dictatorship is an important feature of political violence. At its core, dictatorship comprises rule by the few over the many, with many scholars conceptualizing it as a “residual” to
As such, Part II provides an account for the cross-sectional variation in contemporary levels of democracy between Muslim and non-Muslim societies. See Blattman and Miguel (), Haggard and Kaufman (), Kalyvas (), and Svolik () for excellent reviews of this scholarship. Throughout this book, I will use the terms dictatorship, nondemocracy, autocracy/ authoritarianism interchangeably. I am not unique in investigating some of these tensions. In evaluating the relationship between oil and democracy, Benjamin Smith and David Waldner () identify the “scholarly dissensus,” highlighting conceptual, theoretical, and empirical challenges in existing scholarship. While civil wars occur most frequently in dictatorships, this is not universal. Democracies too may experience civil conflict, such as the American Civil War. Thus, I argue that dictatorship is a strong “correlate” (rather than source) of civil war.
Published online by Cambridge University Press
Analytical Framework
democracy. In this tradition, dictatorships and democracies are contrasted on their procedural approach to the selection of leaders (Dahl , Schumpeter ), often by clearly articulating the minimal electoral requirements for democracy (Alvarez et al. , Boix ). Empirically, this generates a dichotomy between dictatorships and democracies (see e.g., Boix, Miller, and Rosato ). In contrast, other formulations identify and seek to categorize the broader institutional variations that feature across dictatorships, such as differences in the constraints faced by the leader, the competitiveness of leader recruitment/selection, and more generally who can participate in the political process. These variations have been measured on a continuum or with discrete categories. The POLITYindex (Marshall and Jaggers ) is an example of the former, while categorizations, such as “competitive” (Levitsky and Way , ), “electoral” (Schedler ), or “semi-authoritarianism” (Ottaway ) characterize the latter. Turning to their operation, as Milan Svolik () cogently distills, all dictatorships must grapple with two central problems: those of authoritarian control and power-sharing. Each problem relates to a distinct group in a polity: those who rule (“the elite”) and those who are ruled (“the masses”). The challenge of balancing against the threats from the masses (majority) excluded from power constitutes the problem of authoritarian control. In practice, of course, dictators rarely control enough resources by themselves to thwart these challenges and therefore rule with a (small) number of allies, such as military officers, prominent party members, or clerics. This can generate a second, distinct political problem when dictators confront challenges from those with whom they share power. This is the problem of authoritarian power-sharing. A dictator who fails to address these problems faces a heightened risk of losing power through elite defection (coups) and violence from opposition groups (civil wars) and less frequently via social revolution. Moreover, these dual challenges are magnified by the dismal setting of authoritarian politics. First, a dictator lacks an independent authority that can enforce agreements with his governing allies and repressive agents, thus
Gandhi (, chapter ) provides a superb discussion of the historical and contemporary use of dictatorship. Many prominent scholars conceptualize dictatorship as a residual category of democracy. For example, Boix (), Cheibub, Gandhi, and Vreeland (), Przeworski et al. (), and Svolik () define dictatorship as an independent country that fails to satisfy at least one of the following two criteria for democracy: () free and competitive legislative elections and () an executive who is elected either directly in free and competitive presidential elections or indirectly by a legislature in parliamentary systems. A related formulation characterizes who rules in dictatorships, such as the military, a single party, or a personalistic ruler. See, for example, Geddes () and Geddes, Wright, and Frantz (). For example, over the period –, only percent of dictators exited office due to popular uprisings (Svolik , ).
Published online by Cambridge University Press
A Political Economy Approach
elevating the risk of elite defection. Second, violence is ever-present and ultimate arbiter of conflicts. Thus, Svolik (, ) proclaims the “key features of authoritarianism – including institutions, policies, as well as the survival of leaders and regimes – are shaped by the twin problems of power-sharing and control against the backdrop of the dismal conditions under which authoritarian politics takes place.” The dismal setting that dictators face raise their societies prospects for experiencing civil war; a finding, well documented in the empirical civil war literature. Conceptually, civil war entails two-sided violence that pits the incumbent (government) in armed combat against one (or more) opposition groups. In explaining their determinants, many scholars differentiate between “grievances” and “opportunities” as important sources of civil war (e.g., Blattman and Miguel , Collier and Hoeffler , Kalyvas ). On the former, since dictatorships frequently exclude certain groups from power, grievances can be an important driver for conflict. For example, regimes that restrict political participation frequently pursue redistributive policies that perpetuate inequalities and perceptions of deprivation that can mobilize rebellion (Besley and Persson , Cederman, Weidmann, and Gleditsch , Gurr ). Of course, the mere existence of grievances does not imply civil war. As Stathis Kalyvas (, ) observes: “Obviously, the problems with grievances as a determinant of civil war onset is that they seem to be much more prevalent than civil war and that they are very hard to measure directly. Ultimately, it is difficult to escape the conjecture that one has to look for combinations of demand for, and supply of, rebellion (or intention and opportunity).” Greed-based explanations posit that economic conditions comprise an important driver of civil war (Collier and Hoeffler ). One channel posits that economic decline, for example through depressed wages, can both lower the opportunity cost for rebels to mount an insurgency and weaken an incumbent’s capacity to fight back (Miguel, Satyanath, and Sergenti , Powell ). Another channel identifies the motive for the incumbent and opposition to control state resources (“the state prize”), such as foreign aid and a variety of natural resources (Besley and Persson a, b, Le Billon ). For instance, revenues derived from the state’s control of natural resources, such as oil, natural gas, diamonds, and timber, can incite violence from all sides. Revenues from natural resources can finance an incumbent’s repressive
The literature on civil war identifies different types of groups based on their: ethnic, linguistic, and religious identification/fragmentation, geographic location (e.g., urban versus rural, near natural resources), social/economic class and modes of production (i.e., levels of vertical and horizontal inequality in societies), and ties to foreign powers. Differences in the composition, size, and/or intensity of grievances among individuals in these groups can motivate armed violence.
Published online by Cambridge University Press
Analytical Framework
capacity, such as the size of its army. Similarly, opposition groups may seek to control the natural resources to finance their rebellion. Empirically, several studies document a higher likelihood, intensity, and duration of civil war in countries with natural resources (e.g., Le Billon , Lei and Michaels , Synder ). Greed-based explanations, of course, do not exclude the importance of political grievances. In practice, the presence of both motives – grievances and greed – heighten the incidence of civil war. Consequently, dictatorships with valuable state prizes (e.g., oil revenues, foreign aid receipts) often face an elevated risk of civil war. The preceding overview of political violence highlight two important points. First, dictatorships entail the rule of the few over the many, in which the regime confronts two problems – of control and power-sharing – in a dismal setting of ever-present violence. Second, dictatorships especially those with valuable state prizes, also face an increased risk of two-sided violence in the form of civil war. B Rents and Political Regimes Rents and Governance Given the political challenges described above, how can incumbents better ensure the longevity of their power? Scholars across multiple disciplines have identified the essential role of buoyant public finances in helping governments (rulers) remain in power. In describing the primacy of revenues for all governments, Margaret Levi (, ) describes how they can benefit a ruler personally and/or his subjects: Rulers may use the funds to line their own pockets or to promote their personal power. They may use the funds to support social or personal ends. They have ideological ends they wish to promote. They may be altruistic. Randomness of ends characterizes rational choices in general, and rulers operate with a wide range of them. It’s by the means of the state and its revenues that rulers achieve their personal and social ends.
As Levi subsequently elucidates, the importance of revenues thus incentivizes rulers to be “predatory” and extract as much revenue as possible. In doing so, however, rulers face constraints. For example, maximizing state revenue involves the cost of extraction or a trade-off between extraction costs (e.g., deadweight loss associated with taxation) and economic development (e.g., growth). Minimizing the transaction costs associated with taxation, therefore, can compel rulers to derive their revenue from alternate sources. Scholars frequently refer to these alternate sources of nontax government revenue as unearned income or rents (e.g., Ahmed , , Morrison , Smith ). Whereas tax revenue entails the extraction of income from
The incumbent could also transfer funds to regime supporters and/or “buy off” rebel groups. Since this sub-section builds on Levi’s () seminal work, I adopt her use of the term “rulers” to refer to governments/regimes.
Published online by Cambridge University Press
A Political Economy Approach
economic agents (e.g., wage earners, firms) and associated transactions (e.g., value-added taxes, trade tariffs), rents may entail revenues derived through the state’s exploitation of natural resources (e.g., oil, minerals) and external transfers (e.g., economic and military assistance from foreign governments). Rents can include revenues from oil because it can be generated through the operation of state-owned oil enterprises rather than through taxes. Rents also include income from abroad (e.g., sovereign borrowing, foreign aid) and intergovernmental grants at the subnational level. For many governments, rents are nontrivial and account for a large share of their expenditures, such as in Nicaragua ( percent), Bolivia ( percent), Republic of Congo ( percent), Burundi ( percent), Romania ( percent), Syria ( percent), Iran ( percent), Nepal ( percent), Pakistan ( percent), and Japan ( percent). Rents and Democracy The generation and composition of government finances from tax and nontax sources has political implications. For instance, in premodern Europe, the need to finance interstate wars compelled European sovereigns to develop tax collection systems. According to Charles Tilly (), taxation required bargaining between revenue-maximizing leaders and their citizens through which rulers relinquished some influence over state policies in exchange for taxes. With more warfare over time and thus a greater reliance on tax revenue, rulers tended to become more accountable to their populations. This bargaining mechanism thus generates a pathway from taxation to more democratic governance; and by extension, governments that generate their revenue from nontax sources are less likely to feature democratic characteristics. This framework linking taxation to democracy has been applied to explain modern political development in various regions, including Africa (Guyer ), Asia (Bernstein and Lu ), the Middle East (Crystal ), Europe and North America (Scheve and Stasavage ), and the former Soviet Union (Gehlbach ).
To be clear, tax revenue entails the extraction of income from individuals (e.g., wages, salaries), firms, land and capital owners, and associated economic transactions via consumption (e.g., value-added taxes) and international trade (e.g., tariffs), among others. Note, my definition of rents excludes remittances (which accrue directly to individuals). However, through an expenditure switching strategy, governments may be able to incorporate remittance income into their revenue base, thus comprising a type of nontax income (Ahmed , ). State-owned enterprises control percent of the world’s oil production and percent of its reserves (Ivanhoe , Tordo et al. ). Examples include Saudi Aramco, China National Petroleum Corporation, Nigerian National Petroleum Corporation, and Petroleos de Venezuela. Based on calculations from Morrison (, ) using data averaged over the period, –. Bates and Lien () formalize this bargaining process and identify two important considerations. First, greater asset mobility may weaken the sovereign’s ability to tax. If the tax rate is too high, citizens may shift their assets abroad (an argument further explored by Boix ). Second, the relationship between taxation and representation may be conditional on the amount of spending that citizens (expect to) receive in return.
Published online by Cambridge University Press
Analytical Framework
While taxation tends to be positively associated with democracy, it frequently generates political contestation within societies. Indeed, redistributive conflicts – typically regarding the level of taxation and associated transfers from the “elites” to the “masses” – are critical elements in prominent theories of political instability and regime transitions (e.g., Acemoglu and Robinson , Boix , Moore ). According to Kevin Morrison (), two tensions underlie this redistributive conflict: the need for governments to spend while minimizing their tax burden. The availability of nontax revenue (rents) can assuage these tensions: “Nontax revenues enable governments to reduce their taxation levels while increasing government spending, thereby increasing the stability of democracies, dictatorships, and their leaders” (Morrison , ). To understand the logic, consider how rents can strengthen democracy in Daron Acemoglu and James Robinson’s () theory of democratic transitions. They posit that elites democratize as means to credibly commit to future redistribution rather than suffer revolution in which the masses expropriate the assets held by the elites. In this context, the availability of rents can help lower a regime’s tax burden, thus relieving the costs of redistribution for elites and facilitating democracy. Thad Dunning () applies this logic in articulating how oil rents can foster “crude democracy”: By reducing the extent to which democratic majorities redistribute private income or wealth away from the rich elites through taxation, land reform, or other measures, resource rents can reduce the economic cost of democracy to elites. . . . This occurs because both resource rents may reduce the current redistributive costs of democracy and because resources may endogenously lower the extent to which future democratic regimes may redistribute private wealth from elites when there is not a coup threat; resource rents thus shape elite expectations about redistribution under future democratic regimes.
Rents and dictatorship. While rents may lower the costs of democratization in theory, there is a large body of research documenting how rents tend to entrench dictatorship. One strand of this literature identifies the role rents can play in introducing authoritarianism in existing democracies. For instance, rents can engender a strong “incumbent advantage” that can reduce political competition and introduce creeping authoritarianism (Wantchekon ). In clientelistic settings, one pathway is via targeted redistribution: The incumbent can transfer rents directly to their supporters (e.g., co-ethnics) and exclude other groups from the direct benefits of rents (Jablonski ). Consequently,
This logic also applies to Boix’s () theory of democratization. Relatedly, Menaldo () argues whether natural resources enhance democracy is conditional on preexisting institutional quality. Rents in countries with good/democratic institutions do not harm democratic governance (e.g., Norway). In support of this conjecture, Dunning shows that natural resource rents, particularly from oil, supported “crude democracy” in several Latin American countries, most notably Venezuela.
Published online by Cambridge University Press
A Political Economy Approach
the recipients of rents may be more likely to turn out and vote for the incumbent. Another strand of this literature discusses the channels through which rents strengthen dictatorship. Much of this scholarship originates from scholars of the Middle East investigating whether the prevalence of rents fostered authoritarian governance in that region (Beblawi , Chaudhry , Madhavy ). The central argument in this “rentier state” literature posits that governments funded by external rents are relieved from the need to raise taxes, which makes them less accountable to their citizens. As a result, these governments are less likely to deploy the rents in ways that promote economic development. Focusing on the political ramifications, scholars have refined the theoretical channels through which rents can hinder democracy. Michael Ross (, –) identifies and tests three broad “effects” through which rents can foster dictatorship: a “rentier effect,” through which rents “relieve the social pressures that might lead to demands for greater accountability” from the broader population; a “repression effect,” through which rents may empower “governments to spend more on internal security and so block the population’s democratic aspirations”; and a diminution of a “modernization effect,” through which “resource-led growth does not lead to higher education levels and greater occupational specialization. . . [and thus] should fail to bring about democracy.” As Ross concludes, these effects are not mutually exclusive as “all three explanations, or any combination of them, may be simultaneously valid.” In delineating these effects, Ross elucidates how rents can help dictators mitigate their dual challenges of authoritarian control and power-sharing, for instance, through his decomposition of the rentier effect into three distinct mechanisms: a taxation, spending, and group formation effect. Consistent with Tilly () and my earlier discussion, the taxation effect argues that rents reduce a government’s need to tax their population for revenues and consequently leads their populations to demand less political accountability. In
Mahdavy (, ) first advanced the contemporary concept of the rentier state as a state that receives substantial rents from foreign individuals, concerns, or governments. Beblawi (, ) developed a more precise conceptualization of a rentier state. In such a state, () the rents are paid by foreign actors, () these rents accrue directly to the state, () only a few domestic actors are engaged in the generation of this rent (wealth), and () the majority are only involved in the distribution or utilization of the wealth. Chaudhry’s () seminal work broadens the theoretical and empirical scope, arguing how oil wealth affects governance both in oil producers and labor-exporting countries in the Middle East. For the latter, Chaudhry argues migration to oil producers from nonproducers and associated remittance inflows can affect governance in laborexporting countries (Ahmed provides causal evidence in support of this argument). For more recent treatments on the relationship between rents – mainly oil – and democracy in the Middle East, see Cammett et al. () and Smith and Waldner (, chapter ). While Ross discusses these effects with regards to oil wealth, the insights can be generalized to other rents.
Published online by Cambridge University Press
Analytical Framework
combination this can enhance the dictator’s control over society. The spending effect posits that rents allow governments to spend more on patronage, which in turn dampens latent pressures for democratization. These transfers may reward a small and critical group of regime supporters (e.g., military commanders, clerics, co-ethnic leaders) and/or fund social services that benefit the broader population. The former, which supplies targeted private benefits, can help dictators reduce the problem of authoritarian power-sharing by retaining the loyalty of key regime supporters. In contrast, if a government uses its rents to fund generous social benefits, it may be an effective instrument of authoritarian control. For example, Entelis () describes how the Saudi Arabian government uses its oil wealth to fund a generous welfare state that reduces pressures for democratization. Relatedly, the group formation effect argues that when rents are buoyant, the government can use its largesse to prevent the formation of social groups that are independent from the state and thus, be inclined to demand greater political representation. Heterodox Views Some scholars point to the mixed evidence between rents – particularly, oil wealth – and democracy/dictatorship to suggest that rents may have no independent (causal) effect on democracy or dictatorship. For example, examining data over two centuries, Stephen Haber and Victor Menaldo () find a tenuous relationship between oil wealth and dictatorship and argue that there is no “oil curse,” at least over the long run; and in some instances, oil rents may comprise a political “blessing.” In a similar vein, Benjamin Smith and David Waldner (, ) highlight a “scholarly dissensus” in existing scholarship linking oil to dictatorship. They maintain:
Moreover, the utility of this strategy is magnified in increasingly smaller power-sharing coalitions. This is because the per capita transfers can be higher to coalition members. Smith (, ) summarizes this logic: When coalition size is small, leaders can greatly reward their essential supporters by supplying private goods [targeted benefits]. However, as coalition size becomes large, private goods become more thinly spread, and leaders can more efficiently reward their supporters through the provision of public goods.” Bueno de Mesquita et al. () and Lake and Baum () provide empirical evidence of this trade-off. Ahmed (, chapter ) provides a numerical example. As discussed shortly, providing social benefits for the masses (e.g., public goods) may not be optimal for dictators. A more politically beneficial strategy for a dictator is to fund repression and targeted transfers (e.g., Smith ). A variant of the group formation effect builds on Putnam’s () work on social capital. Here rents can stifle the formation of civic groups that lie above the family but below the state. These civic groups are often important drivers for greater democratic accountability. Andersen and Ross () rebut this conclusion, finding that oil wealth may be a blessing in (some instances) prior to . However, after there is stronger evidence pointing to an oil curse. Smith and Waldner () reach this conclusion by pointing to several conceptual (measurement), empirical, and theoretical issues in extant scholarship.
Published online by Cambridge University Press
A Political Economy Approach
That when one looks closely at specific claims made in the literature, and the specific methodological warrants of those claims, the conclusion that a political resource curse exists appears less unassailable than would be the case when all the diverse findings are pooled together without sufficiently discriminating appraisal.
Smith and Waldner’s synopsis engages with the empirical evidence linking oil to politics across time and space (regions). Their analysis reveals that a political resource curse exists primarily in the Middle East and North Africa, which they attribute to factors related to institutions and coalitional politics. To the extent that the institutional legacy of Muslim conquest may explain the configuration of institutions and coalitions (e.g., relationship between the ruler, the military, and religious clergy) in many contemporary Muslim societies (as explored in Part II of this book), Smith and Waldner’s inferences may be compatible with those advanced in this book. C Conditional Theories of Rents and Political (In)stability While rents can help dictators mitigate their dual challenges of authoritarian control and power-sharing, the specific policies they permit (e.g., kleptocracy, repression) may incite political instability, such as revolutions and civil war. As noted earlier, dictatorship and its accompanying politics, often creates a climate conducive for violence. With this environment in mind, existing scholarship identifies several pathways linking rents and regime type to political instability. Variation in Institutions One pathway evaluates the effects of rents on political stability as institutions vary. In the “perils of unearned income,” Alastair Smith () formalizes these dynamics within the selectorate model of politics (Bueno de Mesquita et al. ). A critical parameter in the selectorate model is the size of the “winning coalition” (W) required for a leader to remain in power. Small W systems characterize dictatorships, where the leader’s optimal strategy is to
For the latter, they study the relationship between oil and dictatorship (and the prospects of democratic transitions) in four regions – Latin America, the Middle East, Africa, and Southeast Asia. Here I focus on civil war, but rents may also incite war between states. See for example, Caselli, Morelli, and Rohner (), and Colgan (). A related explanation focuses on the strength of opposition forces and the timing of oil wealth. Focusing on oil producers, Benjamin Smith () posits that variation in the strength of opposition forces (to the government) in late industrializers can affect how rulers manage their oil wealth with divergent political implications when oil wealth declines. Where opposition forces are strong in late industrializers (e.g., Indonesia), the ruler undertakes policies (e.g., more effective taxation of non-oil income) that reduce its probability of losing power when oil income declines. In contrast, where opposition forces are weak in late industrializers (e.g., Iran), the ruler is prone to adopting policies (e.g., patronage to selected/narrow groups) that increase its probability of losing power when oil income declines.
Published online by Cambridge University Press
Analytical Framework
provide targeted benefits to coalition members and to reduce the provision of public goods (e.g., welfare transfers, rule of law). One of Smith’s key findings identifies a non-monotonic relationship, where the effect of rents on revolutionary threats varies with the size of W. Referring to rents as “free resources,” Smith’s () comparative statics reveal that “large coalition systems are relatively more efficient than small coalitions at converting free resources into societal benefits and economic activity because free resources fund public goods rather than corruption, graft, and kleptocracy. From the perspective of citizens outside of the winning coalition, if the level of free resources rises, then the desirability of a large coalition increases relative to a small coalition system . . . This encourages the citizens to support revolution” (). Of course, if the costs of rebellion are prohibitively high, the availability of rents may not necessarily induce revolution. As Smith (, ) observes, these costs can be large in small W systems: The citizens’ decision to rebel is affected by two factors: the difference between their welfare under the current institutions relative to the welfare they expect under postrevolutionary institutions and the probability of the revolution succeeding. When coalition size is small, leaders provide few public goods. This leaves the citizens extremely dissatisfied with their lot relative to what they could obtain under postrevolutionary institutions. This gives them a great desire to rebel. However, the citizens also factor the probability of success into their decision to rebel. When the coalition is small, leaders provide few of the public goods which help revolutionaries coordinate and organize. As a result, when W is small, citizens choose not to rebel; not because they don’t want change, but they are unlikely to succeed.
The confluence of these comparative statics generates an inverted U-shaped relationship between rents, coalition size, and revolutionary threats. In small W and large W systems, rents are less likely to induce revolution. However, in the intermediate range of coalition size, greater rents incite revolution. Of course, as Smith () also shows, the availability of free resources can empower governments to finance their repressive capacity to crush rebellion.43 In conjunction with the U-shaped relationship, this policy response implies that relatively robust dictatorships (e.g., monarchies or personalist dictatorships with smaller W) can effectively deter revolutionary threats induced by rents. In short, rents can monotonically increase political stability in dictatorships. In subsequent research, Bruce Bueno de Mesquita and Alastair Smith () provide empirical corroboration for this prediction.
Figure in Smith () demonstrates this. In Smith’s model, governments can also contract their supply of public goods which can make it more difficult for individuals and groups to collectively organize. Note, the non-monotonic effect formalized by Smith () and empirically supported by Bueno de Mesquita and Smith () varies by coalition size. It does not vary by changes in the level of rents. This distinction is important as the model I introduce in Section . generates nonmonotonic effects that vary by rents and coalition size. I conceptualize the latter as cohesive institutions.
Published online by Cambridge University Press
A Political Economy Approach
Variation in Rents Another pathway evaluates the effect of institutions on political stability as rents vary. Here, the existing literature identifies two divergent effects associated with changing levels of rents on civil war. The first is the converse of the politically stabilizing role of rents discussed in the previous section. Here a reduction in rents weakens the government’s capacity to thwart insurgency and/or funds to buy off regime supporters and rivals. Relatedly, rents can mitigate civil conflict by providing a “cushion” to government spending during slumps in the “private” economy. One such example is a well-timed foreign aid package (or opportune increase in commodity prices) that can provide governments with the necessary financial windfall to make rebellion less attractive for opposition groups (Savun and Tirone ). Of course, if these rents disappear the prospect for civil war may be elevated. Citing the commitment problem articulated by Robert Powell (), Richard Nielsen et al. () argue that a severe drop in government revenues (e.g., stemming from a rapid decline in foreign aid) can weaken the government’s capacity to effectively buy off rebel groups and contribute to “inefficient conflict.” They leverage matching techniques to provide cross-national corroboration, documenting how negative aid shocks accelerate the likelihood of civil war. The other path links rents to heightened conflict. This account posits that rents increase the size of the government “pie,” and if there are multiple groups dividing the pie, rent-seeking can contribute to increased fighting over it (Grossman ). Several studies formalize this central idea, identifying the critical role of preexisting political institutions in shaping conflictual behavior. For example, Aaron Tornell and Philip Lane () formalize a “voracity effect” by which an increase in a government’s fiscal resources (e.g., due to a natural resource boom, foreign aid windfall) incites groups within societies with “weak” institutions to compete violently for and squander the windfall and reduce economic growth. Relatedly, Jacob Svensson () develops a repeated game with stochastic shocks in unearned income, where the increase in rent-seeking arises due to coordination failure across interest groups. This coordination failure is often heightened in ethnically diverse societies. For example, Roland Hodler () constructs a model where a surge in rents fosters Cournot-type competition in the form of group fighting in ethnically fragmented societies with a corresponding elevated risk of conflict and lower growth. In these models, political institutions are unable to get incumbent governments to credibly commit to allocating state resources to stave off violence. As I discuss below, this commitment problem is characteristic of societies with less cohesive political institutions. More generally, the conditional effects of rents on political stability (associated with the models above) fall under a broader theory of an “institutions curse” in which rents received in societies with less cohesive institutions are
Published online by Cambridge University Press
Analytical Framework
prone to pernicious political economy (Menaldo ). As discussed above, in some of these accounts a reduction in rents leads to civil war, while in others higher rents do. Moreover, the likelihood of civil war in the presence of an increase or decrease in rents often depends on the quality of “cohesive political institutions.”
D Cohesive Institutions But what are cohesive political institutions? Cohesive institutions describe how an incumbent government “shares” state resources within its territory (Besley and Persson a). In general, societies with more cohesive institutions allocate a more equitable share of state resources to groups that are out of power (i.e., opposition groups). As Besley and Persson (b) note, real-world examples of societies with more cohesive institutions are those with “more minority protection through a system of constitutional checks and balances, through a parliamentary form of government, or through a proportional electoral system” (). This categorization clearly situates democracies as being more cohesive and dictatorships as being less cohesive. In formal models, the degree of cohesive institutions can be captured with a bounded (continuous) parameter; for example, θ 2 ½0, 1 where a higher value of θ corresponds to more cohesive institutions. Thus, if θ ¼ 1=2, then state resources are transferred equally between supporters of the incumbent and the opposition. In contrast, if θ ¼ 0 the incumbent allocates all the state’s resources to its supporters and none to supporters of the opposition. Empirically, scholars have documented important determinants/correlates of cohesive institutions. For instance, governments in more ethnically, religious, and/or linguistically diverse societies tend to supply higher amounts (shares) of public goods to individuals that share the same group affiliation/identity as the incumbent (Alesina, Baqir, and Easterly , Baldwin and Huber , Habyarimana et al. ). Thus, many “fractionalized” societies exhibit less cohesive political institutions.
Although he does not use the terminology of cohesive institutions, Victor Menaldo () provides an excellent account of an institutions curse associated with natural resource rents. Faisal Z. Ahmed () develops a parallel account of an institutions curse with various forms of international capital (e.g., remittances, foreign aid, and investment). Relatedly, correlates of domestic political institutions can also generate a broader “conditional resource curse.” This may arise from other features of a country’s domestic political economy, including the strength of opposition forces (Smith ), the level of income inequality (Dunning ), ownership structure of natural resources (Luong and Weinthal ), the quality of electoral institutions (Mahdavi ), and degree of economic integration (Kurtz and Brooks ). An alternative categorization of cohesive institutions can situate it along Acemoglu and Robinson’s () concept of “inclusive” institutions. As these scholars document, societies with more inclusive institutions share power and state resources (public finances) across various societal groups.
Published online by Cambridge University Press
A Political Economy Approach
Subsidies and transfers (% Expenditures) 20 40 60 80
SPN GMY CZR SWZ NTH UKR VEN MLD
BLR
0
ITA CAN POL SLV SER SWD CHL BUL KZKRUS MON ROK USA URU ALB SLO BEL BRA AZE MAC CRO UGA HUN EST GRG UKG BOS FIN RUM MAL IND NOR ALG SAF LAT EQG GRC AUS FRN IRE TUN ARM INS TAJ NEW JPN COLTUR AUL BOT KUW BOL IRN TRIISR ETI BNG NIC MAS BUI PAN POR MOR PER DEN PAR KYR SRI NIG PNG CYP COS LES GUALEB SEN GHA ZIM FJI CEN SIN SYR CAM EGYANG QAT SWA JOR TOG LBR ZAM THI PHI SAL HON PAK RWA NAM GAM SIEBEN MAG KEN MLI BAH LAO BFO OMA BHU AFGIVO ZAI JAM
0
.2
.4 .6 POLITY (normalized)
.8
1
FIGURE . Democracy and public goods Note: Country average based on author’s calculations. Data on subsidies and transfers is from the World Bank ().
As noted above, another important correlate of cohesive institutions is a society’s quality of democracy: Societies with more cohesive institutions tend to exhibit a higher quality of democratic governance and institutions. This association is unsurprising as many theories predict that governments in democracies will allocate a greater share of their resources to public goods, such as expenditures on welfare payments and subsidies to the population (e.g., Acemoglu and Robinson , Bueno de Mesquita et al. , Lake and Baum , Olson ). Given this incentive, democratic governments correspondingly tend to exert greater “tax effort”: They derive greater share of their revenues from taxation (Besley and Persson a). Using data from the World Bank, Figures . and . corroborate these conjectures for a global cross section of countries. Looking first at government spending, Figure . shows that some of the least democratic countries (e.g., Bhutan, Zaire) allocate less than percent of their expenditures on social transfers. In sharp contrast, the most democratic countries, on average, spend
Greater tax effort is also exerted in resource-abundant democracies, such as Norway and the United States.
Published online by Cambridge University Press
Analytical Framework
60
AUL JPN NEW INS
USA CAN
SAF
Tax income (% revenue) 20 40
QAT ANG
TRI
MAL
ZIM
PNG
TUR
UKG DEN BEL IRE VEN MEX LBR FJI NAM ISR ZAM ITA SYR SPN KEN JAM SIN VIE ROK BOT SLO NTH FIN SWA KZK THI BRA IND AUS GUA CHN NIG MON COL LAO CYP NOR MORALG HON AZE SAL POR POL GHA EQG TOG RUM FRN SEN IVO LES PANPERCHL ETI BEN KYR HUN ZAI MYA SIE EGY RWA CZR GMY TUN BUL ARM GRC SWD BFO PAK NIC EST SWZ MAS BUIUGA SLV BNG LEB SRI COS UKR MLI MAC ARG LAT IRN OMA CEN MAG NEP SER JOR BLR GAM PAR CRO GRG KUW BHU URU ALBCAM AFG RUS BOL BAH TAJ MLD BOS
0
PHI
0
.2
.4 .6 POLITY (normalized)
.8
1
FIGURE . Democracies exert greater tax effort Note: Country average based on author’s calculations. Data on tax income is from the World Bank ().
over percent of their budgets on social transfers. This finding is not driven by greater per capita wealth in more democratic societies. The positive relationship between democracy and social transfers remains statistically robust in a regression that controls for log GDP per capita, as well as with other common measures of democracy, such as constraints on the executive (Acemoglu and Johnson , Gledistch and Ward ). Relatedly, Figure . shows that governments in more democratic polities exert greater tax effort. In this figure, the positive association between the share of government revenue collected from income taxes and a higher normalized POLITY score is consistent with predictions from several prominent political economy models of public finance based on the median voter theorem (Meltzer and Richards ), electoral incentives (Lizzeri and Persico ), credible commitments to redistribution (Acemoglu and Robinson ), and selectorate theory (Bueno de Mesquita et al. ). Taken together, the purpose of the Figures . and . is to simply illustrate how variation in democracy captures the core feature of cohesive institutions in the contemporary era: the extent to which state resources are equitably
Calculated percentage based on average across countries with a normalized polity score greater than or equal to ..
Published online by Cambridge University Press
A Political Economy Approach
distributed across individuals (groups) that support and oppose the incumbent. That said, it is worth noting that cohesive institutions may also be correlated with (and endogenous to) many historical and contemporary conditions within a given society, such as its climate/geography (), precolonial state capacity (Hariri ), “accumulated democratic experience” (Gerring et al. ), factor endowments (Acemoglu and Robinson , Engerman and Sokoloff , Menaldo ), colonial legacy (Olsson ), and structure of the international system (Boix ). And, as the cited scholars demonstrate, these conditions are in turn highly correlated with democracy in the contemporary period. The covariance of these conditions implies that isolating their separate effects poses a critical empirical challenge; a task that Part II of this book addresses in greater detail. As a preview, this is achieved through the research designs (e.g., using panel data, instrumental variables) and choice of control variables. For example, since many correlates of cohesive institutions are country-specific and timeinvariant (e.g., precolonial state capacity and history, colonial legacy, climate, and geography), in econometric analysis with panel data, they can be accounted with country fixed effects. Similarly, controlling for year (or period) fixed effects can capture the effects of important temporal factors that affect all societies, such as the structure of the international system.
E Theoretical Puzzle The discussion in the preceding sections generates an important theoretical puzzle stemming from three core quandaries: () Rents can be a source of political stability, especially as a means for dictatorships to increase their repressive capacity and provide transfers to their key supporters. () Some theories, however, predict that higher levels of rents heighten the propensity for civil war, while others predict the opposite: A drop in rents makes civil war more likely. () These politically destabilizing effects tend to be elevated in societies with less cohesive societies. Surprisingly, these divergent paths have yet to be reconciled in a unified analytical framework. Thus, I seek to answer the following theoretical puzzle: Under what conditions should an increase or decrease in rents lead to political violence? In collaborative work with Daniel Schwab and Eric Werker, we advance a formal model to reconcile this tension in existing theoretical work and also
For example, the use of instrumental variables can help mitigate concerns with omitted variables (Angrist and Pischke ).
Published online by Cambridge University Press
Analytical Framework
provide a peaceful pathway for democratization (Ahmed, Schwab, and Werker ). In particular, the model derives plausible equilibrium conditions under which variation in rents and the quality of cohesive institutions can lead to differential trajectories of democracy (peace), dictatorship (repression), and/or civil war in a unified theoretical framework. Building on the now canonical model of state capacity and political violence developed by Timothy Besley and Torsten Persson (, , a,b), the model is driven by competition for political power between two groups within a country – an incumbent and an opposition – where the winner distributes state resources disproportionately to its own members depending on the cohesiveness of institutions. Consistent with the discussion in the previous section, in the model more cohesive institutions correspond to a more equitable distribution of state resources across society by the incumbent. The model assumes that it is not possible to credibly commit in advance to a more equitable redistribution, so both groups have an incentive to use violence to increase their likelihood of gaining or holding power. In doing so, this assumption captures an important dimension of politics in democracies and dictatorships (Svolik ). Like Besley and Persson’s formulation, my model generates three outcomes: peace/ democracy (no violence), repression (one-sided violence inflicted by the incumbent), and civil war (two-sided violence involving the opposition fighting the incumbent). While the model identifies nontax income (rents) as an important source of these political outcomes, I am careful to capture other important features that can affect politics, such as per capita income (through wages). The model extends the framework from Besley and Persson to allow for the possibility that the incumbent can deter attacks by the opposition when government resources are sufficiently high. The model is essentially a one-shot sequential game, and when rents are moderate, an incumbent will respond to an insurgency with only moderate violence, which encourages an attack by the insurgents. When rents dwarf the costs of fielding a large army, the incumbent would be willing to spend enough to squash an insurgency, and the insurgents are deterred from attacking. To preview my predictions, whereas existing models predict a hierarchy of violence in which higher levels of rents monotonically increase the likelihood of political stability (e.g., Smith ) or conflict in dictatorships (e.g., Besley and Persson b), I provide conditions for differential and non-monotonic
In the analysis to follow, references to “my model” (or using “I”) reflect collaborative work with Daniel Schwab and Eric Werker. In extensions to the model (e.g., where players may incorporate “common knowledge” acquired via repeated interactions), there may be Pareto dominant equilibria. The degree of cohesion can be shaped by several factors, including historical legacy (e.g., colonial history, premodern state capacity), factor endowments (e.g., land inequality, natural resources), and ethnolinguistic and religious fractionalization. The empirical analysis in subsequent chapters will discuss and control for these factors.
Published online by Cambridge University Press
A Political Economy Approach
effects. My model generates two key predictions. First, consistent with many existing accounts I show that higher levels of rents raise an autocrat’s repressive capacity and ability to maintain authoritarian rule. Second, upon the reduction of these rents (to medium or low levels), these regimes are susceptible to challenges from opposition groups. In particular, the flexibility of the model allows for two paths from a reduction in the level of rents: a heightened incidence of civil war or peaceful transitions to democracy. The particular outcome hinges on the preexisting cohesion of political institutions. Violence is more likely in societies where preexisting institutions are less cohesive, while a peaceful transition to democracy is more likely in societies whose preexisting institutions are more cohesive. To the best of my knowledge, this differential and non-monotonic effect is novel in the theoretical literature on rents and political stability. For interested readers, I present this model in the next section. Section . distills the model’s main empirical implications and previews the research designs to evaluate them.
. In this section, I advance a parsimonious game theoretical model to illustrate how () higher levels of rents can entrench dictatorship and () the conditions under which their decline can engender civil conflict or democracy. For the latter, I show that countries with less cohesive institutions are prone to conflict when rents decline, while those with more cohesive institutions can transition peacefully to democracy. A Basic Setup As already noted, my model closely follows a series of theoretical formulations of state capacity and political violence developed by Besley and Persson (, , a,b). In any society, there are two groups ðI, PÞ, each with a population normalized to . Each member of the population earns a wage ðwÞ which is set to . The incumbent (denoted, I) starts the period in power, while the other group ðPÞ is in the opposition. Ending the period in power allows the incumbent to favor its own group when distributing state resources ðRÞ, which is an exogenous parameter. State resources ðRÞ can be thought of as nontax government income (rents) that may come from domestic sources (e.g., income from natural resources,
For instance, Bueno de Mesquita and Smith (), Dunning (), Menaldo (), Morrison (), and Ross (, ) do not generate these predictions. Since the model’s comparative statics will focus on analyzing variation in cohesive institutions and rents setting wages to simplifies the subsequent mathematical derivations. That said, allowing wages to vary would not affect the comparative statics analysis of institutions and rents.
Published online by Cambridge University Press
Analytical Framework
such as oil) and from abroad (e.g., foreign aid, sovereign borrowing). In certain settings, state resources could also include external security guarantees – “geopolitical rents” – available to the incumbent, such as the presence of foreign troops in their territory and diplomatic cover to pursue repressive domestic policies (e.g., imprison political opponents). Unlike material resources such as revenues from natural resources or from foreign aid, geopolitical rents may be tied exclusively to safeguarding the incumbent and are not shared with the opposition. For example, the presence of US troops in certain Persian Gulf countries may serve to better ensure US foreign policy preferences (e.g., a stable supply of oil, peaceful relations with Israel) and tacitly help support the incumbent regime’s continued rule. If that incumbent loses power to the opposition, the US may withdraw military support; in which case, geopolitical rents would be unavailable to the opposition. Both the incumbent and the opposition have access to a violence technology: an army of size A 2 ð0, 1Þ that costs wA to raise. I assume that there are two sizes of armies available for the incumbent, L and H (for low and high), and one for the opposition, which is the same as the incumbent’s low level. Limiting army sizes to discrete options is primarily for tractability, but it does also have some real-world applicability; for example, a government may decide between allowing rebels to hold zero or one city, or they may choose whether or not to use its air force. Allowing the incumbent access to violence levels , L, and H means that there is a possibility that an incumbent’s response to violence is dependent on the level of rents. I assume that the opposition first chooses whether to attack or not, which the incumbent observes, and then the incumbent chooses its own level of violence. If the investment needed for the incumbent to field a large army is sufficiently close to its investment for a small army, then the opposition will never attack. The reason is that the incumbent would employ a large army against an attack because the expenditure would be small relative to the gain
I abstract from the important question of intragroup inequality and assume that all members of a group receive an equal share of the group’s resources. I provide evidence consistent with this conjecture in Chapter . Egypt is an example in a nonoil-producing exporter. During the Arab Spring, the US government effectively withdrew support from the democratically elected Morsi government after the demise of the US-backed Mubarak regime. Upon the resumption of military dictatorship, the USA resumed its support of the regime. This is an important modification to Besley and Persson (): In that paper, the incumbent can choose to attack or not, but has no flexibility on army size if it does attack. In that paper and in this chapter, assumptions on the parameters and the conflict function ensure that in equilibrium the incumbent always responds violently to an insurgency. In Besley and Persson (a,b), the choice of violence is continuous, so the incumbent’s choice of violence can depend on unearned income. However, other assumptions in that paper rule out the possibility that opposition violence increases as unearned income decreases. This timing is the same as Besley and Persson () but differs from Besley and Persson (b) where the two groups attack simultaneously.
Published online by Cambridge University Press
A Political Economy Approach
from retaining power, and so the opposition would be deterred. As I show in the model’s proofs (in the chapter’s Appendix), the relevant threshold is that there is never two-sided violence when H < 3L. Since opposition groups do sometimes mount an insurgency, I assume H > 3L. This ensures that two-sided violence does occur for some values of R. Power changes hands stochastically according to the conflict function γ AP , AI 2 ½0, 1, where either side can improve their chances by investing in violence. The following specifies the functional form as: 8 I P > >0 A > A P I < I P γ A ,A ¼ 1 A < A > > : 1 AI ¼ AP 2 This functional form means the group choosing the higher level of violence wins power with certainty, and if they choose the same level, then power changes hands with probability 12. At the end of the period, the level of rents R is divided between the two groups, and utility is linear in consumption, meaning I can treat the agents as risk neutral. The group that ends the period out of power gets θR, and the inpower group gets ð1−θÞR, where 0 θ 12 represents equality of institutions. Importantly, a higher value of θ implies institutions are more cohesive. Note that θ is assumed to be “exogenous” in the model, which is unlikely to be the case empirically. Cognizant of this concern, Part II of this book strives to establish a plausibly exogenous determinant of cohesive institutions that is unrelated (“causally prior”) to contemporary levels of rents. An important distinction between the two groups is thatthe group that started in opposition pays for their entire military expenditure AP , while payment for the incumbent group comes out of the public purse before resources are distributed. Then the game ends and payoffs are realized.
It would be straightforward to introduce incumbent advantage by assuming that power changes hands with probability α when both groups show the same level of violence for some α < 12, but the results do not meaningfully change, so I retain the simpler formulation. Specifically, I use the percentage of territory conquered by Islamic armies (corresponding to contemporary country boundaries) during the period – CE to “explain” contemporary levels of democracy. This percentage of conquered territory is clearly uncorrelated with timevarying rent streams in the modern era. The results would be unchanged if financing for the incumbent’s army came from a combination of public and private funding. It is worth noting that the equilibrium outcomes I identify in this single-shot game are also subgame perfect Nash equilibrium (SPNE) in a repeated game where each generation cares about other generations (to some degree). It can be shown that in a repeated game, the equilibrium outcomes identified in the single-shot setting comprise the “worst case” equilibrium. Future work could investigate the existence of Pareto dominant SPNE in the repeated game. For example, if one were to take the standard approach to repeated games and presume that the history of play is common knowledge between the players throughout the game, then one could
Published online by Cambridge University Press
Analytical Framework Opposition Peace
Attack
Incumbent Peace
Payoffs:
]
FIGURE
. Game tree
, ]
Incumbent Repression
[q(R–L), (1–q)(R–L)]
Attack with L
[
(
) (
,
)
Attack with H
]
[q(R–H)–L, (1–q)(R–H)]
The two groups’ respective payoffs are the sum of their wage, their expected receipts of nontax income, and for the opposition, the cost of financing an army. Since the wage is always one by assumption, it is omitted for clarity. The payoff for the incumbent is: V I AP , AI ¼ 1 γ AP , AI ð1 θÞ R AI þ γ AP , AI θ R AI and the opposition’s payoff is: V P AP , AI ¼ 1 γ AP , AI θ R AI þ γ AP , AI ð1−θÞ R AI −AP :
B Optimal Investment in Violence In this section I solve for the optimal choice of violence for each group. Since this is a one period game with two actions, I start at the end of the game and work backward. Figure . illustrates the corresponding game tree, where the opposition moves first. The last row in the figure states the payoffs for each series of actions, with the opposition’s payoff listed first in parentheses. I first consider the case where the opposition does not attempt an insurgency. The incumbent then chooses between peace and using one-sided violence, which is interpreted as repression. A peaceful incumbent realizes a payoff of: 1 1 R V I ð0, 0Þ ¼ ð1−θÞR þ θR ¼ 2 2 2
use the Folk Theorem to establish the existence of Pareto superior SPNE. I leave this interesting line of inquiry to future work.
Published online by Cambridge University Press
A Political Economy Approach
TABLE
. Incumbent’s optimal choice in political violence
R
Response to peace
Response to violence
L 1−2θ
ð1−θÞ R 2L1−2θ
L
L
L
L
H
0R L 1−2θ
2Lð1−θÞ 1−2θ
1−θÞ−L R 2H ð1−2θ
1−θÞ−L R > 2Hð1−2θ
Note: The first column shows a range of values for R. The next two columns show the incumbent’s optimal response to peaceful or aggressive opposition. Recall, L is a low deployment of armed forces. H is a high deployment of armed forces.
while a repressive incumbent is guaranteed to retain power, generating a payoff of: V I ð0, LÞ ¼ ð1 θÞðR LÞ: Note that the conflict function means the incumbent would never use a high level of violence against a peaceful incumbent, because high or low violence both guarantee success. The main aim of the model is to clarify the impact of rents on violence choices, so I allow R to vary and hold all other parameters fixed. The incumbent will be peaceful in response to peaceful opposition when: V I ð0, 0Þ V I ð0, LÞ which simplifies to:
ð1−θÞ Peace is the incumbent’s optimal response to peace if 0 R 2L1−2θ .
ð1−θÞ Low violence is the incumbent’s optimal response to peace if R > 2L1−2θ .
I assume that an indifferent group will break ties by choosing the lower level of violence. Similarly, if the opposition is violent, then the incumbent chooses between peace, low violence, and high violence. In the appendix to this chapter, I show Table . presents the incumbent’s optimal response to a peaceful or violent opposition: I now solve for the opposition’s optimal choice for every value of R. If L , then the opposition knows the incumbent will be peaceful, 0 R 1−2θ regardless if it attempts an insurgency. Thus, the opposition chooses between V P ð0, 0Þ ¼ 12 R and V P ðL, 0Þ ¼ ð1 θÞðR LÞ. Over the relevant range of R, the former value is larger; meaning that when L both the opposition and incumbent will be peaceful. This process can 0 R 1−2θ be continued for every value of R, by solving the incumbent’s response to peace and violence, and then determining which choice maximizes the opposition payoff. Table . presents the payoffs (see the chapter’s Appendix for the proofs).
Published online by Cambridge University Press
Analytical Framework TABLE
. Incumbent and opposition optimal choices
Interval
R
Inc. response to
Inc. response to L
Opp.
Inc.
L 0 R 1−2θ
L 1−2θ
L
2Lð1−θÞ 1−2θ
L
L
L
2Lð1−θÞþL 1−2θ
L
L
L
L
L
H
H
ð1−θÞ R 2L1−2θ
R>
1−θÞþL < R 2Lð1−2θ 1−θÞ−L < R 2Hð1−2θ
2H ð1−θÞ−L 1−2θ
Note: The first column is a label for the range of R (unearned resources) being described. The second column is that range. The next two columns are the incumbent’s response to a peaceful or violent opposition. The final two columns are the actions that the opposition and the incumbent would actually take. The first two rows are both labeled as Interval because the outcome (both groups peaceful) is the same in both.
As R ranges from to infinity, there are four relevant intervals. In the first interval, both groups are peaceful because the prize for winning power is too small to justify fighting. There are two rows in Table . that make up Interval because the incumbent would be peaceful in response to rebellion in the first row and violent in response to rebellion in the second row. But in both rows the actual outcome would be peace on both sides. The incumbent will be tempted to use violence at a lower threshold of R than the opposition because the incumbent funds its army using public funds, and in Interval the opposition is peaceful while the incumbent uses the smaller army size against peace or violence, so the result is repression. In Interval the opposition would still always use a small army but now R is sufficiently large that the opposition is also willing to use, and both groups use low violence (civil conflict). In Interval R has increased to the point where the incumbent would employ a high level of violence in response to rebellion, but low violence in response to peace. It is disastrous for the opposition if the incumbent uses a high level of violence, because the incumbent is guaranteed to hold power and uses up state resources as well. This discourages the opposition from attacking, so the result is low-level repression again, the same as in Interval . These dynamics may arise in natural resource dependent countries where an incumbent and opposition fight over the rents (e.g., Angola, South Sudan, Colombia). This result – that violence can fall as R increases – is driven by two factors. First, since the incumbent has two choices of army size, as R rises to Interval the incumbent’s response to an attack can increase from a small to a large
As I demonstrate in Chapter , higher resource dependence on oil revenue is not a robust determinant of high intensity civil war in the Persian Gulf.
Published online by Cambridge University Press
A Political Economy Approach
army size, which can deter the opposition. Second, the result stems from the sequential nature of the groups’ decisions, because the incumbent cannot respond to the opposition if they act simultaneously.
C Transitions to Democracy or Civil War The aim of Table . is to depict how the incumbent’s and opposition’s optimal choices change as R varies, holding the size of armies (L, H) and cohesive institutions ðθÞ fixed. Figure . also shows optimal choices but allows θ to vary. Specifically, I plot the four intervals depicted in Table ., while θ ranges from . and . and L and H remain fixed at . and . respectively. Going forward, I use the term “transition” to describe how changes in the model’s exogenous parameters (R or θ) leads to changes in violence (e.g., movements from repression to civil war, repression to peace). Figure . demonstrates how a country can transition between peace, repression, and conflict as rents and cohesive institutions vary. In general, higher levels of rents tend to increase the intensity of political violence: Holding fixed the level of θ, “moving up” the y-axis makes civil conflict more likely. This monotonic effect is consistent with the predictions of Besley and Persson (, , a,b). That said, Figure . demonstrates the possibility of nonmonotonic effects whereby a reduction in rents can make peace or civil conflict more likely. Whether a country experiences peace or conflict depends critically on the preexisting level of institutional cohesion. Transition to Democracy To observe a transition to democracy associated with a decline in rents, consider a country initially in Interval (low repression). In this instance, a reduction in rents can shift the country to Interval in which the incumbent decommissions its army (i.e., AI changes from L to ) and peace is achieved: a condition conducive for democracy. Moreover, observe that this peaceful transition to democracy is more likely to occur at higher levels θ. Since θ is “exogenous” in this model, this implies that a reduction in rents in societies with a higher preexisting (exogeneous) level of cohesive institutions can experience a peaceful democracy.
In Besley and Persson (, b), it is not possible for a decrease in rents to lead to an increase in violence. The incumbent in Besley and Persson () has only one possible army size so the deterrent effect observed in my model cannot occur. The two sides choose army size simultaneously in Besley and Persson (a), so there cannot be an exactly analogous deterrence effect, and the authors’ assumptions limiting the strategic complementarities and substitutabilities in the conflict function ensure that both groups’ army choices are weakly decreasing as R decreases. These numerical values are chosen for illustrative purposes, and other examples give similar results.
Published online by Cambridge University Press
2.5
Analytical Framework
R
1.5
2
Repression (Interval 4)
1
Conflict (Interval 3)
0.5
Repression (Interval 2)
0
Peace (Interval 1)
0.2
0.25
0.3 q
0.35
0.4
FIGURE . Political violence as a function of cohesive institutions ðθÞ and rents (R) Notes: This figure simulates the violence outcomes as a function of θ (cohesiveness of institutions) and R (rents). The values of L and H (fraction of workforce deployed for low and high army size) are set to . and ., respectively. In Interval , both groups are peaceful. In Interval , the incumbent uses low violence in response to peace, and would have used low violence in response to violence from the opposition. Both groups use low violence in Interval . In Interval , the incumbent uses high violence in response to an attack, so the opposition is deterred and the result is peace from the opposition and low violence from the incumbent (repression). The values of R range from to infinity.
Transition to Civil Conflict Of course, Figure . also illustrates that a reduction in rents can also make civil conflict more likely. To see this, consider a country initially in Interval of “low repression.” In this case, a reduction in R can shift a country to Interval in which both the incumbent and opposition deploy small armies and engage in low intensity warfare. Moreover, being in Interval in the first place is more likely to occur at lower levels of θ. And since θ is assumed to be exogenous in the model, this implies that a reduction in rents in societies with a lower preexisting level of cohesive institutions may heighten the propensity for civil war. To the best of
In Chapter , I argue these features capture important dynamics potentially explaining why many Muslim foreign aid recipients experienced heightened internal political strife – characterized at the extreme with high intensity civil war – in the s and s and remained nondemocratic.
Published online by Cambridge University Press
A Political Economy Approach
my knowledge, these differential and non-monotonic effects associated with a decline in rents – peaceful transition to democracy or heightened propensity for civil war – is novel in the theoretical literature on rents and political stability. Preserving Peace (Democracy) Finally, Figure . also shows that more cohesive institutions can preserve peace at higher levels of rents: a finding consistent with existing models (e.g., Besley and Persson b). When θ ¼ 0:2 there will be peace only when R 0:2667, but when θ ¼ 0:4, an incumbent will not be tempted to repress the opposition as long as R 0:6. This behavior stems from the expectation that the incumbent will be treated fairly even if it loses power, an expectation that underlies the peaceful transition of power that occurs in democracies. Moreover, this means that if two countries are initially both in Interval (repression) with the same level of R, the country with a lower θ will require a larger decline in R to shift to Interval (peace/democracy). In other words, more cohesive institutions can provide some protections against rent-driven civil conflict; a finding consistent with existing models of rent-seeking and political (in)stability (e.g., Lane and Tornell , Svensson ).
. A Propositions The discussion in the preceding section is summarized in the following three empirical implications and two corollaries: Proposition : Rents and Dictatorship Higher levels of rents can foster dictatorship (and other correlates, such as political repression), particularly in societies where preexisting institutions are less cohesive. This corresponds to moving from Interval to Interval or . Proposition : Reduced Rents and Political Stability Reduced levels of rents can shift a country from conflict or repression into democracy (i.e., movement from Intervals – to Interval ), but cannot move a country out of democracy. Proposition : Reduced Rents and Political Transitions as Cohesive Institutions Vary The threshold R below which democracy exists is greater where institutions are more cohesive (i.e., θ is greater). Empirically, this can imply two divergent outcomes that hinge on the cohesion of preexisting institutions: Corollary A: A reduced level of rents is more likely to foster civil conflict in societies where preexisting institutions are less cohesive. This corresponds to a movement from Interval to Interval .
Published online by Cambridge University Press
Analytical Framework
Corollary B: While the same reduction in rents is more likely to lead to democratization in societies where preexisting institutions are more cohesive. This corresponds to a movement from Interval or to Interval . B Empirical Evaluation: A Road Map I evaluate these propositions in Parts II (“Conquest”) and III (“Rents”) of this book to show why so many Muslim majority societies are prone to dictatorship and civil conflict. In evaluating these propositions, I am cognizant that both rents and institutions are endogenous. Accordingly, each part will articulate the challenge posed by endogeneity and present identification strategies to better establish causal relationships. I briefly preview the research designs below. Conquest In Part II (“Conquest”), I establish a plausibly exogenous source of variation in cohesive institutions for contemporary Muslim societies. This is relevant as the level (quality) of cohesive institutions is assumed to be exogenous in my formal model. In Chapters –, I use qualitative historical evidence and cross-national statistical analysis to document the institutional legacy of Muslim conquest. The empirics strive to show how the initial period of Muslim military conquest following the death of Prophet Muhammad (– CE) established authoritarian political structures and then trace its subsequent trajectory of centralized autocracy in “conquered” territories through to the present. Since the process of military conquest and consolidation of centralized autocracy predated the discovery and distribution of rents (e.g., oil, receipts of foreign aid), the historical legacy of these institutions can serve as a source of exogenous variation in cohesive institutions in many contemporary Muslim societies. Thus, the institutional legacy of Muslim conquest may help explain the crosssectional difference in democracy between Muslim and non-Muslim societies; that is, the persistent democratic deficit depicted in Figure .. I interpret this finding to imply that the legacy of Muslim conquest fostered less cohesive
For example, the extraction of natural resource and subsequent generation of rents can be affected by domestic institutions. Menaldo () documents the higher levels of foreign investment and subsequent resource extraction in oil-producing democracies. Thus, the direction of this endogeneity suggests the effect of oil wealth on autocracy may be downward biased. Chapter discusses the emergence and persistence of an “Islamic equilibrium.” The chapter also presents cross-national causal evidence using a novel differences-in-differences research design. In Chapter , I examine the process of reconquest in medieval Spain and adoption of Islam in Indonesia and Malaysia (through trade) to document how the process of military conquest and subsequent rule in conquered territories mattered for the introduction and persistence of the classical Islamic equilibrium.
Published online by Cambridge University Press
A Political Economy Approach
institutions in Muslim societies (exposed to conquest) in the contemporary period. Rents This exogenous source of variation in preexisting cohesive institutions in the modern period allows me to evaluate how variation in rents affects variation in dictatorship and civil war in contemporary Muslim and non-Muslim societies over time (as depicted in Figures . and .). Thus, I can directly test propositions – and corollaries A and B. In Part III of this book (comprised of Chapters –), I demonstrate how temporal variation in rents – oil revenues in oil producers and foreign aid in non-oil producers – explains patterns of dictatorship and civil conflict in Muslim “conquest societies” that are consistent with this chapter’s motivating trends in Figures .–. (and test propositions and A). To test proposition B, I evaluate the effect of rents in nonMuslim societies and demonstrate how their higher levels of preexisting cohesive institutions facilitated their peaceful transitions to democracy when their rents declined. Chapter demonstrates how oil revenues have sustained dictatorship in oilproducing societies that were exposed to Muslim conquest without elevating the propensity for civil conflict. This latter finding can be attributed in part to the presence of geopolitical rents from the United States that have strengthened the repressive capacity of governments in oil-producing Muslim countries to thwart domestic insurgencies. To the best of my knowledge, introducing geopolitical rents is new to the oil curse literature. Endogeneity, of course, looms large in identifying causal relationships. To mitigate these concerns, I follow Kris Ramsay () and use out-of-region disaster damage as an instrumental variable for oil wealth to estimate a causal relationship between oil and political violence, particularly in conquered Muslim territories. This chapter also provide evidence that oil producers with more cohesive institutions (i.e., those not exposed to Muslim conquest) have been able to transition to greater democracy. A finding consistent with proposition . In Chapters and , I probe the effect of rents in non-oil-producing developing countries. In these chapters I focus on rents that emanate from abroad – foreign aid in Muslim societies, “petrodollar” investments in Latin America, and trade subsidies in Eastern Europe – that also co-vary with movements in world oil prices. This second ‘fact’ is attractive in drawing causal inferences. Since oil prices are plausibly exogenous to economic and political conditions in non-oil-producing developing countries, I exploit their variation to predict movements in rents to explain variation in political violence in non-oil
I use data on US troop deployments and qualitative accounts of the US government’s strategic engagement in the Persian Gulf to justify this argument. Limiting the sample to non-oil-producing countries purges the direct effect of oil revenues on politics.
Published online by Cambridge University Press
Analytical Framework
producing countries. I do so by employing difference-in-differences and instrumental variables research designs. In Chapter , I leverage a quasi-natural experiment of oil price driven aid disbursements to demonstrate how higher amounts of foreign aid entrenched autocracy in non-oil-producing Muslim countries (many of whom were conquered by Muslim armies), but their subsequent decline led to civil war. These dynamics are consistent with proposition A (and Figure .). Chapter evaluates proposition B. Using case studies – from Latin America and former Soviet Eastern Europe – and cross-national statistical analysis, the chapter argues that societies with preexisting institutions that are more cohesive transitioned to democracy when their rents declined. The chapter’s empirical analysis looks beyond Muslim societies – thus, providing a broader evaluation of the book’s theoretical arguments to a global sample of countries.
. This chapter presented the book’s analytical framework. It began by describing several empirical patterns that motivate the empirical and theoretical puzzles. These can be distilled to: () What explains the persistent democratic deficit between Muslim and non-Muslim societies? () Across Muslim societies, why do differences in a country’s rent structure (i.e., oil versus non-oil production) and its temporal variation explain movements in dictatorship and civil conflict? Motivated by the second question, the rest of the chapter advanced a theory linking variation in rents and exogenous (preexisting) cohesive institutions to varying intensities of political violence (dictatorship and civil war). Using a formal model, I generated several propositions, three of which may “answer” the two puzzles described above: () Higher levels of rents entrench dictatorship in societies whose preexisting institutions are less cohesive. () A decline in rents elevates the prospect of civil war in societies whose preexisting institutions are less cohesive. () A decline in rents can foster a peaceful transition to democracy in societies whose preexisting institutions are more cohesive. I ended the chapter by previewing the research designs to evaluate these propositions and empirical puzzles. My aim in Part II of the book is to demonstrate how the spread of Islam through military conquest (between
Chapter also introduces a new source of exogenous variation to measure preexisting cohesive institutions in non-Muslim societies.
Published online by Cambridge University Press
A Political Economy Approach
and ) introduced an institutional trajectory that created a ‘sticky’ set of less cohesive institutions in contemporary Muslim societies. Thus, I leverage the historical legacy of Muslim conquest as a plausibly exogenous source of preexisting cohesive institutions to evaluates propositions and corollaries A and B in the contemporary period. Part III of the book does just that by examining how variation in rents explains the persistence of a democratic deficit between Muslim and non-Muslim countries and variation in political violence across oil and non-oil-producing Muslim societies.
Published online by Cambridge University Press
Appendix A
0
POLITY (normalized) 0.2 0.4 0.6
0.8
Additional Figures
1970
1975
1980
1985
1990 Year
Oil FIGURE A
1995
2000
2005
2010
Non-oil
Democracy in non-Muslim countries
Published online by Cambridge University Press
0.35 0.3 0.25 0.2 0.15
0
0.1
0.05
POLITY (normalized)
Civil war 0.1 0.15
0.2
0.25
A Political Economy Approach
1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 Year Civil war
Democracy and civil war in Muslim oil producers
100
0.04 Civil war
0
50
0.02 0 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 Year Civil war FIGURE A
Aid per capita
Civil war and foreign aid in non-Muslim non-oil producers
Published online by Cambridge University Press
Aid per capita
0.06
150
FIGURE A
POLITY (normalized)
0
0
20
0.05
Civil war 0.1 0.15
40 60 80 Oil price (2015 US$)
0.2
100
0.25
Analytical Framework
1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 Year Civil war FIGURE A
Oil price (2015 US$)
Civil war and oil prices in Muslim oil producers
Published online by Cambridge University Press
Appendix B Proofs for the Formal Model
In this appendix, I present the proofs for Section ..
. First, it is important to confirm the ranges of R presented in Table . cover the entire (non-negative) real line without overlap for every valid parameter value of θ, L, and H. That is, I need to show that 2Lð1−θÞ 2Lð1−θÞþL 1−θÞ−L L < 2H ð1−2θ . 1−2θ < 1−2θ < 1−2θ The first inequality holds because 0 < θ < 12 by assumption, so 2ð1−θÞ > 1. The second inequality is true because L > 0. To verify the third inequality, note that 2H ð1−θÞ > 6Lð1−θÞ, because of the assumption that H > 3L. This means 2H ð1−θÞ−L ð1−θÞ 1−θÞ−L 1−θÞþL > 2L1−2θ þ 4Lð1−2θ > 2Lð1−2θ , and implies that Table . covers the 1−2θ non-negative real line without overlap. The assumption that H > 3L guarantees the lower bound for R in Interval of Table . is in fact lower than the upper bound, which ensures that there are values of R which generate twosided violence. I next verify that the stated strategies for the incumbent are in fact optimal for the incumbent. In Section ., I showed that:
ð1−θÞ Peace is the incumbent’s optimal response to peace if 0 R 2L1−2θ .
ð1−θÞ Low violence is the incumbent’s optimal response to peace if R > 2L1−2θ .
On the other hand, if the opposition is violent, then the incumbent chooses between peace, low violence, and high violence. The payoffs are:
V I ðL, 0Þ ¼ θR These proofs are from Ahmed, Schwab, and Werker ().
Published online by Cambridge University Press
Analytical Framework
V I ðL, LÞ ¼ 12 ð1−θÞðR−LÞ þ 12 θðR−LÞ ¼ 12 ðR−LÞ V I ðL, H Þ ¼ ð1−θÞðR−H Þ.
It is straightforward to show that the first value is highest when 1−θÞ−L L L 0 R 1−2θ , the second highest when 1−2θ < R 2Hð1−2θ , and the last value
1−θÞ−L , which confirms the values presented in Table .. is highest when R > 2Hð1−2θ
. Table . shows the optimal choices of both groups as a function of R. The incumbent’s choices in each region follows from Table ., so the only remaining task it so verify the opposition’s choices, conditional on how the incumbent would respond to peace and violence. The opposition’s payoffs for each of the possible outcomes is:
V P ð0, 0Þ ¼ 12 R V P ð0, LÞ ¼ θðR−LÞ V P ð0, H Þ ¼ θðR−H Þ V P ðL, 0Þ ¼ ð1−θÞR−L V P ðL, LÞ ¼ 12 ð1−θÞðR−LÞ þ 12 θðR−LÞ−L ¼ 12 ðR−LÞ−L V P ðL, H Þ ¼ θðR−H Þ−L.
L Consider the first row of Table . where 0 R 1−2θ . The incumbent will be peaceful regardless of whether the opposition attacks, so the opposition 2L , chooses between 12 R andð1−θÞR−L. The former is higher whenever R 1−2θ which includes the entire row of Table .. Next consider the range where the incumbent is peaceful in response to peace but uses low violence in response to violence. The opposition chooses between V P ð0, 0Þ and V P ðL, LÞ. The former is always higher, meaning that in the second row of Table ., the opposition will be peaceful. Next, consider the range of R where the incumbent will use low violence regardless of the opposition’s choice, which spans the third and fourth rows of Table .. A peaceful opposition group will receive V P ð0, LÞ, while an aggressive opposition group will receive V P ðL, LÞ. The former is higher when 1−θÞþL R 2Lð1−2θ , which confirms that in row of Table . the opposition will choose peace and in row it will fight. Finally, consider the region where the incumbent will use low violence in response to peace and high violence in response to violence. The opposition is guaranteed to lose power, so it prefers to remain peaceful, which confirms the final row of Table .
Published online by Cambridge University Press
Published online by Cambridge University Press
Analytical Framework
In Chapter , I identified two tragic characteristics of many contemporary Muslim societies: the pervasiveness of dictatorship and an elevated propensity of civil war. Among political economists, the prevalence of these outcomes – dictatorship and civil conflict – captures the central features of societies afflicted with political violence. Accordingly, in this chapter, I adopt this terminology and develop an analytical framework to better understand why many Muslim societies are inflicted with varying and heightened levels of political violence. I begin with several figures to motivate the analysis at the heart of this book. These figures demonstrate the cross-sectional and temporal variation in political violence between Muslim and non-Muslim societies and across Muslim societies with different rent structures (i.e., oil versus non-oil producers). These figures suggest a relationship between foreign-based rents – notably, oil revenues and foreign aid – and patterns of dictatorship and civil war within Muslim societies. With this empirical motivation, I then advance a theory linking variation in preexisting cohesive institutions and rents to explain patterns of dictatorship, democracy, and civil war and transitions between these outcomes. This discussion proceeds in a stepwise fashion.
For example, in a series of theoretical formulations, Besley and Persson (, , a,b) model a hierarchy of political violence that ranges from peace, repression (one-sided violence), and civil war (two-sided violence). Relatedly, Svolik (, ) describes how dictators govern in societies in which “violence is an ever-present and ultimate arbiter of conflicts in authoritarian politics.” Rents refer to nontax government revenue. This includes revenues from the extraction and sale of natural resources (e.g., oil, natural gas) that can accrue to national governments and the receipt of foreign economic and military aid. Rents can also include sovereign borrowing and subnational transfers (Morrison ). In some instances, rents can be non-economic, for example geopolitical rents in the form of security guarantees (e.g., foreign troop deployments) that can augment a
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
Analytical Framework
I first distill the rich scholarship on rents and political stability, identifying an important tension in existing theoretical formulations: whether a reduction in rents contributes to violence or not. This discussion in turn provides the rationale for advancing a formal theory (model) to help reconcile this tension. Whereas existing models predict a monotonic effect linking greater rents to political violence, my model allows for non-monotonic effects. Consistent with these existing theories, I show that higher levels of rents can sustain autocracy, but unlike these models, I show how declines in rents can generate divergent political trajectories. Specifically, I articulate the conditions under which a decline in rents can generate a peaceful transition to democracy or engender civil war. The model highlights the important role of preexisting cohesive institutions in shaping how a decline in rents can lead to these divergent outcomes: democracy or civil war. To the best of my knowledge, the model advanced in this chapter is the first to generate these differential trajectories in a unified analytical framework. The chapter’s unified framework (model) is an important contribution to existing scholarship. As this chapter describes, existing scholarship has tended to study whether rents and institutions affect patterns of democracy/dictatorship or civil war as distinct and separate outcomes (e.g., Hodler , Menaldo , Smith ). This is surprising since regime type and civil war tend to covary: dictatorships are more prone to civil war (e.g., Blattman and Miguel ). I end the chapter by delineating the model’s empirical implications and preview the research designs to evaluate their capacity to explain patterns of dictatorship and civil war in Muslim and non-Muslim societies (which is the focus of Parts II and III of the book).
. : In this section, I present several figures describing patterns of political violence in Muslim and non-Muslim societies that motivate the core questions (“puzzles”) that this book seeks to explain. The first figure plots the average
government’s capacity to maintain political control. For example, the stationing of US troops in various countries in the Persian Gulf (e.g., Saudi Arabia, Kuwait) can serve to safeguard stable world oil supplies and also serve to deter domestic insurgencies (that could disrupt these oil supplies). While I refer to this model with “my” and “I” in this chapter, it was developed in collaborative work with Daniel Schwab and Eric Werker. See Ahmed, Schwab, and Werker (). Societies with more cohesive institutions are those that share more state resources with groups out of power (e.g., supporters of opposition political parties, different ethnic or religious groups). I discuss cohesive institutions in Section ..
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
A Political Economy Approach
level of democracy in Muslim- and non-Muslim-majority societies since . To measure a country’s level (quality) of democracy, I use its composite POLITY score (Marshall and Jaggers ). This index and its sub-indices are used widely in empirical research on political regimes, institutional design, variation, and durability. The POLITY score combines information about the competitiveness of and constraints on executive recruitment with information about the extent of political competition. The measure ranges from − (full dictatorship) to + (consolidated democracy). I normalize each country’s annual score on a [,] scale where a higher value corresponds to greater democracy. Countries where percent or more of their population identify with Islam are classified as Muslim.
A Democratic Deficit in Muslim Societies Figure . reports the average normalized POLITY score across Muslim and non-Muslim countries on an annual basis since . The figure shows that Muslim societies have, on average, been less democratic than non-Muslim societies. For example, using conventional classification schemes in economics and political science, the typical Muslim country has always been a dictatorship: its normalized POLITY has not exceeded .. In contrast, non-Muslim societies have exhibited greater temporal variation in their normalized POLITY score. While the typical non-Muslim country started out as a democracy in , it became less democratic over the subsequent decade. In the mid-s, non-Muslim societies gradually became more democratic on average, a pattern consistent with the so-called Third Wave of democratization (Huntington ). This upward trend accelerated after the end of the Cold War in and has stabilized since . Despite some improvements in democratic governance in Muslim societies after , Figure . notably reveals a persistent “democratic deficit” in Muslim countries (relative to their non-Muslim counterparts).
See for example, Acemoglu and Robinson (), Ahmed (), Besley and Persson (a), Dunning (), and Morrison ().Bueno de Mesquita et al. () use sub-indices of the composite POLITY score to construct their measure of selectorate and coalition size. Alternate measures of democracy (e.g., Boix, Miller, and Rosato ) are contrasted and evaluated in relation to POLITY. Note that my use of “typical” refers to the statistical average across each group of countries (Muslim, non-Muslim). Scholars often use cutoffs in the POLITY index to differentiate dictatorships from democracies. Economists frequently classify countries with a normalized POLITY score greater than or equal to . as a democracy and those below . as a dictatorship. In contrast, political scientists employ a higher cutoff for democracy: a normalized POLITY score equal to or exceeding . (which corresponds to a POLITY score equal to or above +).
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
0
.2
POLITY (normalized) .4 .6
.8
1
Analytical Framework
1960
1965
1970
1975
1980
1985 Year
1990
Non-Muslim FIGURE
1995
2000
2005
2010
Muslim
. Persistent democratic deficit in Muslim societies
B Cohesiveness in Muslim Societies The democratic deficit identified in Figure . is associated with less cohesive institutions in Muslim societies. Figure . provides some suggestive evidence using welfare spending as a measure of cohesive institutions (like Figure .). The figure shows that governments in Muslim societies, on average, allocate a persistently lower percentage of their expenditure on welfare spending relative to non-Muslim societies. Over the entire period, the difference in shares (of about percentage points) is statistically different from zero (with a p-value < .). This implies that Muslim societies tend to be significantly less cohesive relative to non-Muslim societies. It is worth emphasizing that reliable crossnational data on welfare spending is available from and is therefore less extensive than measures of democracy, which are available from (and for many countries for decades prior to ). As democracy and cohesiveness tend to be positively associated (see Figure .), my analysis going forward will
Recall that cohesiveness refers to the proportion of state revenue (expenditure) shared between supporters of the ruling government and those in opposition. In societies with less cohesive institutions, the ruler is less egalitarian in sharing state revenue with supporters and opposition groups. That is, the ruler tends to spend a lower share of expenditure on broad-based subsidies and transfers to the masses.
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
0
Subsidies and transfers (% Expenditures) 10 20 30 40
A Political Economy Approach
1990
1995
2000 Year Non-Muslim
FIGURE
2005
2010
Muslim
. Cohesiveness in Muslim and non-Muslim societies
use democracy as a proxy of cohesiveness, primarily since there is more extensive cross-national and temporal data coverage for this measure.
C Political (In)Stability in Muslim Societies The relative stability of dictatorship in Muslim societies suggested in Figure ., however, seems to mask differences in their underlying rent structure. To see this, Figure . plots the average normalized POLITY score across oil and nonoil-producing Muslim societies (I use British Petroleum’s designation of oilproducing countries). This figure reveals several interesting patterns regarding the relative stability of dictatorship across oil and non-oil-producing Muslim societies. First, while oil and non-oil-producing Muslim societies exhibited, on average, similar levels of democracy in , oil producers became increasingly less democratic in the s. Second, despite some improvements in democracy in the s, non-oil-producing Muslim societies reversed this trend in the s. And by the early s, Muslim non-oil producers were less democratic than their oil-producing countries. Third, Muslim non-oil producers experienced another reversal: in the mid-s they started to become slightly more
As I discuss in Chapter this trend may be due to the nationalization of oil companies in the s and the subsequent surge in oil income to national governments (Mahdavi ).
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
0
POLITY (normalized) .2 .4 .6
.8
Analytical Framework
1960
1965
1970
1975
1980
Oil FIGURE
1985 Year
1990
1995
2000
2005
2010
Non-oil
. Democracy in Muslim oil and non-oil producers
democratic and by the early s passed the threshold (barely) for democracy. Fourth, Muslim oil producers did not experience these improvements in democracy. By , these societies tended to be less democratic than they were in . The greater variation in democracy among Muslim non-oil producers prompts greater scrutiny. Figures . and . provide some suggestive explanations. Figure . plots the average incidence of civil war in Muslim non-oil producers (left axis) against their average level of democracy (right scale). It shows that Muslim non-oil producers were relatively peaceful in the s and s (when they also became less democratic), but became increasingly conflict-prone in the mid-s until the late s. During this period, these
Similar patterns are not observed in non-Muslim countries. Since the s, both oil and nonoil-producing countries, on average, have become more democratic over time. Many of these oil producers have been consolidated democracies since the s (e.g., Norway, United States) and only recently discovered significant quantities of oil (e.g., the United Kingdom in the North Sea in the s). See Figure A in this chapter’s appendix. Data on civil war is from the Armed Conflict Database (Harbom, Melander, and Wallensteen ). Following convention, conflicts exceeding , battle deaths per annum comprise civil war. In contrast, battle deaths below the , threshold comprise civil conflict. Similar patterns are not observed in Muslim oil producers (see Figure A in the chapter’s Appendix). These countries have remained relatively stable politically compared to Muslim non-oil producers. In Muslim oil producers, the average incidence of civil war has hovered
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
0
.2
.05
Civil war .1
.15
.3 .4 .5 POLITY (normalized)
.2
.6
A Political Economy Approach
1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 Year Civil war FIGURE
POLITY (normalized)
. Civil war and democracy in Muslim non-oil producers
societies were twice as likely to experience civil war relative to the prior two decades. Concomitantly, these societies started to become more democratic as the average normalized POLITY score started to rise. With these improvements in democracy, by the early s civil strife in these countries subsided significantly. A potential source for the patterns in Figure . may be attributable to variation in a specific type of unearned (nontax) government income received in many non-oil-producing Muslim societies: foreign aid. As Neumayer () and Werker, Ahmed, and Cohen () document, these countries have been the recipients of significant amounts of foreign aid from Western and Arab donors. Figure . shows that movements in foreign aid per capita are correlated with political stability in Muslim non-oil producers. The figure suggests
between . and . since and has been driven primarily by sustained violence in Algeria and Sudan. Despite some fluctuation – both increases and decreases – over time, the average level of democracy in Muslim oil producers has not changed since . These patterns are not present in non-Muslim non-oil producers (see Figure A in the chapter’s appendix). First, the average incidence of civil war in non-Muslim countries has been significantly lower (e.g., . at its maximum in ). Second, a decline in aid is not associated with an uptick in civil war. As I describe and formalize later in this chapter, this null association is theoretically plausible because non-Muslim countries could have (relatively) more cohesive institutions. The analysis in Chapter provides empirical support.
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
0
0
.05
50
Civil war .1
100 Aid per capita
.15
150
.2
200
Analytical Framework
1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 Year Civil war
Aid per capita
FIGURE . Foreign aid and civil war in Muslim non-oil producers Note: Aid per capita reports the annual average value among Muslim non-oil producers in in US dollars. These figures are based on the author’s calculations using data from the World Bank.
buoyant amounts of foreign aid may “buy” political stability, while their decline may engender conflict. Moreover, notice how a significant decline in foreign aid in the mid-s coincides with the uptick in civil war in Muslim non-oil producers. Figures .–. generate several questions at the heart of this book. For instance, why is there a persistent democratic deficit in Muslim societies relative to non-Muslim societies? What explains its temporal variation, particularly its widening since the s? For example, why did many Muslim societies miss the Third Wave of democratization? Looking across Muslim societies, why have oil producers remained more politically stable compared to non-oil-producing Muslim societies? Why might a country’s rent structure matter? Moreover, why have non-oil-producing Muslim countries experienced greater
In contrast, a reduction in oil prices (and associated oil rents) seems to be unassociated with, and notably has not triggered, heightened violence in Muslim oil producers (see Figure A in the chapter’s appendix). As I argue in Chapter , this may stem from the provision of geopolitical rents (e.g., implied US military support for the government) in Muslim oil producers that has effectively deterred opposition groups from engaging in combat against the government.
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
A Political Economy Approach
political upheaval in the form of civil war but remained relatively nondemocratic? To answer these questions we need to explain () the persistence of nondemocratic political structures in Muslim societies relative to non-Muslim societies and () the possible role of rents in explaining variations in political violence – dictatorship and civil war – across oil and non-oil-producing Muslim societies. In Part II of this book, I provide an answer to the first question by demonstrating how the institutional legacy of Muslim conquest – a process that occurred over a millennia ago – helps explain the persistence of autocracy in many contemporary Muslim societies relative to non-Muslim societies (i.e., the democratic deficit in the cross section of countries). In the remainder of this chapter, I focus on the second question by explaining how rents can affect variations in political violence both across countries and over time, and how this may be related to the quality of preexisting cohesive institutions.
. There is an expansive literature on the causes and consequences of dictatorship and civil war. Rather than summarizing this rich scholarship, my aim in this section is to distill the analytical features that are relevant to understanding how rents can explain variation in political violence. This discussion highlights some of the prominent theoretical and supporting evidence and showcases some of the mixed findings. In doing so, I identify several tensions in this existing scholarship, which serves as a motivation to formulate my own theoretical formulation (model) to reconcile these tensions. Before delving into this, I first define key terms. A Political Violence Both as a cause and consequence, dictatorship is an important feature of political violence. At its core, dictatorship comprises rule by the few over the many, with many scholars conceptualizing it as a “residual” to
As such, Part II provides an account for the cross-sectional variation in contemporary levels of democracy between Muslim and non-Muslim societies. See Blattman and Miguel (), Haggard and Kaufman (), Kalyvas (), and Svolik () for excellent reviews of this scholarship. Throughout this book, I will use the terms dictatorship, nondemocracy, autocracy/ authoritarianism interchangeably. I am not unique in investigating some of these tensions. In evaluating the relationship between oil and democracy, Benjamin Smith and David Waldner () identify the “scholarly dissensus,” highlighting conceptual, theoretical, and empirical challenges in existing scholarship. While civil wars occur most frequently in dictatorships, this is not universal. Democracies too may experience civil conflict, such as the American Civil War. Thus, I argue that dictatorship is a strong “correlate” (rather than source) of civil war.
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
Analytical Framework
democracy. In this tradition, dictatorships and democracies are contrasted on their procedural approach to the selection of leaders (Dahl , Schumpeter ), often by clearly articulating the minimal electoral requirements for democracy (Alvarez et al. , Boix ). Empirically, this generates a dichotomy between dictatorships and democracies (see e.g., Boix, Miller, and Rosato ). In contrast, other formulations identify and seek to categorize the broader institutional variations that feature across dictatorships, such as differences in the constraints faced by the leader, the competitiveness of leader recruitment/selection, and more generally who can participate in the political process. These variations have been measured on a continuum or with discrete categories. The POLITYindex (Marshall and Jaggers ) is an example of the former, while categorizations, such as “competitive” (Levitsky and Way , ), “electoral” (Schedler ), or “semi-authoritarianism” (Ottaway ) characterize the latter. Turning to their operation, as Milan Svolik () cogently distills, all dictatorships must grapple with two central problems: those of authoritarian control and power-sharing. Each problem relates to a distinct group in a polity: those who rule (“the elite”) and those who are ruled (“the masses”). The challenge of balancing against the threats from the masses (majority) excluded from power constitutes the problem of authoritarian control. In practice, of course, dictators rarely control enough resources by themselves to thwart these challenges and therefore rule with a (small) number of allies, such as military officers, prominent party members, or clerics. This can generate a second, distinct political problem when dictators confront challenges from those with whom they share power. This is the problem of authoritarian power-sharing. A dictator who fails to address these problems faces a heightened risk of losing power through elite defection (coups) and violence from opposition groups (civil wars) and less frequently via social revolution. Moreover, these dual challenges are magnified by the dismal setting of authoritarian politics. First, a dictator lacks an independent authority that can enforce agreements with his governing allies and repressive agents, thus
Gandhi (, chapter ) provides a superb discussion of the historical and contemporary use of dictatorship. Many prominent scholars conceptualize dictatorship as a residual category of democracy. For example, Boix (), Cheibub, Gandhi, and Vreeland (), Przeworski et al. (), and Svolik () define dictatorship as an independent country that fails to satisfy at least one of the following two criteria for democracy: () free and competitive legislative elections and () an executive who is elected either directly in free and competitive presidential elections or indirectly by a legislature in parliamentary systems. A related formulation characterizes who rules in dictatorships, such as the military, a single party, or a personalistic ruler. See, for example, Geddes () and Geddes, Wright, and Frantz (). For example, over the period –, only percent of dictators exited office due to popular uprisings (Svolik , ).
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
A Political Economy Approach
elevating the risk of elite defection. Second, violence is ever-present and ultimate arbiter of conflicts. Thus, Svolik (, ) proclaims the “key features of authoritarianism – including institutions, policies, as well as the survival of leaders and regimes – are shaped by the twin problems of power-sharing and control against the backdrop of the dismal conditions under which authoritarian politics takes place.” The dismal setting that dictators face raise their societies prospects for experiencing civil war; a finding, well documented in the empirical civil war literature. Conceptually, civil war entails two-sided violence that pits the incumbent (government) in armed combat against one (or more) opposition groups. In explaining their determinants, many scholars differentiate between “grievances” and “opportunities” as important sources of civil war (e.g., Blattman and Miguel , Collier and Hoeffler , Kalyvas ). On the former, since dictatorships frequently exclude certain groups from power, grievances can be an important driver for conflict. For example, regimes that restrict political participation frequently pursue redistributive policies that perpetuate inequalities and perceptions of deprivation that can mobilize rebellion (Besley and Persson , Cederman, Weidmann, and Gleditsch , Gurr ). Of course, the mere existence of grievances does not imply civil war. As Stathis Kalyvas (, ) observes: “Obviously, the problems with grievances as a determinant of civil war onset is that they seem to be much more prevalent than civil war and that they are very hard to measure directly. Ultimately, it is difficult to escape the conjecture that one has to look for combinations of demand for, and supply of, rebellion (or intention and opportunity).” Greed-based explanations posit that economic conditions comprise an important driver of civil war (Collier and Hoeffler ). One channel posits that economic decline, for example through depressed wages, can both lower the opportunity cost for rebels to mount an insurgency and weaken an incumbent’s capacity to fight back (Miguel, Satyanath, and Sergenti , Powell ). Another channel identifies the motive for the incumbent and opposition to control state resources (“the state prize”), such as foreign aid and a variety of natural resources (Besley and Persson a, b, Le Billon ). For instance, revenues derived from the state’s control of natural resources, such as oil, natural gas, diamonds, and timber, can incite violence from all sides. Revenues from natural resources can finance an incumbent’s repressive
The literature on civil war identifies different types of groups based on their: ethnic, linguistic, and religious identification/fragmentation, geographic location (e.g., urban versus rural, near natural resources), social/economic class and modes of production (i.e., levels of vertical and horizontal inequality in societies), and ties to foreign powers. Differences in the composition, size, and/or intensity of grievances among individuals in these groups can motivate armed violence.
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
Analytical Framework
capacity, such as the size of its army. Similarly, opposition groups may seek to control the natural resources to finance their rebellion. Empirically, several studies document a higher likelihood, intensity, and duration of civil war in countries with natural resources (e.g., Le Billon , Lei and Michaels , Synder ). Greed-based explanations, of course, do not exclude the importance of political grievances. In practice, the presence of both motives – grievances and greed – heighten the incidence of civil war. Consequently, dictatorships with valuable state prizes (e.g., oil revenues, foreign aid receipts) often face an elevated risk of civil war. The preceding overview of political violence highlight two important points. First, dictatorships entail the rule of the few over the many, in which the regime confronts two problems – of control and power-sharing – in a dismal setting of ever-present violence. Second, dictatorships especially those with valuable state prizes, also face an increased risk of two-sided violence in the form of civil war. B Rents and Political Regimes Rents and Governance Given the political challenges described above, how can incumbents better ensure the longevity of their power? Scholars across multiple disciplines have identified the essential role of buoyant public finances in helping governments (rulers) remain in power. In describing the primacy of revenues for all governments, Margaret Levi (, ) describes how they can benefit a ruler personally and/or his subjects: Rulers may use the funds to line their own pockets or to promote their personal power. They may use the funds to support social or personal ends. They have ideological ends they wish to promote. They may be altruistic. Randomness of ends characterizes rational choices in general, and rulers operate with a wide range of them. It’s by the means of the state and its revenues that rulers achieve their personal and social ends.
As Levi subsequently elucidates, the importance of revenues thus incentivizes rulers to be “predatory” and extract as much revenue as possible. In doing so, however, rulers face constraints. For example, maximizing state revenue involves the cost of extraction or a trade-off between extraction costs (e.g., deadweight loss associated with taxation) and economic development (e.g., growth). Minimizing the transaction costs associated with taxation, therefore, can compel rulers to derive their revenue from alternate sources. Scholars frequently refer to these alternate sources of nontax government revenue as unearned income or rents (e.g., Ahmed , , Morrison , Smith ). Whereas tax revenue entails the extraction of income from
The incumbent could also transfer funds to regime supporters and/or “buy off” rebel groups. Since this sub-section builds on Levi’s () seminal work, I adopt her use of the term “rulers” to refer to governments/regimes.
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
A Political Economy Approach
economic agents (e.g., wage earners, firms) and associated transactions (e.g., value-added taxes, trade tariffs), rents may entail revenues derived through the state’s exploitation of natural resources (e.g., oil, minerals) and external transfers (e.g., economic and military assistance from foreign governments). Rents can include revenues from oil because it can be generated through the operation of state-owned oil enterprises rather than through taxes. Rents also include income from abroad (e.g., sovereign borrowing, foreign aid) and intergovernmental grants at the subnational level. For many governments, rents are nontrivial and account for a large share of their expenditures, such as in Nicaragua ( percent), Bolivia ( percent), Republic of Congo ( percent), Burundi ( percent), Romania ( percent), Syria ( percent), Iran ( percent), Nepal ( percent), Pakistan ( percent), and Japan ( percent). Rents and Democracy The generation and composition of government finances from tax and nontax sources has political implications. For instance, in premodern Europe, the need to finance interstate wars compelled European sovereigns to develop tax collection systems. According to Charles Tilly (), taxation required bargaining between revenue-maximizing leaders and their citizens through which rulers relinquished some influence over state policies in exchange for taxes. With more warfare over time and thus a greater reliance on tax revenue, rulers tended to become more accountable to their populations. This bargaining mechanism thus generates a pathway from taxation to more democratic governance; and by extension, governments that generate their revenue from nontax sources are less likely to feature democratic characteristics. This framework linking taxation to democracy has been applied to explain modern political development in various regions, including Africa (Guyer ), Asia (Bernstein and Lu ), the Middle East (Crystal ), Europe and North America (Scheve and Stasavage ), and the former Soviet Union (Gehlbach ).
To be clear, tax revenue entails the extraction of income from individuals (e.g., wages, salaries), firms, land and capital owners, and associated economic transactions via consumption (e.g., value-added taxes) and international trade (e.g., tariffs), among others. Note, my definition of rents excludes remittances (which accrue directly to individuals). However, through an expenditure switching strategy, governments may be able to incorporate remittance income into their revenue base, thus comprising a type of nontax income (Ahmed , ). State-owned enterprises control percent of the world’s oil production and percent of its reserves (Ivanhoe , Tordo et al. ). Examples include Saudi Aramco, China National Petroleum Corporation, Nigerian National Petroleum Corporation, and Petroleos de Venezuela. Based on calculations from Morrison (, ) using data averaged over the period, –. Bates and Lien () formalize this bargaining process and identify two important considerations. First, greater asset mobility may weaken the sovereign’s ability to tax. If the tax rate is too high, citizens may shift their assets abroad (an argument further explored by Boix ). Second, the relationship between taxation and representation may be conditional on the amount of spending that citizens (expect to) receive in return.
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
Analytical Framework
While taxation tends to be positively associated with democracy, it frequently generates political contestation within societies. Indeed, redistributive conflicts – typically regarding the level of taxation and associated transfers from the “elites” to the “masses” – are critical elements in prominent theories of political instability and regime transitions (e.g., Acemoglu and Robinson , Boix , Moore ). According to Kevin Morrison (), two tensions underlie this redistributive conflict: the need for governments to spend while minimizing their tax burden. The availability of nontax revenue (rents) can assuage these tensions: “Nontax revenues enable governments to reduce their taxation levels while increasing government spending, thereby increasing the stability of democracies, dictatorships, and their leaders” (Morrison , ). To understand the logic, consider how rents can strengthen democracy in Daron Acemoglu and James Robinson’s () theory of democratic transitions. They posit that elites democratize as means to credibly commit to future redistribution rather than suffer revolution in which the masses expropriate the assets held by the elites. In this context, the availability of rents can help lower a regime’s tax burden, thus relieving the costs of redistribution for elites and facilitating democracy. Thad Dunning () applies this logic in articulating how oil rents can foster “crude democracy”: By reducing the extent to which democratic majorities redistribute private income or wealth away from the rich elites through taxation, land reform, or other measures, resource rents can reduce the economic cost of democracy to elites. . . . This occurs because both resource rents may reduce the current redistributive costs of democracy and because resources may endogenously lower the extent to which future democratic regimes may redistribute private wealth from elites when there is not a coup threat; resource rents thus shape elite expectations about redistribution under future democratic regimes.
Rents and dictatorship. While rents may lower the costs of democratization in theory, there is a large body of research documenting how rents tend to entrench dictatorship. One strand of this literature identifies the role rents can play in introducing authoritarianism in existing democracies. For instance, rents can engender a strong “incumbent advantage” that can reduce political competition and introduce creeping authoritarianism (Wantchekon ). In clientelistic settings, one pathway is via targeted redistribution: The incumbent can transfer rents directly to their supporters (e.g., co-ethnics) and exclude other groups from the direct benefits of rents (Jablonski ). Consequently,
This logic also applies to Boix’s () theory of democratization. Relatedly, Menaldo () argues whether natural resources enhance democracy is conditional on preexisting institutional quality. Rents in countries with good/democratic institutions do not harm democratic governance (e.g., Norway). In support of this conjecture, Dunning shows that natural resource rents, particularly from oil, supported “crude democracy” in several Latin American countries, most notably Venezuela.
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
A Political Economy Approach
the recipients of rents may be more likely to turn out and vote for the incumbent. Another strand of this literature discusses the channels through which rents strengthen dictatorship. Much of this scholarship originates from scholars of the Middle East investigating whether the prevalence of rents fostered authoritarian governance in that region (Beblawi , Chaudhry , Madhavy ). The central argument in this “rentier state” literature posits that governments funded by external rents are relieved from the need to raise taxes, which makes them less accountable to their citizens. As a result, these governments are less likely to deploy the rents in ways that promote economic development. Focusing on the political ramifications, scholars have refined the theoretical channels through which rents can hinder democracy. Michael Ross (, –) identifies and tests three broad “effects” through which rents can foster dictatorship: a “rentier effect,” through which rents “relieve the social pressures that might lead to demands for greater accountability” from the broader population; a “repression effect,” through which rents may empower “governments to spend more on internal security and so block the population’s democratic aspirations”; and a diminution of a “modernization effect,” through which “resource-led growth does not lead to higher education levels and greater occupational specialization. . . [and thus] should fail to bring about democracy.” As Ross concludes, these effects are not mutually exclusive as “all three explanations, or any combination of them, may be simultaneously valid.” In delineating these effects, Ross elucidates how rents can help dictators mitigate their dual challenges of authoritarian control and power-sharing, for instance, through his decomposition of the rentier effect into three distinct mechanisms: a taxation, spending, and group formation effect. Consistent with Tilly () and my earlier discussion, the taxation effect argues that rents reduce a government’s need to tax their population for revenues and consequently leads their populations to demand less political accountability. In
Mahdavy (, ) first advanced the contemporary concept of the rentier state as a state that receives substantial rents from foreign individuals, concerns, or governments. Beblawi (, ) developed a more precise conceptualization of a rentier state. In such a state, () the rents are paid by foreign actors, () these rents accrue directly to the state, () only a few domestic actors are engaged in the generation of this rent (wealth), and () the majority are only involved in the distribution or utilization of the wealth. Chaudhry’s () seminal work broadens the theoretical and empirical scope, arguing how oil wealth affects governance both in oil producers and labor-exporting countries in the Middle East. For the latter, Chaudhry argues migration to oil producers from nonproducers and associated remittance inflows can affect governance in laborexporting countries (Ahmed provides causal evidence in support of this argument). For more recent treatments on the relationship between rents – mainly oil – and democracy in the Middle East, see Cammett et al. () and Smith and Waldner (, chapter ). While Ross discusses these effects with regards to oil wealth, the insights can be generalized to other rents.
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
Analytical Framework
combination this can enhance the dictator’s control over society. The spending effect posits that rents allow governments to spend more on patronage, which in turn dampens latent pressures for democratization. These transfers may reward a small and critical group of regime supporters (e.g., military commanders, clerics, co-ethnic leaders) and/or fund social services that benefit the broader population. The former, which supplies targeted private benefits, can help dictators reduce the problem of authoritarian power-sharing by retaining the loyalty of key regime supporters. In contrast, if a government uses its rents to fund generous social benefits, it may be an effective instrument of authoritarian control. For example, Entelis () describes how the Saudi Arabian government uses its oil wealth to fund a generous welfare state that reduces pressures for democratization. Relatedly, the group formation effect argues that when rents are buoyant, the government can use its largesse to prevent the formation of social groups that are independent from the state and thus, be inclined to demand greater political representation. Heterodox Views Some scholars point to the mixed evidence between rents – particularly, oil wealth – and democracy/dictatorship to suggest that rents may have no independent (causal) effect on democracy or dictatorship. For example, examining data over two centuries, Stephen Haber and Victor Menaldo () find a tenuous relationship between oil wealth and dictatorship and argue that there is no “oil curse,” at least over the long run; and in some instances, oil rents may comprise a political “blessing.” In a similar vein, Benjamin Smith and David Waldner (, ) highlight a “scholarly dissensus” in existing scholarship linking oil to dictatorship. They maintain:
Moreover, the utility of this strategy is magnified in increasingly smaller power-sharing coalitions. This is because the per capita transfers can be higher to coalition members. Smith (, ) summarizes this logic: When coalition size is small, leaders can greatly reward their essential supporters by supplying private goods [targeted benefits]. However, as coalition size becomes large, private goods become more thinly spread, and leaders can more efficiently reward their supporters through the provision of public goods.” Bueno de Mesquita et al. () and Lake and Baum () provide empirical evidence of this trade-off. Ahmed (, chapter ) provides a numerical example. As discussed shortly, providing social benefits for the masses (e.g., public goods) may not be optimal for dictators. A more politically beneficial strategy for a dictator is to fund repression and targeted transfers (e.g., Smith ). A variant of the group formation effect builds on Putnam’s () work on social capital. Here rents can stifle the formation of civic groups that lie above the family but below the state. These civic groups are often important drivers for greater democratic accountability. Andersen and Ross () rebut this conclusion, finding that oil wealth may be a blessing in (some instances) prior to . However, after there is stronger evidence pointing to an oil curse. Smith and Waldner () reach this conclusion by pointing to several conceptual (measurement), empirical, and theoretical issues in extant scholarship.
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
A Political Economy Approach
That when one looks closely at specific claims made in the literature, and the specific methodological warrants of those claims, the conclusion that a political resource curse exists appears less unassailable than would be the case when all the diverse findings are pooled together without sufficiently discriminating appraisal.
Smith and Waldner’s synopsis engages with the empirical evidence linking oil to politics across time and space (regions). Their analysis reveals that a political resource curse exists primarily in the Middle East and North Africa, which they attribute to factors related to institutions and coalitional politics. To the extent that the institutional legacy of Muslim conquest may explain the configuration of institutions and coalitions (e.g., relationship between the ruler, the military, and religious clergy) in many contemporary Muslim societies (as explored in Part II of this book), Smith and Waldner’s inferences may be compatible with those advanced in this book. C Conditional Theories of Rents and Political (In)stability While rents can help dictators mitigate their dual challenges of authoritarian control and power-sharing, the specific policies they permit (e.g., kleptocracy, repression) may incite political instability, such as revolutions and civil war. As noted earlier, dictatorship and its accompanying politics, often creates a climate conducive for violence. With this environment in mind, existing scholarship identifies several pathways linking rents and regime type to political instability. Variation in Institutions One pathway evaluates the effects of rents on political stability as institutions vary. In the “perils of unearned income,” Alastair Smith () formalizes these dynamics within the selectorate model of politics (Bueno de Mesquita et al. ). A critical parameter in the selectorate model is the size of the “winning coalition” (W) required for a leader to remain in power. Small W systems characterize dictatorships, where the leader’s optimal strategy is to
For the latter, they study the relationship between oil and dictatorship (and the prospects of democratic transitions) in four regions – Latin America, the Middle East, Africa, and Southeast Asia. Here I focus on civil war, but rents may also incite war between states. See for example, Caselli, Morelli, and Rohner (), and Colgan (). A related explanation focuses on the strength of opposition forces and the timing of oil wealth. Focusing on oil producers, Benjamin Smith () posits that variation in the strength of opposition forces (to the government) in late industrializers can affect how rulers manage their oil wealth with divergent political implications when oil wealth declines. Where opposition forces are strong in late industrializers (e.g., Indonesia), the ruler undertakes policies (e.g., more effective taxation of non-oil income) that reduce its probability of losing power when oil income declines. In contrast, where opposition forces are weak in late industrializers (e.g., Iran), the ruler is prone to adopting policies (e.g., patronage to selected/narrow groups) that increase its probability of losing power when oil income declines.
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
Analytical Framework
provide targeted benefits to coalition members and to reduce the provision of public goods (e.g., welfare transfers, rule of law). One of Smith’s key findings identifies a non-monotonic relationship, where the effect of rents on revolutionary threats varies with the size of W. Referring to rents as “free resources,” Smith’s () comparative statics reveal that “large coalition systems are relatively more efficient than small coalitions at converting free resources into societal benefits and economic activity because free resources fund public goods rather than corruption, graft, and kleptocracy. From the perspective of citizens outside of the winning coalition, if the level of free resources rises, then the desirability of a large coalition increases relative to a small coalition system . . . This encourages the citizens to support revolution” (). Of course, if the costs of rebellion are prohibitively high, the availability of rents may not necessarily induce revolution. As Smith (, ) observes, these costs can be large in small W systems: The citizens’ decision to rebel is affected by two factors: the difference between their welfare under the current institutions relative to the welfare they expect under postrevolutionary institutions and the probability of the revolution succeeding. When coalition size is small, leaders provide few public goods. This leaves the citizens extremely dissatisfied with their lot relative to what they could obtain under postrevolutionary institutions. This gives them a great desire to rebel. However, the citizens also factor the probability of success into their decision to rebel. When the coalition is small, leaders provide few of the public goods which help revolutionaries coordinate and organize. As a result, when W is small, citizens choose not to rebel; not because they don’t want change, but they are unlikely to succeed.
The confluence of these comparative statics generates an inverted U-shaped relationship between rents, coalition size, and revolutionary threats. In small W and large W systems, rents are less likely to induce revolution. However, in the intermediate range of coalition size, greater rents incite revolution. Of course, as Smith () also shows, the availability of free resources can empower governments to finance their repressive capacity to crush rebellion.43 In conjunction with the U-shaped relationship, this policy response implies that relatively robust dictatorships (e.g., monarchies or personalist dictatorships with smaller W) can effectively deter revolutionary threats induced by rents. In short, rents can monotonically increase political stability in dictatorships. In subsequent research, Bruce Bueno de Mesquita and Alastair Smith () provide empirical corroboration for this prediction.
Figure in Smith () demonstrates this. In Smith’s model, governments can also contract their supply of public goods which can make it more difficult for individuals and groups to collectively organize. Note, the non-monotonic effect formalized by Smith () and empirically supported by Bueno de Mesquita and Smith () varies by coalition size. It does not vary by changes in the level of rents. This distinction is important as the model I introduce in Section . generates nonmonotonic effects that vary by rents and coalition size. I conceptualize the latter as cohesive institutions.
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
A Political Economy Approach
Variation in Rents Another pathway evaluates the effect of institutions on political stability as rents vary. Here, the existing literature identifies two divergent effects associated with changing levels of rents on civil war. The first is the converse of the politically stabilizing role of rents discussed in the previous section. Here a reduction in rents weakens the government’s capacity to thwart insurgency and/or funds to buy off regime supporters and rivals. Relatedly, rents can mitigate civil conflict by providing a “cushion” to government spending during slumps in the “private” economy. One such example is a well-timed foreign aid package (or opportune increase in commodity prices) that can provide governments with the necessary financial windfall to make rebellion less attractive for opposition groups (Savun and Tirone ). Of course, if these rents disappear the prospect for civil war may be elevated. Citing the commitment problem articulated by Robert Powell (), Richard Nielsen et al. () argue that a severe drop in government revenues (e.g., stemming from a rapid decline in foreign aid) can weaken the government’s capacity to effectively buy off rebel groups and contribute to “inefficient conflict.” They leverage matching techniques to provide cross-national corroboration, documenting how negative aid shocks accelerate the likelihood of civil war. The other path links rents to heightened conflict. This account posits that rents increase the size of the government “pie,” and if there are multiple groups dividing the pie, rent-seeking can contribute to increased fighting over it (Grossman ). Several studies formalize this central idea, identifying the critical role of preexisting political institutions in shaping conflictual behavior. For example, Aaron Tornell and Philip Lane () formalize a “voracity effect” by which an increase in a government’s fiscal resources (e.g., due to a natural resource boom, foreign aid windfall) incites groups within societies with “weak” institutions to compete violently for and squander the windfall and reduce economic growth. Relatedly, Jacob Svensson () develops a repeated game with stochastic shocks in unearned income, where the increase in rent-seeking arises due to coordination failure across interest groups. This coordination failure is often heightened in ethnically diverse societies. For example, Roland Hodler () constructs a model where a surge in rents fosters Cournot-type competition in the form of group fighting in ethnically fragmented societies with a corresponding elevated risk of conflict and lower growth. In these models, political institutions are unable to get incumbent governments to credibly commit to allocating state resources to stave off violence. As I discuss below, this commitment problem is characteristic of societies with less cohesive political institutions. More generally, the conditional effects of rents on political stability (associated with the models above) fall under a broader theory of an “institutions curse” in which rents received in societies with less cohesive institutions are
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
Analytical Framework
prone to pernicious political economy (Menaldo ). As discussed above, in some of these accounts a reduction in rents leads to civil war, while in others higher rents do. Moreover, the likelihood of civil war in the presence of an increase or decrease in rents often depends on the quality of “cohesive political institutions.”
D Cohesive Institutions But what are cohesive political institutions? Cohesive institutions describe how an incumbent government “shares” state resources within its territory (Besley and Persson a). In general, societies with more cohesive institutions allocate a more equitable share of state resources to groups that are out of power (i.e., opposition groups). As Besley and Persson (b) note, real-world examples of societies with more cohesive institutions are those with “more minority protection through a system of constitutional checks and balances, through a parliamentary form of government, or through a proportional electoral system” (). This categorization clearly situates democracies as being more cohesive and dictatorships as being less cohesive. In formal models, the degree of cohesive institutions can be captured with a bounded (continuous) parameter; for example, θ 2 ½0, 1 where a higher value of θ corresponds to more cohesive institutions. Thus, if θ ¼ 1=2, then state resources are transferred equally between supporters of the incumbent and the opposition. In contrast, if θ ¼ 0 the incumbent allocates all the state’s resources to its supporters and none to supporters of the opposition. Empirically, scholars have documented important determinants/correlates of cohesive institutions. For instance, governments in more ethnically, religious, and/or linguistically diverse societies tend to supply higher amounts (shares) of public goods to individuals that share the same group affiliation/identity as the incumbent (Alesina, Baqir, and Easterly , Baldwin and Huber , Habyarimana et al. ). Thus, many “fractionalized” societies exhibit less cohesive political institutions.
Although he does not use the terminology of cohesive institutions, Victor Menaldo () provides an excellent account of an institutions curse associated with natural resource rents. Faisal Z. Ahmed () develops a parallel account of an institutions curse with various forms of international capital (e.g., remittances, foreign aid, and investment). Relatedly, correlates of domestic political institutions can also generate a broader “conditional resource curse.” This may arise from other features of a country’s domestic political economy, including the strength of opposition forces (Smith ), the level of income inequality (Dunning ), ownership structure of natural resources (Luong and Weinthal ), the quality of electoral institutions (Mahdavi ), and degree of economic integration (Kurtz and Brooks ). An alternative categorization of cohesive institutions can situate it along Acemoglu and Robinson’s () concept of “inclusive” institutions. As these scholars document, societies with more inclusive institutions share power and state resources (public finances) across various societal groups.
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
A Political Economy Approach
Subsidies and transfers (% Expenditures) 20 40 60 80
SPN GMY CZR SWZ NTH UKR VEN MLD
BLR
0
ITA CAN POL SLV SER SWD CHL BUL KZKRUS MON ROK USA URU ALB SLO BEL BRA AZE MAC CRO UGA HUN EST GRG UKG BOS FIN RUM MAL IND NOR ALG SAF LAT EQG GRC AUS FRN IRE TUN ARM INS TAJ NEW JPN COLTUR AUL BOT KUW BOL IRN TRIISR ETI BNG NIC MAS BUI PAN POR MOR PER DEN PAR KYR SRI NIG PNG CYP COS LES GUALEB SEN GHA ZIM FJI CEN SIN SYR CAM EGYANG QAT SWA JOR TOG LBR ZAM THI PHI SAL HON PAK RWA NAM GAM SIEBEN MAG KEN MLI BAH LAO BFO OMA BHU AFGIVO ZAI JAM
0
.2
.4 .6 POLITY (normalized)
.8
1
FIGURE . Democracy and public goods Note: Country average based on author’s calculations. Data on subsidies and transfers is from the World Bank ().
As noted above, another important correlate of cohesive institutions is a society’s quality of democracy: Societies with more cohesive institutions tend to exhibit a higher quality of democratic governance and institutions. This association is unsurprising as many theories predict that governments in democracies will allocate a greater share of their resources to public goods, such as expenditures on welfare payments and subsidies to the population (e.g., Acemoglu and Robinson , Bueno de Mesquita et al. , Lake and Baum , Olson ). Given this incentive, democratic governments correspondingly tend to exert greater “tax effort”: They derive greater share of their revenues from taxation (Besley and Persson a). Using data from the World Bank, Figures . and . corroborate these conjectures for a global cross section of countries. Looking first at government spending, Figure . shows that some of the least democratic countries (e.g., Bhutan, Zaire) allocate less than percent of their expenditures on social transfers. In sharp contrast, the most democratic countries, on average, spend
Greater tax effort is also exerted in resource-abundant democracies, such as Norway and the United States.
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
Analytical Framework
60
AUL JPN NEW INS
USA CAN
SAF
Tax income (% revenue) 20 40
QAT ANG
TRI
MAL
ZIM
PNG
TUR
UKG DEN BEL IRE VEN MEX LBR FJI NAM ISR ZAM ITA SYR SPN KEN JAM SIN VIE ROK BOT SLO NTH FIN SWA KZK THI BRA IND AUS GUA CHN NIG MON COL LAO CYP NOR MORALG HON AZE SAL POR POL GHA EQG TOG RUM FRN SEN IVO LES PANPERCHL ETI BEN KYR HUN ZAI MYA SIE EGY RWA CZR GMY TUN BUL ARM GRC SWD BFO PAK NIC EST SWZ MAS BUIUGA SLV BNG LEB SRI COS UKR MLI MAC ARG LAT IRN OMA CEN MAG NEP SER JOR BLR GAM PAR CRO GRG KUW BHU URU ALBCAM AFG RUS BOL BAH TAJ MLD BOS
0
PHI
0
.2
.4 .6 POLITY (normalized)
.8
1
FIGURE . Democracies exert greater tax effort Note: Country average based on author’s calculations. Data on tax income is from the World Bank ().
over percent of their budgets on social transfers. This finding is not driven by greater per capita wealth in more democratic societies. The positive relationship between democracy and social transfers remains statistically robust in a regression that controls for log GDP per capita, as well as with other common measures of democracy, such as constraints on the executive (Acemoglu and Johnson , Gledistch and Ward ). Relatedly, Figure . shows that governments in more democratic polities exert greater tax effort. In this figure, the positive association between the share of government revenue collected from income taxes and a higher normalized POLITY score is consistent with predictions from several prominent political economy models of public finance based on the median voter theorem (Meltzer and Richards ), electoral incentives (Lizzeri and Persico ), credible commitments to redistribution (Acemoglu and Robinson ), and selectorate theory (Bueno de Mesquita et al. ). Taken together, the purpose of the Figures . and . is to simply illustrate how variation in democracy captures the core feature of cohesive institutions in the contemporary era: the extent to which state resources are equitably
Calculated percentage based on average across countries with a normalized polity score greater than or equal to ..
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
A Political Economy Approach
distributed across individuals (groups) that support and oppose the incumbent. That said, it is worth noting that cohesive institutions may also be correlated with (and endogenous to) many historical and contemporary conditions within a given society, such as its climate/geography (), precolonial state capacity (Hariri ), “accumulated democratic experience” (Gerring et al. ), factor endowments (Acemoglu and Robinson , Engerman and Sokoloff , Menaldo ), colonial legacy (Olsson ), and structure of the international system (Boix ). And, as the cited scholars demonstrate, these conditions are in turn highly correlated with democracy in the contemporary period. The covariance of these conditions implies that isolating their separate effects poses a critical empirical challenge; a task that Part II of this book addresses in greater detail. As a preview, this is achieved through the research designs (e.g., using panel data, instrumental variables) and choice of control variables. For example, since many correlates of cohesive institutions are country-specific and timeinvariant (e.g., precolonial state capacity and history, colonial legacy, climate, and geography), in econometric analysis with panel data, they can be accounted with country fixed effects. Similarly, controlling for year (or period) fixed effects can capture the effects of important temporal factors that affect all societies, such as the structure of the international system.
E Theoretical Puzzle The discussion in the preceding sections generates an important theoretical puzzle stemming from three core quandaries: () Rents can be a source of political stability, especially as a means for dictatorships to increase their repressive capacity and provide transfers to their key supporters. () Some theories, however, predict that higher levels of rents heighten the propensity for civil war, while others predict the opposite: A drop in rents makes civil war more likely. () These politically destabilizing effects tend to be elevated in societies with less cohesive societies. Surprisingly, these divergent paths have yet to be reconciled in a unified analytical framework. Thus, I seek to answer the following theoretical puzzle: Under what conditions should an increase or decrease in rents lead to political violence? In collaborative work with Daniel Schwab and Eric Werker, we advance a formal model to reconcile this tension in existing theoretical work and also
For example, the use of instrumental variables can help mitigate concerns with omitted variables (Angrist and Pischke ).
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
Analytical Framework
provide a peaceful pathway for democratization (Ahmed, Schwab, and Werker ). In particular, the model derives plausible equilibrium conditions under which variation in rents and the quality of cohesive institutions can lead to differential trajectories of democracy (peace), dictatorship (repression), and/or civil war in a unified theoretical framework. Building on the now canonical model of state capacity and political violence developed by Timothy Besley and Torsten Persson (, , a,b), the model is driven by competition for political power between two groups within a country – an incumbent and an opposition – where the winner distributes state resources disproportionately to its own members depending on the cohesiveness of institutions. Consistent with the discussion in the previous section, in the model more cohesive institutions correspond to a more equitable distribution of state resources across society by the incumbent. The model assumes that it is not possible to credibly commit in advance to a more equitable redistribution, so both groups have an incentive to use violence to increase their likelihood of gaining or holding power. In doing so, this assumption captures an important dimension of politics in democracies and dictatorships (Svolik ). Like Besley and Persson’s formulation, my model generates three outcomes: peace/ democracy (no violence), repression (one-sided violence inflicted by the incumbent), and civil war (two-sided violence involving the opposition fighting the incumbent). While the model identifies nontax income (rents) as an important source of these political outcomes, I am careful to capture other important features that can affect politics, such as per capita income (through wages). The model extends the framework from Besley and Persson to allow for the possibility that the incumbent can deter attacks by the opposition when government resources are sufficiently high. The model is essentially a one-shot sequential game, and when rents are moderate, an incumbent will respond to an insurgency with only moderate violence, which encourages an attack by the insurgents. When rents dwarf the costs of fielding a large army, the incumbent would be willing to spend enough to squash an insurgency, and the insurgents are deterred from attacking. To preview my predictions, whereas existing models predict a hierarchy of violence in which higher levels of rents monotonically increase the likelihood of political stability (e.g., Smith ) or conflict in dictatorships (e.g., Besley and Persson b), I provide conditions for differential and non-monotonic
In the analysis to follow, references to “my model” (or using “I”) reflect collaborative work with Daniel Schwab and Eric Werker. In extensions to the model (e.g., where players may incorporate “common knowledge” acquired via repeated interactions), there may be Pareto dominant equilibria. The degree of cohesion can be shaped by several factors, including historical legacy (e.g., colonial history, premodern state capacity), factor endowments (e.g., land inequality, natural resources), and ethnolinguistic and religious fractionalization. The empirical analysis in subsequent chapters will discuss and control for these factors.
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
A Political Economy Approach
effects. My model generates two key predictions. First, consistent with many existing accounts I show that higher levels of rents raise an autocrat’s repressive capacity and ability to maintain authoritarian rule. Second, upon the reduction of these rents (to medium or low levels), these regimes are susceptible to challenges from opposition groups. In particular, the flexibility of the model allows for two paths from a reduction in the level of rents: a heightened incidence of civil war or peaceful transitions to democracy. The particular outcome hinges on the preexisting cohesion of political institutions. Violence is more likely in societies where preexisting institutions are less cohesive, while a peaceful transition to democracy is more likely in societies whose preexisting institutions are more cohesive. To the best of my knowledge, this differential and non-monotonic effect is novel in the theoretical literature on rents and political stability. For interested readers, I present this model in the next section. Section . distills the model’s main empirical implications and previews the research designs to evaluate them.
. In this section, I advance a parsimonious game theoretical model to illustrate how () higher levels of rents can entrench dictatorship and () the conditions under which their decline can engender civil conflict or democracy. For the latter, I show that countries with less cohesive institutions are prone to conflict when rents decline, while those with more cohesive institutions can transition peacefully to democracy. A Basic Setup As already noted, my model closely follows a series of theoretical formulations of state capacity and political violence developed by Besley and Persson (, , a,b). In any society, there are two groups ðI, PÞ, each with a population normalized to . Each member of the population earns a wage ðwÞ which is set to . The incumbent (denoted, I) starts the period in power, while the other group ðPÞ is in the opposition. Ending the period in power allows the incumbent to favor its own group when distributing state resources ðRÞ, which is an exogenous parameter. State resources ðRÞ can be thought of as nontax government income (rents) that may come from domestic sources (e.g., income from natural resources,
For instance, Bueno de Mesquita and Smith (), Dunning (), Menaldo (), Morrison (), and Ross (, ) do not generate these predictions. Since the model’s comparative statics will focus on analyzing variation in cohesive institutions and rents setting wages to simplifies the subsequent mathematical derivations. That said, allowing wages to vary would not affect the comparative statics analysis of institutions and rents.
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
Analytical Framework
such as oil) and from abroad (e.g., foreign aid, sovereign borrowing). In certain settings, state resources could also include external security guarantees – “geopolitical rents” – available to the incumbent, such as the presence of foreign troops in their territory and diplomatic cover to pursue repressive domestic policies (e.g., imprison political opponents). Unlike material resources such as revenues from natural resources or from foreign aid, geopolitical rents may be tied exclusively to safeguarding the incumbent and are not shared with the opposition. For example, the presence of US troops in certain Persian Gulf countries may serve to better ensure US foreign policy preferences (e.g., a stable supply of oil, peaceful relations with Israel) and tacitly help support the incumbent regime’s continued rule. If that incumbent loses power to the opposition, the US may withdraw military support; in which case, geopolitical rents would be unavailable to the opposition. Both the incumbent and the opposition have access to a violence technology: an army of size A 2 ð0, 1Þ that costs wA to raise. I assume that there are two sizes of armies available for the incumbent, L and H (for low and high), and one for the opposition, which is the same as the incumbent’s low level. Limiting army sizes to discrete options is primarily for tractability, but it does also have some real-world applicability; for example, a government may decide between allowing rebels to hold zero or one city, or they may choose whether or not to use its air force. Allowing the incumbent access to violence levels , L, and H means that there is a possibility that an incumbent’s response to violence is dependent on the level of rents. I assume that the opposition first chooses whether to attack or not, which the incumbent observes, and then the incumbent chooses its own level of violence. If the investment needed for the incumbent to field a large army is sufficiently close to its investment for a small army, then the opposition will never attack. The reason is that the incumbent would employ a large army against an attack because the expenditure would be small relative to the gain
I abstract from the important question of intragroup inequality and assume that all members of a group receive an equal share of the group’s resources. I provide evidence consistent with this conjecture in Chapter . Egypt is an example in a nonoil-producing exporter. During the Arab Spring, the US government effectively withdrew support from the democratically elected Morsi government after the demise of the US-backed Mubarak regime. Upon the resumption of military dictatorship, the USA resumed its support of the regime. This is an important modification to Besley and Persson (): In that paper, the incumbent can choose to attack or not, but has no flexibility on army size if it does attack. In that paper and in this chapter, assumptions on the parameters and the conflict function ensure that in equilibrium the incumbent always responds violently to an insurgency. In Besley and Persson (a,b), the choice of violence is continuous, so the incumbent’s choice of violence can depend on unearned income. However, other assumptions in that paper rule out the possibility that opposition violence increases as unearned income decreases. This timing is the same as Besley and Persson () but differs from Besley and Persson (b) where the two groups attack simultaneously.
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
A Political Economy Approach
from retaining power, and so the opposition would be deterred. As I show in the model’s proofs (in the chapter’s Appendix), the relevant threshold is that there is never two-sided violence when H < 3L. Since opposition groups do sometimes mount an insurgency, I assume H > 3L. This ensures that two-sided violence does occur for some values of R. Power changes hands stochastically according to the conflict function γ AP , AI 2 ½0, 1, where either side can improve their chances by investing in violence. The following specifies the functional form as: 8 I P > >0 A > A P I < I P γ A ,A ¼ 1 A < A > > : 1 AI ¼ AP 2 This functional form means the group choosing the higher level of violence wins power with certainty, and if they choose the same level, then power changes hands with probability 12. At the end of the period, the level of rents R is divided between the two groups, and utility is linear in consumption, meaning I can treat the agents as risk neutral. The group that ends the period out of power gets θR, and the inpower group gets ð1−θÞR, where 0 θ 12 represents equality of institutions. Importantly, a higher value of θ implies institutions are more cohesive. Note that θ is assumed to be “exogenous” in the model, which is unlikely to be the case empirically. Cognizant of this concern, Part II of this book strives to establish a plausibly exogenous determinant of cohesive institutions that is unrelated (“causally prior”) to contemporary levels of rents. An important distinction between the two groups is thatthe group that started in opposition pays for their entire military expenditure AP , while payment for the incumbent group comes out of the public purse before resources are distributed. Then the game ends and payoffs are realized.
It would be straightforward to introduce incumbent advantage by assuming that power changes hands with probability α when both groups show the same level of violence for some α < 12, but the results do not meaningfully change, so I retain the simpler formulation. Specifically, I use the percentage of territory conquered by Islamic armies (corresponding to contemporary country boundaries) during the period – CE to “explain” contemporary levels of democracy. This percentage of conquered territory is clearly uncorrelated with timevarying rent streams in the modern era. The results would be unchanged if financing for the incumbent’s army came from a combination of public and private funding. It is worth noting that the equilibrium outcomes I identify in this single-shot game are also subgame perfect Nash equilibrium (SPNE) in a repeated game where each generation cares about other generations (to some degree). It can be shown that in a repeated game, the equilibrium outcomes identified in the single-shot setting comprise the “worst case” equilibrium. Future work could investigate the existence of Pareto dominant SPNE in the repeated game. For example, if one were to take the standard approach to repeated games and presume that the history of play is common knowledge between the players throughout the game, then one could
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
Analytical Framework Opposition Peace
Attack
Incumbent Peace
Payoffs:
]
FIGURE
. Game tree
, ]
Incumbent Repression
[q(R–L), (1–q)(R–L)]
Attack with L
[
(
) (
,
)
Attack with H
]
[q(R–H)–L, (1–q)(R–H)]
The two groups’ respective payoffs are the sum of their wage, their expected receipts of nontax income, and for the opposition, the cost of financing an army. Since the wage is always one by assumption, it is omitted for clarity. The payoff for the incumbent is: V I AP , AI ¼ 1 γ AP , AI ð1 θÞ R AI þ γ AP , AI θ R AI and the opposition’s payoff is: V P AP , AI ¼ 1 γ AP , AI θ R AI þ γ AP , AI ð1−θÞ R AI −AP :
B Optimal Investment in Violence In this section I solve for the optimal choice of violence for each group. Since this is a one period game with two actions, I start at the end of the game and work backward. Figure . illustrates the corresponding game tree, where the opposition moves first. The last row in the figure states the payoffs for each series of actions, with the opposition’s payoff listed first in parentheses. I first consider the case where the opposition does not attempt an insurgency. The incumbent then chooses between peace and using one-sided violence, which is interpreted as repression. A peaceful incumbent realizes a payoff of: 1 1 R V I ð0, 0Þ ¼ ð1−θÞR þ θR ¼ 2 2 2
use the Folk Theorem to establish the existence of Pareto superior SPNE. I leave this interesting line of inquiry to future work.
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
A Political Economy Approach
TABLE
. Incumbent’s optimal choice in political violence
R
Response to peace
Response to violence
L 1−2θ
ð1−θÞ R 2L1−2θ
L
L
L
L
H
0R L 1−2θ
2Lð1−θÞ 1−2θ
1−θÞ−L R 2H ð1−2θ
1−θÞ−L R > 2Hð1−2θ
Note: The first column shows a range of values for R. The next two columns show the incumbent’s optimal response to peaceful or aggressive opposition. Recall, L is a low deployment of armed forces. H is a high deployment of armed forces.
while a repressive incumbent is guaranteed to retain power, generating a payoff of: V I ð0, LÞ ¼ ð1 θÞðR LÞ: Note that the conflict function means the incumbent would never use a high level of violence against a peaceful incumbent, because high or low violence both guarantee success. The main aim of the model is to clarify the impact of rents on violence choices, so I allow R to vary and hold all other parameters fixed. The incumbent will be peaceful in response to peaceful opposition when: V I ð0, 0Þ V I ð0, LÞ which simplifies to:
ð1−θÞ Peace is the incumbent’s optimal response to peace if 0 R 2L1−2θ .
ð1−θÞ Low violence is the incumbent’s optimal response to peace if R > 2L1−2θ .
I assume that an indifferent group will break ties by choosing the lower level of violence. Similarly, if the opposition is violent, then the incumbent chooses between peace, low violence, and high violence. In the appendix to this chapter, I show Table . presents the incumbent’s optimal response to a peaceful or violent opposition: I now solve for the opposition’s optimal choice for every value of R. If L , then the opposition knows the incumbent will be peaceful, 0 R 1−2θ regardless if it attempts an insurgency. Thus, the opposition chooses between V P ð0, 0Þ ¼ 12 R and V P ðL, 0Þ ¼ ð1 θÞðR LÞ. Over the relevant range of R, the former value is larger; meaning that when L both the opposition and incumbent will be peaceful. This process can 0 R 1−2θ be continued for every value of R, by solving the incumbent’s response to peace and violence, and then determining which choice maximizes the opposition payoff. Table . presents the payoffs (see the chapter’s Appendix for the proofs).
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
Analytical Framework TABLE
. Incumbent and opposition optimal choices
Interval
R
Inc. response to
Inc. response to L
Opp.
Inc.
L 0 R 1−2θ
L 1−2θ
L
2Lð1−θÞ 1−2θ
L
L
L
2Lð1−θÞþL 1−2θ
L
L
L
L
L
H
H
ð1−θÞ R 2L1−2θ
R>
1−θÞþL < R 2Lð1−2θ 1−θÞ−L < R 2Hð1−2θ
2H ð1−θÞ−L 1−2θ
Note: The first column is a label for the range of R (unearned resources) being described. The second column is that range. The next two columns are the incumbent’s response to a peaceful or violent opposition. The final two columns are the actions that the opposition and the incumbent would actually take. The first two rows are both labeled as Interval because the outcome (both groups peaceful) is the same in both.
As R ranges from to infinity, there are four relevant intervals. In the first interval, both groups are peaceful because the prize for winning power is too small to justify fighting. There are two rows in Table . that make up Interval because the incumbent would be peaceful in response to rebellion in the first row and violent in response to rebellion in the second row. But in both rows the actual outcome would be peace on both sides. The incumbent will be tempted to use violence at a lower threshold of R than the opposition because the incumbent funds its army using public funds, and in Interval the opposition is peaceful while the incumbent uses the smaller army size against peace or violence, so the result is repression. In Interval the opposition would still always use a small army but now R is sufficiently large that the opposition is also willing to use, and both groups use low violence (civil conflict). In Interval R has increased to the point where the incumbent would employ a high level of violence in response to rebellion, but low violence in response to peace. It is disastrous for the opposition if the incumbent uses a high level of violence, because the incumbent is guaranteed to hold power and uses up state resources as well. This discourages the opposition from attacking, so the result is low-level repression again, the same as in Interval . These dynamics may arise in natural resource dependent countries where an incumbent and opposition fight over the rents (e.g., Angola, South Sudan, Colombia). This result – that violence can fall as R increases – is driven by two factors. First, since the incumbent has two choices of army size, as R rises to Interval the incumbent’s response to an attack can increase from a small to a large
As I demonstrate in Chapter , higher resource dependence on oil revenue is not a robust determinant of high intensity civil war in the Persian Gulf.
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
A Political Economy Approach
army size, which can deter the opposition. Second, the result stems from the sequential nature of the groups’ decisions, because the incumbent cannot respond to the opposition if they act simultaneously.
C Transitions to Democracy or Civil War The aim of Table . is to depict how the incumbent’s and opposition’s optimal choices change as R varies, holding the size of armies (L, H) and cohesive institutions ðθÞ fixed. Figure . also shows optimal choices but allows θ to vary. Specifically, I plot the four intervals depicted in Table ., while θ ranges from . and . and L and H remain fixed at . and . respectively. Going forward, I use the term “transition” to describe how changes in the model’s exogenous parameters (R or θ) leads to changes in violence (e.g., movements from repression to civil war, repression to peace). Figure . demonstrates how a country can transition between peace, repression, and conflict as rents and cohesive institutions vary. In general, higher levels of rents tend to increase the intensity of political violence: Holding fixed the level of θ, “moving up” the y-axis makes civil conflict more likely. This monotonic effect is consistent with the predictions of Besley and Persson (, , a,b). That said, Figure . demonstrates the possibility of nonmonotonic effects whereby a reduction in rents can make peace or civil conflict more likely. Whether a country experiences peace or conflict depends critically on the preexisting level of institutional cohesion. Transition to Democracy To observe a transition to democracy associated with a decline in rents, consider a country initially in Interval (low repression). In this instance, a reduction in rents can shift the country to Interval in which the incumbent decommissions its army (i.e., AI changes from L to ) and peace is achieved: a condition conducive for democracy. Moreover, observe that this peaceful transition to democracy is more likely to occur at higher levels θ. Since θ is “exogenous” in this model, this implies that a reduction in rents in societies with a higher preexisting (exogeneous) level of cohesive institutions can experience a peaceful democracy.
In Besley and Persson (, b), it is not possible for a decrease in rents to lead to an increase in violence. The incumbent in Besley and Persson () has only one possible army size so the deterrent effect observed in my model cannot occur. The two sides choose army size simultaneously in Besley and Persson (a), so there cannot be an exactly analogous deterrence effect, and the authors’ assumptions limiting the strategic complementarities and substitutabilities in the conflict function ensure that both groups’ army choices are weakly decreasing as R decreases. These numerical values are chosen for illustrative purposes, and other examples give similar results.
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
2.5
Analytical Framework
R
1.5
2
Repression (Interval 4)
1
Conflict (Interval 3)
0.5
Repression (Interval 2)
0
Peace (Interval 1)
0.2
0.25
0.3 q
0.35
0.4
FIGURE . Political violence as a function of cohesive institutions ðθÞ and rents (R) Notes: This figure simulates the violence outcomes as a function of θ (cohesiveness of institutions) and R (rents). The values of L and H (fraction of workforce deployed for low and high army size) are set to . and ., respectively. In Interval , both groups are peaceful. In Interval , the incumbent uses low violence in response to peace, and would have used low violence in response to violence from the opposition. Both groups use low violence in Interval . In Interval , the incumbent uses high violence in response to an attack, so the opposition is deterred and the result is peace from the opposition and low violence from the incumbent (repression). The values of R range from to infinity.
Transition to Civil Conflict Of course, Figure . also illustrates that a reduction in rents can also make civil conflict more likely. To see this, consider a country initially in Interval of “low repression.” In this case, a reduction in R can shift a country to Interval in which both the incumbent and opposition deploy small armies and engage in low intensity warfare. Moreover, being in Interval in the first place is more likely to occur at lower levels of θ. And since θ is assumed to be exogenous in the model, this implies that a reduction in rents in societies with a lower preexisting level of cohesive institutions may heighten the propensity for civil war. To the best of
In Chapter , I argue these features capture important dynamics potentially explaining why many Muslim foreign aid recipients experienced heightened internal political strife – characterized at the extreme with high intensity civil war – in the s and s and remained nondemocratic.
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
A Political Economy Approach
my knowledge, these differential and non-monotonic effects associated with a decline in rents – peaceful transition to democracy or heightened propensity for civil war – is novel in the theoretical literature on rents and political stability. Preserving Peace (Democracy) Finally, Figure . also shows that more cohesive institutions can preserve peace at higher levels of rents: a finding consistent with existing models (e.g., Besley and Persson b). When θ ¼ 0:2 there will be peace only when R 0:2667, but when θ ¼ 0:4, an incumbent will not be tempted to repress the opposition as long as R 0:6. This behavior stems from the expectation that the incumbent will be treated fairly even if it loses power, an expectation that underlies the peaceful transition of power that occurs in democracies. Moreover, this means that if two countries are initially both in Interval (repression) with the same level of R, the country with a lower θ will require a larger decline in R to shift to Interval (peace/democracy). In other words, more cohesive institutions can provide some protections against rent-driven civil conflict; a finding consistent with existing models of rent-seeking and political (in)stability (e.g., Lane and Tornell , Svensson ).
. A Propositions The discussion in the preceding section is summarized in the following three empirical implications and two corollaries: Proposition : Rents and Dictatorship Higher levels of rents can foster dictatorship (and other correlates, such as political repression), particularly in societies where preexisting institutions are less cohesive. This corresponds to moving from Interval to Interval or . Proposition : Reduced Rents and Political Stability Reduced levels of rents can shift a country from conflict or repression into democracy (i.e., movement from Intervals – to Interval ), but cannot move a country out of democracy. Proposition : Reduced Rents and Political Transitions as Cohesive Institutions Vary The threshold R below which democracy exists is greater where institutions are more cohesive (i.e., θ is greater). Empirically, this can imply two divergent outcomes that hinge on the cohesion of preexisting institutions: Corollary A: A reduced level of rents is more likely to foster civil conflict in societies where preexisting institutions are less cohesive. This corresponds to a movement from Interval to Interval .
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
Analytical Framework
Corollary B: While the same reduction in rents is more likely to lead to democratization in societies where preexisting institutions are more cohesive. This corresponds to a movement from Interval or to Interval . B Empirical Evaluation: A Road Map I evaluate these propositions in Parts II (“Conquest”) and III (“Rents”) of this book to show why so many Muslim majority societies are prone to dictatorship and civil conflict. In evaluating these propositions, I am cognizant that both rents and institutions are endogenous. Accordingly, each part will articulate the challenge posed by endogeneity and present identification strategies to better establish causal relationships. I briefly preview the research designs below. Conquest In Part II (“Conquest”), I establish a plausibly exogenous source of variation in cohesive institutions for contemporary Muslim societies. This is relevant as the level (quality) of cohesive institutions is assumed to be exogenous in my formal model. In Chapters –, I use qualitative historical evidence and cross-national statistical analysis to document the institutional legacy of Muslim conquest. The empirics strive to show how the initial period of Muslim military conquest following the death of Prophet Muhammad (– CE) established authoritarian political structures and then trace its subsequent trajectory of centralized autocracy in “conquered” territories through to the present. Since the process of military conquest and consolidation of centralized autocracy predated the discovery and distribution of rents (e.g., oil, receipts of foreign aid), the historical legacy of these institutions can serve as a source of exogenous variation in cohesive institutions in many contemporary Muslim societies. Thus, the institutional legacy of Muslim conquest may help explain the crosssectional difference in democracy between Muslim and non-Muslim societies; that is, the persistent democratic deficit depicted in Figure .. I interpret this finding to imply that the legacy of Muslim conquest fostered less cohesive
For example, the extraction of natural resource and subsequent generation of rents can be affected by domestic institutions. Menaldo () documents the higher levels of foreign investment and subsequent resource extraction in oil-producing democracies. Thus, the direction of this endogeneity suggests the effect of oil wealth on autocracy may be downward biased. Chapter discusses the emergence and persistence of an “Islamic equilibrium.” The chapter also presents cross-national causal evidence using a novel differences-in-differences research design. In Chapter , I examine the process of reconquest in medieval Spain and adoption of Islam in Indonesia and Malaysia (through trade) to document how the process of military conquest and subsequent rule in conquered territories mattered for the introduction and persistence of the classical Islamic equilibrium.
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
A Political Economy Approach
institutions in Muslim societies (exposed to conquest) in the contemporary period. Rents This exogenous source of variation in preexisting cohesive institutions in the modern period allows me to evaluate how variation in rents affects variation in dictatorship and civil war in contemporary Muslim and non-Muslim societies over time (as depicted in Figures . and .). Thus, I can directly test propositions – and corollaries A and B. In Part III of this book (comprised of Chapters –), I demonstrate how temporal variation in rents – oil revenues in oil producers and foreign aid in non-oil producers – explains patterns of dictatorship and civil conflict in Muslim “conquest societies” that are consistent with this chapter’s motivating trends in Figures .–. (and test propositions and A). To test proposition B, I evaluate the effect of rents in nonMuslim societies and demonstrate how their higher levels of preexisting cohesive institutions facilitated their peaceful transitions to democracy when their rents declined. Chapter demonstrates how oil revenues have sustained dictatorship in oilproducing societies that were exposed to Muslim conquest without elevating the propensity for civil conflict. This latter finding can be attributed in part to the presence of geopolitical rents from the United States that have strengthened the repressive capacity of governments in oil-producing Muslim countries to thwart domestic insurgencies. To the best of my knowledge, introducing geopolitical rents is new to the oil curse literature. Endogeneity, of course, looms large in identifying causal relationships. To mitigate these concerns, I follow Kris Ramsay () and use out-of-region disaster damage as an instrumental variable for oil wealth to estimate a causal relationship between oil and political violence, particularly in conquered Muslim territories. This chapter also provide evidence that oil producers with more cohesive institutions (i.e., those not exposed to Muslim conquest) have been able to transition to greater democracy. A finding consistent with proposition . In Chapters and , I probe the effect of rents in non-oil-producing developing countries. In these chapters I focus on rents that emanate from abroad – foreign aid in Muslim societies, “petrodollar” investments in Latin America, and trade subsidies in Eastern Europe – that also co-vary with movements in world oil prices. This second ‘fact’ is attractive in drawing causal inferences. Since oil prices are plausibly exogenous to economic and political conditions in non-oil-producing developing countries, I exploit their variation to predict movements in rents to explain variation in political violence in non-oil
I use data on US troop deployments and qualitative accounts of the US government’s strategic engagement in the Persian Gulf to justify this argument. Limiting the sample to non-oil-producing countries purges the direct effect of oil revenues on politics.
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
Analytical Framework
producing countries. I do so by employing difference-in-differences and instrumental variables research designs. In Chapter , I leverage a quasi-natural experiment of oil price driven aid disbursements to demonstrate how higher amounts of foreign aid entrenched autocracy in non-oil-producing Muslim countries (many of whom were conquered by Muslim armies), but their subsequent decline led to civil war. These dynamics are consistent with proposition A (and Figure .). Chapter evaluates proposition B. Using case studies – from Latin America and former Soviet Eastern Europe – and cross-national statistical analysis, the chapter argues that societies with preexisting institutions that are more cohesive transitioned to democracy when their rents declined. The chapter’s empirical analysis looks beyond Muslim societies – thus, providing a broader evaluation of the book’s theoretical arguments to a global sample of countries.
. This chapter presented the book’s analytical framework. It began by describing several empirical patterns that motivate the empirical and theoretical puzzles. These can be distilled to: () What explains the persistent democratic deficit between Muslim and non-Muslim societies? () Across Muslim societies, why do differences in a country’s rent structure (i.e., oil versus non-oil production) and its temporal variation explain movements in dictatorship and civil conflict? Motivated by the second question, the rest of the chapter advanced a theory linking variation in rents and exogenous (preexisting) cohesive institutions to varying intensities of political violence (dictatorship and civil war). Using a formal model, I generated several propositions, three of which may “answer” the two puzzles described above: () Higher levels of rents entrench dictatorship in societies whose preexisting institutions are less cohesive. () A decline in rents elevates the prospect of civil war in societies whose preexisting institutions are less cohesive. () A decline in rents can foster a peaceful transition to democracy in societies whose preexisting institutions are more cohesive. I ended the chapter by previewing the research designs to evaluate these propositions and empirical puzzles. My aim in Part II of the book is to demonstrate how the spread of Islam through military conquest (between
Chapter also introduces a new source of exogenous variation to measure preexisting cohesive institutions in non-Muslim societies.
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
A Political Economy Approach
and ) introduced an institutional trajectory that created a ‘sticky’ set of less cohesive institutions in contemporary Muslim societies. Thus, I leverage the historical legacy of Muslim conquest as a plausibly exogenous source of preexisting cohesive institutions to evaluates propositions and corollaries A and B in the contemporary period. Part III of the book does just that by examining how variation in rents explains the persistence of a democratic deficit between Muslim and non-Muslim countries and variation in political violence across oil and non-oil-producing Muslim societies.
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
Appendix A
0
POLITY (normalized) 0.2 0.4 0.6
0.8
Additional Figures
1970
1975
1980
1985
1990 Year
Oil FIGURE A
1995
2000
2005
2010
Non-oil
Democracy in non-Muslim countries
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
0.35 0.3 0.25 0.2 0.15
0
0.1
0.05
POLITY (normalized)
Civil war 0.1 0.15
0.2
0.25
A Political Economy Approach
1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 Year Civil war
Democracy and civil war in Muslim oil producers
100
0.04 Civil war
0
50
0.02 0 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 Year Civil war FIGURE A
Aid per capita
Civil war and foreign aid in non-Muslim non-oil producers
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
Aid per capita
0.06
150
FIGURE A
POLITY (normalized)
0
0
20
0.05
Civil war 0.1 0.15
40 60 80 Oil price (2015 US$)
0.2
100
0.25
Analytical Framework
1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 Year Civil war FIGURE A
Oil price (2015 US$)
Civil war and oil prices in Muslim oil producers
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
Appendix B Proofs for the Formal Model
In this appendix, I present the proofs for Section ..
. First, it is important to confirm the ranges of R presented in Table . cover the entire (non-negative) real line without overlap for every valid parameter value of θ, L, and H. That is, I need to show that 2Lð1−θÞ 2Lð1−θÞþL 1−θÞ−L L < 2H ð1−2θ . 1−2θ < 1−2θ < 1−2θ The first inequality holds because 0 < θ < 12 by assumption, so 2ð1−θÞ > 1. The second inequality is true because L > 0. To verify the third inequality, note that 2H ð1−θÞ > 6Lð1−θÞ, because of the assumption that H > 3L. This means 2H ð1−θÞ−L ð1−θÞ 1−θÞ−L 1−θÞþL > 2L1−2θ þ 4Lð1−2θ > 2Lð1−2θ , and implies that Table . covers the 1−2θ non-negative real line without overlap. The assumption that H > 3L guarantees the lower bound for R in Interval of Table . is in fact lower than the upper bound, which ensures that there are values of R which generate twosided violence. I next verify that the stated strategies for the incumbent are in fact optimal for the incumbent. In Section ., I showed that:
ð1−θÞ Peace is the incumbent’s optimal response to peace if 0 R 2L1−2θ .
ð1−θÞ Low violence is the incumbent’s optimal response to peace if R > 2L1−2θ .
On the other hand, if the opposition is violent, then the incumbent chooses between peace, low violence, and high violence. The payoffs are:
V I ðL, 0Þ ¼ θR These proofs are from Ahmed, Schwab, and Werker ().
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
Analytical Framework
V I ðL, LÞ ¼ 12 ð1−θÞðR−LÞ þ 12 θðR−LÞ ¼ 12 ðR−LÞ V I ðL, H Þ ¼ ð1−θÞðR−H Þ.
It is straightforward to show that the first value is highest when 1−θÞ−L L L 0 R 1−2θ , the second highest when 1−2θ < R 2Hð1−2θ , and the last value
1−θÞ−L , which confirms the values presented in Table .. is highest when R > 2Hð1−2θ
. Table . shows the optimal choices of both groups as a function of R. The incumbent’s choices in each region follows from Table ., so the only remaining task it so verify the opposition’s choices, conditional on how the incumbent would respond to peace and violence. The opposition’s payoffs for each of the possible outcomes is:
V P ð0, 0Þ ¼ 12 R V P ð0, LÞ ¼ θðR−LÞ V P ð0, H Þ ¼ θðR−H Þ V P ðL, 0Þ ¼ ð1−θÞR−L V P ðL, LÞ ¼ 12 ð1−θÞðR−LÞ þ 12 θðR−LÞ−L ¼ 12 ðR−LÞ−L V P ðL, H Þ ¼ θðR−H Þ−L.
L Consider the first row of Table . where 0 R 1−2θ . The incumbent will be peaceful regardless of whether the opposition attacks, so the opposition 2L , chooses between 12 R andð1−θÞR−L. The former is higher whenever R 1−2θ which includes the entire row of Table .. Next consider the range where the incumbent is peaceful in response to peace but uses low violence in response to violence. The opposition chooses between V P ð0, 0Þ and V P ðL, LÞ. The former is always higher, meaning that in the second row of Table ., the opposition will be peaceful. Next, consider the range of R where the incumbent will use low violence regardless of the opposition’s choice, which spans the third and fourth rows of Table .. A peaceful opposition group will receive V P ð0, LÞ, while an aggressive opposition group will receive V P ðL, LÞ. The former is higher when 1−θÞþL R 2Lð1−2θ , which confirms that in row of Table . the opposition will choose peace and in row it will fight. Finally, consider the region where the incumbent will use low violence in response to peace and high violence in response to violence. The opposition is guaranteed to lose power, so it prefers to remain peaceful, which confirms the final row of Table .
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
https://doi.org/10.1017/9781009367509.003 Published online by Cambridge University Press
PART II
THE INSTITUTIONAL LEGACY OF MUSLIM CONQUEST
https://doi.org/10.1017/9781009367509.004 Published online by Cambridge University Press
https://doi.org/10.1017/9781009367509.004 Published online by Cambridge University Press
The Conquest Equilibrium
The analytical framework in the previous chapter identifies two parameters that affect political violence: variation in rents and cohesive institutions. In particular, the formal model predicts that rents have a magnified effect on dictatorship and civil war in societies whose preexisting institutions tend to be less cohesive. In the formal model, these institutions were taken as exogenous; that is they are causally prior to the reception of rents. Therefore, to evaluate whether these predictions explain patterns of political violence in many contemporary Muslim societies necessitates establishing that these societies tend to exhibit preexisting institutions that are less cohesive today. This is the aim of Chapters – (Part II). Accordingly, this chapter is the first of three that strives to explain why so many contemporary Muslim societies have institutions that tend to be less cohesive compared to non-Muslim societies. Together, the collective aim of Chapters – is to provide an exogenous source of less cohesive institutions in Muslim societies (that predate the use of rents in the contemporary period) to evaluate the predictions from the formal model in Chapter . In this chapter, I provide a historical narrative for the emergence of an “equilibrium” of authoritarian power structures in the medieval period that set many Muslim societies on a trajectory toward less democratic politics in the contemporary period. The subsequent two chapters ( and ) provide statistical corroboration. The approach in this chapter is inspired by burgeoning research in political economy identifying the long-run effects of “critical junctures” in history on contemporary political and economic conditions (e.g., Nunn , Soifer
In the previous chapter, I provided evidence that less cohesive institutions are associated with less democracy; and Muslim societies tend to be less cohesive and less democratic, on average, compared to non-Muslim societies (Figures . and .).
https://doi.org/10.1017/9781009367509.005 Published online by Cambridge University Press
The Institutional Legacy of Muslim Conquest
). The long-run effect of these critical junctures is frequently “locked in” through economic and political institutions. For example, Daron Acemoglu, Simon Johnson, and James Robinson () argue that the emergence of transatlantic trade in the fifteenth century engendered stronger property rights protection, and, thus, served as a critical juncture in long-run economic development for many European states. In the Americas, Stanley Engerman and Kenneth Sokoloff () emphasize the role of initial factor endowments at the time of European colonialization – particularly human capital and levels of inequality – in shaping initial institutions (and policies) that affected long-run economic and political development. In some instances, past historical events can shape contemporary societies through non-institutional channels, such as cultural norms and beliefs. Nathan Nunn and Leonard Wantchekon (), for instance, demonstrate how the legacy of the Transatlantic and Indian Ocean slave trades has instilled mistrust among African citizens today. In this chapter, I follow an institutional approach like these studies. I identify the initial period of Muslim conquest and the governing institutions it introduced as a critical juncture in the subsequent trajectory of political institutional development in many contemporary Muslim societies. My account is intentionally quite brief as my objective is to concisely describe the institutional trajectory of centralized dictatorship that Muslim conquest introduced and its
According to Hillel Soifer (, ), “the distinct feature of a historical juncture with the potential to be critical is the loosening of the constraints of structure to allow for agency or contingency to shape divergence from the past, or divergence across cases.” Daron Acemoglu, Simon Johnson, and James Robinson () posit that countries where “initial” political institutions (those established before ) placed significant checks on the monarchy, the growth of Atlantic trade strengthened merchant groups’ ability to change institutions to protect property rights. And these changes were central to subsequent economic growth. Stanley Engerman and Kenneth Sokoloff () argue that initial levels of wealth and political power inequality led to the development of institutions (e.g., property rights, public goods provision such as education) that perpetuated these inequalities, furthering their harmful impact on long-run economic development. Nunn and Wantchekon () show that differences in trust levels within contemporary Africa can be traced back to Indian Ocean slave trades. Combining contemporary individual-level survey data with historical data on slave shipments by ethnic group, they find that individuals whose ancestors were heavily raided during the slave trade are less trusting today. Relatedly, several scholars have investigated the long-run determinants of economic institutions and development in Muslim societies. Timur Kuran () argues that Islamic law fostered economic institutions (e.g., inheritance and corporate law, incentives to invest in public goods) that contributed to the “long divergence” in economic development between the Middle East and Europe. Jared Rubin () and Ahmet Kuru () argue that Islamic clerics (ulema) helped “legitimize” inefficient economic institutions in Muslim societies; which in turn helps explain the long divergence. In this chapter, I focus on events that precede those advanced by these scholars and strive to explain how Muslim conquest affected the long-run trajectory of centralized autocratic rule in Muslim societies. My historical account also describes how Muslim conquest and the political institutional equilibrium it ultimately created was the source that elevated Islamic law and the ulema to prominence in Muslim societies in the Middle East and beyond.
https://doi.org/10.1017/9781009367509.005 Published online by Cambridge University Press
The Conquest Equilibrium
persistence ever since. In doing so, I plan to subsequently use variation in Muslim conquest as a plausibly exogenous determinant of preexisting cohesive political institutions in the contemporary period. In short, the institutional trajectory introduced by Muslim conquest may explain a significant share of the variation (or “stock”) in the persistent democratic deficit and cohesiveness in many contemporary Muslim societies (as depicted in Figures . and .). It is worth noting that this chapter’s historical narrative contrasts with those from other regions in which different “initial conditions” in the premodern era may have contributed to a different configuration of institutions and participating actors, such as hereditary landholders and a cadre of well-trained bureaucrats (Finer ). Some scholars argue this in turn contributed to different forms of political centralization – particularly in Europe – with potentially salubrious long-run effects on economic and political development, for example, via the strengthening of fiscal and legal capacity (e.g., Besley and Persson a, Dincecco , Tilly ). In Chapters and I evaluate the empirical veracity of these competing interpretations (as it relates to the institutional legacy of Muslim conquest). I proceed as follows. I begin by describing the initial period of Muslim conquest and then discuss the governing institutions imposed by Muslim rulers in conquered territories. In this second section, I identify two institutional innovations – the use of elite slave soldiers (mamluks) as the backbone of the state’s repressive capacity and a system of nonhereditary land grants (iqta) used to compensate these soldiers. I collectively refer to these as “mamluk institutions” which formed the foundations of a “classical Islamic equilibrium” of centralized, absolutist rule. The third section describes the persistence of this equilibrium configuration from the tenth century to the present, highlighting the role of a critical propagating agent – the ulema (clergy) – in legitimizing absolutist rule in conquered societies and its effect in weakening opposition from other groups in societies (e.g., intellectuals, landed elites, the bourgeoisie). Here, I briefly contrast the political equilibrium in Muslim societies with those in medieval Europe and Imperial China. In this section, I then describe how successive foreign invasions – by the Turkic/Mongols, Ottomans, and Europeans – did not disrupt the underlying equilibrium configuration of authoritarian political structures in Muslim societies. The fourth section describes “negative cases,” where I focus on contemporary Muslim societies that were not exposed to Islam via military conquest. Most notably these include societies in sub-Saharan Africa (e.g., Mali) and Southeast Asia (e.g., Indonesia and Malaysia). This discussion is important as the statistical analysis in Chapters and will show how the absence of mamluk institutions is not a particularly robust determinant of less
For excellent studies on Muslim conquest and the history of Islamic societies, see Crone (), Hodgson (), Kennedy (), Lapidus (), and Pipes ().
https://doi.org/10.1017/9781009367509.005 Published online by Cambridge University Press
The Institutional Legacy of Muslim Conquest
cohesive institutions (dictatorship) in non-conquered Muslim societies today. The final section offers some final remarks.
. In the span of years after the death of Prophet Muhammad in , the Islamic civilization had spread from the deserts of Arabia to the Pyrenees in northern Spain on its western flank and the Bay of Bengal in the east. While Arab armies conquered significant territories by the mid-tenth century, the Islamic Empire consolidated its authority in these lands by . This section defines what I mean by Muslim conquest and provides a brief historical overview of its process. A Concepts Before describing how Muslim conquest proceeded and its impact on governance in conquered societies, it is worthwhile to clarify what I mean by conquest and its participants. Muslims from diverse ethnic backgrounds formed the armies that conquered territories across the Iberian Peninsula, the Maghreb, the Middle East, parts of Central Asia, and the Indian subcontinent. Following the death of Prophet Muhammad in , almost all Muslims were Arabs, and in the early years of the conquests, one could use the term Arab and Muslim interchangeably to describe the armies of the conquest. However, as the empire expanded beyond the borders of the modern Middle East in the late seventh and early eight centuries, such a usage would be misleading, as Arabs formed only a proportion of the Muslim armies that conquered North Africa, Spain, and parts of Central Asia and the Indian subcontinent. As Hugh Kennedy (, ) notes, “what defined these armies were not their Arabness, even if the leaders were Arabs and the language of command and administration Arabic, but their identity as the armies of Islam – that is, religious identity had replaced the ethnic.” After a successful invasion, Muslim conquest imposed new political and religious elites in the conquered societies. Administration of the burgeoning empire was guided by two principles articulated during the reign of the second caliph, Umar (–). The first principle influenced how settlement of conquered territories should proceed. The conquest was followed by a process of settlement in which predominantly nomadic Arabs – and later other Muslims from the empire – took up residence in conquered territories, typically in specially founded garrison cities (Crone ). The second principle stipulated that conquered populations should be disturbed as little as possible (Lapidus , ). Consequently, Muslim conquest did not require the local populations convert to Islam in quick order. Indeed, while conquest and settlement occurred relatively quickly – by in the modern
https://doi.org/10.1017/9781009367509.005 Published online by Cambridge University Press
The Conquest Equilibrium
Middle East and by elsewhere – the conversion of conquered populations to Islam was a slow process. It was not until the tenth and early eleventh centuries that most of the population became Muslim. While conquest and settlement took a few decades, conversion of the local populations took over three hundred years.
B Expansion via Military Conquest Compared to the expansion of other empires (e.g., led by Alexander the Great or Genghis Khan), two “remarkable” features of Muslim conquest were the quick pace in which large swathes of territory came under the control of the caliphate and the “permanence” of its effect on conquered societies (Kennedy , ). According to the Islamic historian, Ira Lapidus (, ), Muslim conquest was “in part the outcome of deliberate state policy and in part accidental.” Muslim conquest started after the death of Prophet Muhammad in . To thwart the disintegration of various Bedouin tribes that the Prophet united under the banner of Islam, the first caliph Abu Bakr – based in Medina – initiated military campaigns against recalcitrant tribes to prevent their defection. While brutal, these ridda wars proved successful and after the battle of alAqraba in , Muslims expanded their sphere of power over the entire Arabian Peninsula (Kennedy , ). In reaching the peninsula’s frontiers, these incursions ultimately spilled over to southern Iraq – in particular, Ubulla and al-Hira, the former capital of the Lakhmid Kingdom – and the Syrian frontiers; and were driven in part by a lust for booty among the nomadic Bedouin tribes and a fundamental principle of early Islam that Muslims should not attack each other. Notably, Muslim raids into the Syrian territories encroached on the boundaries of the Byzantine Empire, ultimately inciting Constantinople to send a major expedition to confront the Muslim forces in southern Palestine. The resulting battle at Ajnadayn (in ) proved to be a watershed for the future trajectory of Muslim conquest. Fighting for the first time as an army – rather than as separate raiding parties – the Muslims emerged victorious, inspiring a grander vision for the caliphate. According to Ira Lapidus (, ): “With this victory their [Muslim forces] ambitions became boundless; they were no
As Kennedy (, ) notes: “Spain and Portugal are the only conquered countries at this time where the spread of Islam has been reversed; by contrast we now think of Egypt as a major source of Arab culture and of Iran a stronghold of militant.” Conquest helped reconcile these two “tensions.” As Kennedy (, ) states: “Once the ridda had been subdued and the tribes of Arabia were brought under the control of Medina, the leadership had no choice but to direct the frenetic military energies of the Bedouin against the Roman and Sasanian empires. The only way of avoiding an implosion was to direct the Muslims against the non-Muslim world.”
https://doi.org/10.1017/9781009367509.005 Published online by Cambridge University Press
The Institutional Legacy of Muslim Conquest
longer raiders on the soil of Syria seeking booty, but contenders for control of settled empires. What began as inter-tribal skirmishing to consolidate a political confederation in Arabia ended up as a full-scale war against two empires.” In the aftermath of the battle of Ajnadayn, Muslim forces attacked the Byzantine province of Syria. By , Muslims took Damascus, with Baalbeck, Homs, and Hama falling soon thereafter. Jerusalem was taken in and by , Muslims controlled northern Syria and some surrounding Mesopotamian towns. The next Byzantine province to succumb to Muslim conquest was Egypt, a strategically important region for Constantinople. Within the span of a year, Muslim forces had taken Heliopolis and Babylon, and controlled the whole of Egypt, except for Alexandria which fell two years later (in ). With the quick succession of victories in Egypt, Muslim raiders eyed the rest of North Africa. While Muslims conquered Tripoli within a few years (by ), they would take another seventy-five years to control the rest of the Maghreb, with the westernmost city, Tangiers, finally capitulating in . Muslims also vied for the most prosperous and populated Byzantine territories in Anatolia and the Balkans. But here, the Byzantines repelled Muslim incursions and turned the tables by engaging in almost continuous border warfare to retake their former territories (e.g., Syria, Egypt). Indeed, the survival of the Byzantine Empire would serve as a permanent barrier to further Muslim expansion into southern and eastern Europe. On the east, however, Muslims decimated the existing Sasanian Empire. In doing so, Muslim forces completely conquered the territories of modern-day Iraq (by ) and Iran (by ) and soon thereafter, penetrated parts of Central Asia and the Indian subcontinent (albeit, with bloodier and longer lasting military campaigns). Over the next several decades, Muslim armies launched new campaigns on a grand scale. To the west, Muslim forces methodologically conquered North Africa between and ; paving the way to absorb much of the Iberian Peninsula between and . In Anatolia in the north, Muslim armies launched three great but failed expeditions (in , , and ) to capture Constantinople. In the Caucasus, Muslims fought the Khazars and after extremely prolonged and bloody campaigns, the capitals of Transoxania (present day Kazakstan, Tajikistan, and Uzbekistan), Bukhara, and Samarqand fell in and . By , Muslim armies controlled territory up to the Chinese frontier (Kennedy , ). Just south, Muslim armies had traversed into the Indian subcontinent and taken control of Sind by . Thus, in slightly over a century after the death of Prophet Muhammad,
Because Egypt was politically centralized but scarcely urbanized, the conquest was essentially instantaneous. At the Battle of Poitiers in , Muslim armies were repelled from southern France. This effectively contained Muslim control to the Iberian Peninsula only. These regions correspond to contemporary Northern Iran, Kazakhstan, Uzbekistan, and Tajikistan.
https://doi.org/10.1017/9781009367509.005 Published online by Cambridge University Press
The Conquest Equilibrium
Muslims controlled the entire modern Middle East, and had incorporated North Africa, Spain, Transoxania, and some of the Indian subcontinent into their empire as well. The expansion of the caliphate via military conquest took many different forms. In most instances, the process was peaceful whereby the people of town and country would agree to the imposition of terms from the Muslim conquerors. This would usually involve the payment of a tribute and a promise not to aid the enemies of Muslims. This form of indirect rule was prevalent in territories that fell to Muslim forces relatively quickly, such as in Sind (Kennedy , ). Yet, in some instances, where the local population did not peacefully succumb to Muslim armies, conquest was more violent. In these cases, conquest meant the violent sack of a city, expropriation of its wealth, and the execution of citizens opposing Muslim rule. The sacking of Istakhr in Fars or Paykand in Transoxania are clear examples of this. Several factors underlay the initial success of the Muslim conquest, including the relative weakness of rival empires, the nomadic advantage of Bedouin (Arab) tribes in warfare, the “easy” terms imposed on conquered societies, the gains from trade through conquest, and the appeal of converting to Islam. Fred Donner (, ), for example, notes the “fortuitous weakness of the Byzantines and Sassanians just when the Muslims began their expansion.” Had Muslims launched their conquest at the beginning of the seventh century, it would have been far more arduous. This is noteworthy because Muslim forces did not possess any specific technological advantage over their adversaries, nor a greater strength in numbers. However, they did possess several military advantages including greater mobility (since the initial fighters were from nomadic tribes), high quality of military leadership and centralized command (first from Medina, then Damascus, and later Baghdad), and a unifying religious motivation (Kennedy , –). In the initial years of conquest, for example, greater mobility meant that Muslim forces could retreat into the desert to take refuge, to regroup after a defeat, or to take the enemy unaware (e.g., with night or early morning raids). Non-military factors also contributed to the success of Muslim conquest. Caliph Umar’s second principle meant that Muslim conquerors respected local customs and practices, provided the indigenous population paid the necessary taxes (that were levied on non-Muslims) and did not aid the enemies of Muslims. Most notably, conquered subjects did not have to convert to Islam, although many did convert (slowly) over time. These relatively “easy” terms contrasted sharply with those of other empires, who frequently imposed their
In stark contrast to other empires, such as the Byzantines, the local populations were not forced to convert to the invader’s religion. See Kennedy () for a thorough discussion. He provides an excellent historical narrative of Muslim/Arab conquests in the first years after the death of Prophet Muhammad.
https://doi.org/10.1017/9781009367509.005 Published online by Cambridge University Press
The Institutional Legacy of Muslim Conquest
state religion on their subjugated populations (e.g., Christianity by the Byzantines). Over time, significant proportions of the indigenous populations – many of whom were Christian – found various tenets of the Islamic faith (e.g., its “openness,” emphasis on equality) appealing and converted. New converts could enjoy economic benefits of joining the umma (community of believers). For example, they would be exempt from the fees imposed on non-Muslims and could gain greater access to trading opportunities that the expanding Islamic Empire offered. In summarizing the “great” Muslim conquests and their transformative impact, Hugh Kennedy (, ) declares: “In the final analysis, the success of the Muslim conquest was a result of the unstable and impoverished nature of the whole post-Roman world into which they came, the hardiness and self-reliance of the Bedouin warriors and the inspiration and open quality of the new religion of Islam.” This account suggests the process of Muslim conquest led to the subsequent centralization of political authority across many geographically and ethnically diverse societies. After the initial period of military conquest (which effectively ended by around ), Muslim rulers introduced important political institutional innovations that led to a “classical Islamic equilibrium” – characterized by nondemocratic political institutions and governance – that has persisted to the modern era. The remainder of this chapter discusses the key features of this institutional equilibrium and its persistence.
. A Existing Institutional Apparatus Muslim conquest starting in the seventh century did not upend existing institutional structures from late antiquity. Consistent with Caliph Umar’s (–) second principle, upon conquering new territories, the caliphate incorporated existing institutional arrangements into its governance, particularly the administrative and monetary structures of the dominant pre-Islamic Empires of the Middle East. Hugh Kennedy (, ) summarizes the degree of institutional incorporation: The Islamic world inherited and maintained a system of public finances, a system with no parallel in the West, where public taxation had collapsed with the Roman empire.
Recently, economists have leveraged these insights to explain the expansion of Islam via opportunities for international trade. For example, Michalopoulos et al. () argue that pre-Islamic regions with more unequal land distribution – like desert areas – were likely to become Muslim to better exploit the gains from trade. Adopting core Islamic principles, such as income redistribution from the rich to the poor, helped mitigate trade disruptions and gain access to trade networks.
https://doi.org/10.1017/9781009367509.005 Published online by Cambridge University Press
The Conquest Equilibrium
Governments collected money and paid salaries in cash . . . [T]he fiscal role of the state ensured the continued vitality of urban life in the Islamic Middle East when it effectively disappeared in north-west Europe and the Byzantine empire.
In other parts of the Islamic Empire, the caliphate incorporated Greek, Coptic, and Persian administrative structures (von Grunebaum , ), drawing heavily from religious minorities to staff the bureaucracies (although, over time they converted to Islam). According to Julia Bray (, ) administrative secretaries “carried on the culture of the old imperial bureaucracies with little change,” often transferring ideas of governance (e.g., Hellenistic and Persian thought) through the circulation of administrative handbooks and other forms of advice literature regarding governance (Morony , ). However, over time, bureaucratic institutions eventually adapted in a way consistent with Islamic principles, particularly in the legal sphere (Morony , ). B Mamluk Institutions While modifying some existing institutional structures, Muslim conquest introduced several institutions tasked with the goal of maintaining political order. These institutions proved effective in consolidating political authority after the initial military campaigns. Two institutional innovations contributed to an equilibrium that centralized power in an absolutist sovereign. The first is the widespread use of slave soldiers (mamluks) as the core of the military elite. Relatedly, a second institution extended state control over agricultural land as means of payment for slave soldiers (iqta). In combination, these two institutions (mamluks and iqta) – which I collectively refer to as mamluk institutions – helped consolidate political authority in the caliph and served as a “long ‘knot’ connecting the early Islamic period to modern times” (Tsugitaka , vii).
I do not claim that the Islamic equilibrium described in this chapter is unique in fostering autocratic “state centralization” during the premodern period. Other pathways are plausible. For example, absolutist rule was the norm in other states, including Europe and China in the premodern era (Anderson , Finer , Chapters –). However, as I argue shortly, a different configuration of institutions (i.e., the absence of mamluk institutions) led to a different political trajectory of increased checks and balances on the sovereign. It is worth noting that in a few modern sub-Saharan African (SSA) countries exposed to Islamic conquest, mamlukism and iqta were not a central feature in their governance. In these territories, the adoption of Islam frequently helped unify kingdoms (particularly across different tribes) and consolidate the absolutist rule; centralization may have helped thwart the spread of democracy through subsequent European colonialization (Hariri ). These include several precolonial African kingdoms such as the Ghana Empire, Mali Kingdom, Kilwa Sultanate in Tanzania, and the Kanem-Bornu Empire in Chad and Nigeria.
https://doi.org/10.1017/9781009367509.005 Published online by Cambridge University Press
The Institutional Legacy of Muslim Conquest
Mamluks Focusing on the first institutional innovation, while not unique in premodern times, the institution of slave soldiers as the backbone of the military elite (mamluks) emerged during the Abbasid period in the ninth century. This institution spread across the conquered territories, becoming a defining feature of their politics for hundreds of years (Crone ) as “all post-Abbasid regimes were made up of an alien military elite” (Lapidus , ). For example, after the demise of the Abbasid Caliphate, foreign-born soldiers helped consolidate power for several successor regimes, including the Buwayhid (in Iran, –), Ghaznavid (spanning Iran, Afghanistan, Pakistan, and parts of Central Asia, –), Qarakhanid (Central Asia, –), and Seljuk (Antanolia, Iraq, Syria, –). In subsequent regimes, mamluks reinforced the sultan’s rule across conquered territories, spanning modern-day Spain (Glick , and ), India (Jackson ), Egypt (Finer , –), Turkey (Barkey ), the Middle East, and Central Asia (Lapidus , –). In the ninth century, the superior bureaucratic and financial position of Muslim rulers enabled them to recruit mamluks from abroad, rather than conscript military support from the local population. As fighters, mamluks possessed elite military skills (e.g., horsemanship, ruggedness from their native lands in Central Asia) that local populations lacked and served as a critical “repressive agent” to thwart internal rebellions (Pipes ). The most sought after mamluks were of tribal origin, imported from areas “marginal to the settled Islamic world” (Crone , ), where they acquired military training as horsemen during their youth (Hogdson , ). These slaves were typically purchased from the Caucasus and Transoxania while still young and
For example, the Byzantine and Chinese Empires employed ‘slave mercenary’ forces (Finer , ). The Abbasids first “imported” slave soldiers in large part to quell internecine fighting (Blaydes , ). In contrast, more precarious fiscal positions compelled European rulers to conscript military support from the local population, who in turn demanded some constraints on the sovereign’s political authority (Strayer ). Several factors made “Turkic mamluks” particularly valuable. Medieval Arab historian Ibn Khaldun argued that relative to soldiers recruited from more settled areas, Turkic mamluks enjoyed a greater ability to develop a sense of group solidarity, or asabiyyah. This solidarity proved useful for training and in their actual combat for the sultan. More recently, scholars have identified the superior use of the stirrup by Turks relative to non-Turks. Daniel Pipes (, ) notes how the introduction of the stirrup “enhanced the power of the peoples living where horses could be raised – primarily in the steppe lands and in deserts – and reduced the strength of peoples living in densely inhabited areas.” Echoing this view, Marshall Hodges (, ) writes the “steppes formed the most outstanding source of young slaves” because of the “boyhood military training as horsemen.” Relatedly, Pipes (, ) observes how living in the mountains, deserts, and steppes of Central Asia and the Caucasus made Turkic mamluks resilient to conditions of extreme hardship, leading them to be exceptionally healthy and lean.
https://doi.org/10.1017/9781009367509.005 Published online by Cambridge University Press
The Conquest Equilibrium
then underwent years of training at the courts of their masters. This training sought to both enhance their military skills and strengthen their loyalty to the sultan. While mamluks were the property of the sultan and thus obedient to him, they did not necessarily experience economic or social hardships (as most slaves typically endure). In Muslim societies, the social position of a mamluk did not reflect his personal servitude/obedience to the sultan, but rather the status of his master. For example, a mamluk could rise to be a general or a minister of the state. Upon cessation of their service to the sultan, mamluks were manumitted, becoming freedmen with limited legal rights to property, personal security, and marriage. As Lapidus (, ) notes, “in this institution the exclusive personal loyalty of the slave or client-soldier to his master was crucial.” During their service to the sultan, particularities of mamlukism strengthened the loyalty of these elite soldiers to the sovereign. Because mamluks were recruited from abroad, links between local elites and the military power structure were weak, thus reducing the likelihood that the military might join with these elites to overthrow the sovereign. Indeed, the continual replenishments of new recruits from Central Asia limited the “local loyalties through marriage, commercial ventures and land purchase” that might otherwise have occurred (Kennedy , –). The confluence of these factors implied that slave soldiers were “outsiders entirely reliant on the state not just for cash but their very survival” (Kennedy , ). For the sultan, mamluks were thought to be the “safest to rely on” (Hodgson , ), offering the best defense of the sovereign’s interests (Lapidus ). Iqta Closely tied to the use of mamluks was the system of nonhereditary grants to tax agricultural land (iqta) that served to pay the elite military slaves. This institution was first introduced in mid-tenth century Iraq and eventually spread across the Islamic Empire, including its core in the greater Middle East (Iran, Afghanistan, Syria, Egypt) and extending to its peripheries in North Africa, Central Asia, Spain, and India (Glick , Jackson , Lapidus , Tsugitaka ). In practice, while iqta was tweaked to better suit local conditions, the use of nonhereditary grants to compensate the military remained central to absolutist rule in conquered Muslim societies after . For
David Ayalon (, –) highlights various policies and social relations that enhanced the dissociation of mamluks from the local population. First, mamluks typically kept their Turkish names, even after converting to Islam and tended to speak Turkish with each other with often limited knowledge of the local language. Second, mamluks were mainly married to female slaves from their countries of origin rather than local women. Third, the sons of mamluks (who did not enjoy mamluk status themselves) tended to marry women from the local populations thus offering one opportunity for assimilation into the local society. This last point alludes to the inability of mamluks to transmit their status to their offspring. Rather, as Pipes (, ) notes the qualities that made mamluks so valuable were not innate but acquired characteristics.
https://doi.org/10.1017/9781009367509.005 Published online by Cambridge University Press
The Institutional Legacy of Muslim Conquest
example, in his analysis of the contemporary Middle East, Sato Tsugitaka (, ) argues: .. If we consider that soyurghal and tuyul in Safavid and Qajar Iran or timar in Ottoman Turkey had essentially the same characteristics as iqta, we can therefore say that the iqta system continued to exist until the early th century in Iran, and until around the seventeenth century in Turkey. Even in Egypt and Syria, when these regions were conquered by the Ottomans, the muqata’a system was introduced anew taking features from the iqta system. The iqta system proper was utilized by rulers for controlling Islamic states and societies from the tenth to the sixteenth centuries.
While my discussion of iqta here focuses on the relationship between the sultan and iqta holders (muqtas), iqta was more than a land system as it had profound effects on state-society relations in Muslim territories. The iqta system generated two important social relationships: the relationship between the sultan and muqtas, and the relationship between muqtas and peasants (fallahun). In practice, this meant the slave soldiers who were granted iqtas were obliged to provide military service to the sultan and were granted the authority to levy taxes on the production of peasants and local townspeople. In describing its institutional variation – both across different Muslim societies and over time – historians have distilled several key features of the iqta system. Claude Cahen (), for example, emphasizes four central characteristics: () the acquisition of iqta revenue by the muqtas and their military service to the sultan were strictly controlled by the government, () iqta holdings were not inherited, () even iqtas held for life were rare, and () independent rules were not formed by muqtas in the local provinces. While Cahen’s characterization highlights the co-dependent relationship between the sultan and his military forces, it emphasizes the sultan’s underlying control of revenues. For example, the sultan had the ability to revoke a mamluk’s iqta holdings at his discretion. The importance of iqta in consolidating absolutist rule helped reduce challenges to the sultanate’s underlying autocratic political structure. Though well compensated, mamluks were unable to transform themselves into a hereditary landed baronage because they were not permitted to transmit their status directly to their offspring (Finer , ). Crucially, upon the death of an iqta holder, ownership of the land returned to the state. As a result, because the land could not be inherited or sold, the income generated from the tax rights gave soldiers a vested interest in the incumbent regime’s longevity. Indeed, as Chris Wickham (, ) notes, despite various attempts by iqta holders to
Chapter in Tsugitaka () provides an overview. Cahen notes that had property been transferable to the offspring of mamluks, iqta would have been like the feudal system in Europe. This reduced the incentive of iqta holders to supply local public goods. Instead, an alternate but inefficient institution – the wafq – emerged (Kuran ).
https://doi.org/10.1017/9781009367509.005 Published online by Cambridge University Press
The Conquest Equilibrium
make land assignments hereditary, Islamic states resisted, mainly by redistributing the iqta frequently and maintaining the bureaucratic assessment of the taxes with the state. Thus, iqta reinforced the loyalty of the elite military forces to an absolutist and centralized sovereign. As Patricia Crone (, ) emphasizes, this concentration of land holdings in the sovereign, in which government service remained disassociated with land ownership, was unique among premodern societies. Even in sultanates led by mamluks (e.g., Egypt from the thirteenth to sixteenth centuries), upon the death of a mamluk caliph, a new cadre of soldiers rotated in with a re-parceling of iqta holdings. For example, in his ‘special note’ on Mamluk Egypt (–), Samuel Finer () characterizes the governing regime as a “one generational military aristocracy of , men” in which there were “periodical reallocations of the iqta after the incumbent’s death” (Finer , and ). As discussed above, mamluk institutions were central in maintaining political order in the Islamic Empire’s “core” of the greater Middle East from the Abbasid Caliphate onward (Tsugitaka Lapidus , –). Their effectiveness was “imported” elsewhere, as governance along “Abbasid lines” existed on the empire’s peripheries. On the eastern edge, the various Delhi sultanates (from to ) were “sustained by a small and factious military aristocracy that ruled over the vast Indian subcontinent” (Lapidus , ). From the outset, this military aristocracy incorporated both iqta and mamluks in which the Delhi state “rested upon an elite corps of Turkish mamluks, who largely provide the military leadership, the provincial governors, and the great officers of state” (Jackson , ). As Muslim rule expanded further under the Mughal regime, the ruling military elite were compensated in cash
In this description of Mamluk Egypt, Finer emphasizes the distinctiveness of the military aristocracy. They consisted of mamluks from Central Asia, who were completely dissociated from the local populations, spoke Turkish, married slave-women from Central Asia, and wore different clothes. Mamluks were compensated with iqta that were non-transferable to their offspring. Finer’s account also describes the prevalence of internal competition among the military aristocracy that led to short-lived political tenures for successive mamluk caliphs. However, he also notes, the superior fighting ability of this caliphate and its ability to repel Europeans during the Crusades. Iqta and mamluks were central to the founding, expansion, and consolidation of Muslim rule in India. Jackson (, ) clearly states this in his introduction: “When Muslim forces under the Ghurid sultan, Mu’izz al-Din Muhammad b. Sam, made their first major breakthrough into Hindustan in the s, they brought with them two institutions that had long since taken root in the Islamic world. One was iqta, or assignment of land or its revenue, in some cases in return for military service (sometimes misrepresented as ‘fief’ on the Western European model). The other was the mamluk, or military slave. Mamluk status, it should be stressed, bore none of the degrading connotations associated with other types of slavery: Mamluks – generally Turks from the Eurasian steppelands – were highly prized by their masters, receiving both institution in the Islamic faith and a rigorous training in martial arts, and were not employed in any menial capacity.”
https://doi.org/10.1017/9781009367509.005 Published online by Cambridge University Press
The Institutional Legacy of Muslim Conquest
or by an estate called a jagir, which was the equivalent of iqta (Lapidus , ). Jagirs were awarded to military officers of the sultan and indigenous elites (e.g., local princes, tribal leaders) to help defray the costs of a military contingent and administrative functions (e.g., the collection of revenue from tenant farmers). On the western flank of the empire, mamluk-like institutions featured in Muslim Spain. To consolidate its political authority after the initial Muslim conquest in the early eighth century, the Umayyad dynasty in Spain followed the “Abbasid pattern” in which it “suppressed local revolts and built up a client army of soldiers coming from north of the Pyrenees” (Lapidus , ). Like their counterparts in other Muslim controlled territories, these soldiers were foreign-born slaves – known locally as saqaliba – from northern Spain, Germany, and the Slavic countries. Land grants were allocated to generate revenue for the central government. Thus, by the tenth century, the caliphate in Spain reached its zenith of “fame, power, and wealth” in which the caliph became “a grand and distant figure” who “ruled through a hierarchy of slaves” (Glick , and ). The use of mamluks was a “Muslim institution” as “there were no mamluks in Spain after the expulsion of Muslims” a few centuries later (Crone , ). D The Political Equilibrium By , the institutional features associated with Muslim conquest had generated a classical Islamic equilibrium. This equilibrium had four main effects. First, the consolidation of land holdings by the sovereign (i.e., through iqta) increased the rents, broadly construed, to the sovereign. Second, it decreased the rents flowing to the sovereign’s potential rivals, such as the commercial classes. Third, mamlukism diminished potential revolutionary threats from the commercial classes and the masses. For instance, the disassociation of the elite military forces from the local population decreased the likelihood mamluks would band with local elites to overthrow the sultan. Fourth, because mamluks were unable to transmit their status and associated benefits to their children, the elite military force increased its dependence on and loyalty to the monarch. Thus, both the economic and political concentration of resources in the sovereign helped the monarch mitigate the dual challenges of authoritarian politics: the problems of authoritarian power sharing and authoritarian control (Svolik ). On the former, the consolidation of rents in the sovereign that were temporarily granted to the military elite helped the sovereign effectively share power with the enforcers of the caliph’s absolutist rule. On the latter, a military elite that was dissociated from the local population increased the
As I discuss shortly, I demark as an approximate endpoint for the first period of Islamic conquest. After , conquered territories faced a series of foreign invasions.
https://doi.org/10.1017/9781009367509.005 Published online by Cambridge University Press
The Conquest Equilibrium
caliphate’s repressive capacity to crush potential rebellion from the masses. Thus, mamluks comprised the ruler’s “repressive agents” (Rubin ). Together, Muslim conquest helped centralize political authority in a caliph with limited checks and balances on his rule.
. A Legitimizing Agents: The Ulema This equilibrium proved remarkably enduring and was supported by an important additional propagating agent after the tenth century: the ulema (clerics). While the ulema emerged as the primary check on the caliph’s power (Hodgson , ), in practice religious elites worked in tandem with the military (the sovereign backed by mamluks) to perpetuate absolutist rule. In their capacity as propagating agents, clerics helped “legitimize” the caliph’s rule to the masses; and in return, the caliph instituted Islamic law (sharia), which elevated the authority of the ulema as the principal interpreters of the law (Rubin , –). The importance of this alliance among the clerics, military, and state (or an “ulema-state alliance” as referred to by Ahmet Kuru ) is noteworthy as clerics were not central agents in the expansion and consolidation of Muslim political authority prior to the tenth century. Rather, clerics gained prominence in the eleventh century as the Abbasid Caliphate sought to counter the rise of Shii states in North Africa, Egypt, and Syria. To unify Sunni Muslims against this threat, two Abbasid caliphs declared a “Sunni creed” whereby nonadherents (i.e., certain Shiis, rationalist theologians, and philosophers) were deemed to be apostates and subject to execution. Subsequently, clerics became integral to the legitimization of Sunni orthodoxy and by extension the caliphate itself. This ulema-state alliance strengthened in successor Sunni states, starting in the Seljuk Empire (–) whose territory spanned most of Central Asia, Iran, Iraq, and Anatolia. Central to Seljuk rule was its explicit expansion of mamluk institutions and co-option of the ulema with generous funding (and in exchange, the ulema’s legitimization of the military state). Ultimately, the ulema-state alliance proved to be effective in maintaining political order and persisted across Muslim societies from the twelfth century onward. In a recent study, Ahmet Kuru (, ) summarizes the alliance’s prevalence across various Muslim regimes and its persistence over time:
The ulema are not discussed as central actors (if at all) in prominent accounts of the Muslim conquests (e.g., Kennedy , Lapidus ). My description of Muslim conquest in the previous section does not mention clerics.
https://doi.org/10.1017/9781009367509.005 Published online by Cambridge University Press
The Institutional Legacy of Muslim Conquest
From the twelfth to the fourteenth century, the Seljuk model of the ulema-state alliance spread to other Sunni states in Andalus, Egypt, and Syria, particularly the Mamluks [caliphs]. The Crusader and Mongol invasions accelerated the spread of this alliance because Muslim communities sought refuge from the chaos of foreign invasion in military and religious authorities. Later, around the sixteenth century, Muslims established three powerful military empires: the Sunni Ottoman, the Shii Safavid, and the Sunni-run (but non-sectarian) Mughal Empires. These empires established versions of the ulema-state alliance in territories extending from the Balkans to Bengal.
While an ulema-military state alliance served to maintain political order in Muslim societies, several recent studies have also traced its long-run socioeconomic effects (Chaney , Kuru , Rubin ). Jared Rubin () argues the alliance stifled technological and economic innovation in Muslim societies; thus, helping to explain the “long divergence” in economic development between Muslim societies and their non-Muslim counterparts, particularly in Europe (Kuran ). For example, Rubin () documents how the alliance delayed the adoption of the printing press – and the subsequent diffusion of knowledge and innovation – in Muslim societies. Because clerics enjoyed monopoly rents through their hand scribing of the Quran and other religious texts, the sovereign impeded the adoption of the printing press which would have threatened the economic welfare of this important legitimizing agent. More generally, the alliance elevated the importance and centrality of Islamic law (which the ulemas demanded) in Muslim societies. This, as Timur Kuran () meticulously traces, fostered “inefficient” economic institutions, such as the wafq (Islamic charitable trust) and particularities of Islam inheritance law (e.g., equal division of property among family members) that led to lower levels of investment and public goods provision. The ulema-state alliance, coupled with mamluk institutions, has also been attributed to political stagnation. In particular, the use of temporary land grants through iqta prevented the emergence of a landed aristocracy that could have challenged and worked to constrain the sultan’s political authority. The use of iqta effectively retained landholdings and its associated stream of revenue within the state (caliph’s) control. The ulema supported the alliance as it was the beneficiary of state patronage (e.g., salaries) and various forms of monopoly rents (e.g., hand scribing of the Quran, other religious texts and official documents). The ulema also enjoyed an elevated status as the legal interpreters of sharia. Consequently, Eric Chaney (, ) contends that in tandem “both the military and religious elites worked to resist the emergence of rival centers of power, such as merchant guilds, that could have facilitated institutional change.”
In this regard, Islamic law and associated institutions, such as the wafq, are aftereffects of the Islamic equilibrium. And as I will argue in the empirical chapters, these institutions are therefore “post-treatment” in evaluating the effect of Muslim conquest on state centralization and contemporary political institutions.
https://doi.org/10.1017/9781009367509.005 Published online by Cambridge University Press
The Conquest Equilibrium
B Absolutist Rule in Europe and China The ability of religious and military elites in Muslim societies to thwart challenges from competing groups differed from the institutional trajectory in medieval Europe and China. Despite lacking mamluk institutions, rulers in these territories pursued alternate paths toward more centralized forms of autocratic governance (and subsequent institutional legacies). In Europe, a series of fiscal shocks in the eighth century compelled monarchs to conscript local forces from landholders rather than collect taxes. Due to their lack of funds, sovereigns often compensated the service of elite troops (i.e., mounted warriors or knights) through permanent land grants (North, Wallis, and Weingast , ). While this method of military recruitment became the “basis for state-building” (Strayer , ), it also facilitated the rise of a landed aristocracy, which served as a check on the monarch’s political authority and the subsequent emergence of parliamentary institutions (Downing , ). The resulting configuration of propagating agents contrasted with those in Muslim societies: “In medieval Europe, the Church, economic elite, and military propagated kings. In the Middle East, religious authorities and the military were the primary propagating agents” (Rubin , ). In Imperial China, by contrast, absolutist rule was propagated with a different configuration of actors and alliances; a configuration that did not resemble mamluk institutions to its west. From the T’ang Empire (–) through successive dynasties (until ), imperial rule exhibited a quintessential palace-type system: the Emperor within his court buttressed by his army on one side and his bureaucracy on the other. As Samuel Finer () describes in his comparative study of government in the intermediate ages, the specialization and competence of the bureaucracy served as a distinctive feature (relative to other pre-industrial empires, such as the Romans, European, and Islamic) in both legitimizing and constraining the Emperor’s authority and helping govern the empire (Finer , –, –). For the former, the rigorous
Blaydes and Chaney () and Strayer () discuss Europe’s divergent institutional trajectory further. The emperor’s inner court functioned like a cabinet of ministers of variable but small size. For many emperors, his ministers came from aristocratic families. For other emperors, particularly during the Ming dynasty, his ministers were palace eunuchs (Finer , ). These individuals did not possess any independent wealth or status; and thus, were completely obedient and loyal to the emperor. In many respects, their loyalty and elevated status mirrored those of the elite slave soldiers (mamluks) in Islamic societies. However, unlike mamluks, eunuchs were not repressive agents (they were counselors to the emperor) and were not extended material benefits, such as arable/taxable land. The specialization and competency of the bureaucracy stemmed in large part from the competitive three-level examination process to recruit civil servants. This merit-based system was innovative for its time. Historians estimate that less than percent of the population had the sufficient qualifications/scores to serve as bureaucrats.
https://doi.org/10.1017/9781009367509.005 Published online by Cambridge University Press
The Institutional Legacy of Muslim Conquest
training of bureaucrats in Confucianist teachings ensured the Emperor adhered to the “Confucian Way.” According to Finer (, ), “the Confucian persuasion, now so dominant and all-pervasive that its infallibility was taken for granted, resembled ancient Judaism, Christianity, and Islam in this, that it articulated standards to which the ruler must conform.” On the former, bureaucrats managed the empire’s day-to-day affairs, particularly by overseeing the governance of subnational units (e.g., provinces, prefectures) and the collection of tax revenue, primarily of small-scale farming. Indeed, writing in the fifteenth century, Father Matteo Ricci (, ) observed the importance of an “aristocracy of the literati” (i.e., the bureaucrats who were learned in Confucian literature) in legitimizing and sustaining absolutist rule. In the Chinese palace system, the military played a largely defensive role (Dincecco and Wang ). It functioned primarily to repel invasions/raids on China’s expansive frontiers and to subdue internal peasant revolts. Unlike their counterparts in Muslim societies, the military elite in China did not govern provinces nor help collect revenue. There was no system equivalent to iqta to compensate their military service; rather, they were paid from tax revenue, either in hard currency (silver) or in-kind (e.g., with food). Thus, while my discussion of Imperial China has been brief, the main point remains: The configuration of actors sustaining absolutism – and state formation more broadly – in this empire differed from that of Muslim societies to its west.
According to Finer (, –), “Confucianism believed in good men, not in laws; in personal ties of loyalty and reciprocity; not in abstract, universalistic rules and regulations.” While in practice, many emperors deviated from these principles – and some were delegitimized and ousted from power – they needed the legitimacy of the ‘Confucianist literati’ (bureaucrats). The bureaucrats, in contrast, did not need the approval of the emperor. It is worth noting that bureaucrats also sought to self-regulate themselves through the Tribunal of Censors. This agency sought to identify and punish corruption and deviation from Confucianist principles. Mark Dincecco and Yuhua Wang () argue differences in the “geography” of political violence between China and Europe may have contributed to their different political trajectories over the long run. In particular, the prevalence of internal rebellions and sufficient military forces in China meant the emperor did not have to make political concessions to fight interstate wars. In contrast, the greater prevalence of interstate wars in Europe compelled the sovereign to grant political concessions to the landed elite in exchange for tax revenue. Tax revenue in imperial China was collected at the village level from individual (family) land holdings. For example, in the T’ang Empire, each individual/family was allotted some land and was taxed at a constant (and relatively low) rate. A comparison with the Americas is less relevant as centralized autocracy in the medieval period was non-existent as there were no formal empires in the Western Hemisphere. The Mayan civilization had largely collapsed by the ninth century and the Aztec and Incan Empires had not consolidated their rule (i.e., Mexico for the Aztecs and the Andes for the Inca) until the early fifteenth century. The arrival of Europeans – primarily the Spanish and Portuguese – in the late fifteenth century and early sixteenth century eventually decimated the indigenous societies (and their political institutions) through warfare and the introduction of diseases. European colonial powers subsequently introduced largely extractive and non-inclusive political institutions in their
https://doi.org/10.1017/9781009367509.005 Published online by Cambridge University Press
The Conquest Equilibrium
C An Autocratic Trajectory after Muslim Conquest The prevalence of absolutism in Muslim societies (after ) features prominently in the writings of many political theorists. For example, in an important passage of The Prince, Machivelli singled out the autocratic configuration underlying Ottoman rule as an institutional order distinct from European states: “The entire Turkish empire is ruled by one master, and all other men are his servants; he divides his kingdom into sandjaks and dispatches various administrators to govern them, whom he transfers and changes at his pleasure . . . they are all slaves, bounden to him.” Highlighting the absence of a hereditary aristocracy, Francis Bacon, a century later, observed how the absence of a hereditary aristocracy differentiated absolutist rule in Turkey from the rest of Europe: “A monarchy where there is no nobility at all, is ever a pure and absolute tyranny; as that of the Turks. For nobility tempers sovereignty and draws the eyes of the people somewhat aside from the line royal.” These observations are even more surprising given the history of foreign “interventions” in conquered territories after Muslim conquest. Despite a series of foreign invasions (primarily from the Mongols and Turkics in central Asia), Ottoman rule, European colonialism, and some periods of institutional reforms (e.g., the Tanzimat era in the nineteenth century Ottoman Empire), the division of political power among a small group of individuals – the sultan, his military, and religious elites – endured in conquest territories. The remainder of this section describes the persistence of this autocratic configuration in conquered Muslim territories from the tenth century through to the present through the waves of foreign intervention. Mongol and Turkic Rule The first wave of foreign interventions began in the late eleventh century, with invasions of Turkic peoples from Central Asia. The most significant of these peoples were the Seljuks, who would eventually enjoy dynastic rule over much of Anatolia, the Levant, Mesopotamia, Persia, and large parts of Central Asia. While some scholars document how this “influx of fresh barbarians” sought to “alter the balance of society” in Islamic territories (Brett , ), they did not abandon the core institutional feature of the classical equilibrium. Foreign fighters remained critical to governance as “the steppes of Central Asia provided the manpower for the Saljuq and Mongol imperial armies.” Indeed, mamluks and iqta lay at the core of Seljuk state philosophy, in which “the empire, together with the subjects, belongs to the Sultan” and that iqta holders were protectors of the land, but not landowners (von Grunebaum , ).
colonies; often with lingering (and pernicious) long-run effects on politics (e.g., Acemoglu and Robinson , Engermann and Sokoloff ). Cited in Anderson (), . Ibid., .
https://doi.org/10.1017/9781009367509.005 Published online by Cambridge University Press
The Institutional Legacy of Muslim Conquest
While a subsequent wave of Mongol invasions further introduced new migrants to conquest territories, Mongol rulers did not fundamentally alter the underlying social and particularly, the governing structures. For example, while the Mongol invasions decimated agricultural productivity, Genghis Khan extended land grants to the Mongol army in the form of iqta (Tsugitaka , ). Moreover, the mass conversion of the Ghazan Khan (the Mongol ruler of Iran) and his army to Islam cemented Mongol rule to the Islamic equilibrium. As Arjomand (, ) notes, after his conversion, Khan “called himself ‘the king of Islam,’ thus adopting the Islamic royalism as the type of regime that was to prevail in the eastern part of the Muslim world” (). That said, both waves of invasions and their associated migration of peoples from Central Asia transformed the Arab orientation of conquest territories by infusing Turkic and Mongol cultural influences with those of the Arabic and Persian traditions (Bosworth , ). Ottoman Rule By the sixteenth century, the Ottoman Empire emerged as a centralized state that was able to repel major conquests from Central Asia. Although the Ottoman Empire introduced a series of institutional reforms, its core institutional structure was built on previous Middle Eastern states. In describing these institutional inheritances, Lisa Blaydes (, ) identifies the most important as “centralized bureaucracies that managed assigning tax rights of state-led land to members of the military elite and continued use of slave soldiers.” In particular, the high degree of political centralization in the Ottoman sultan was critical to holding together a multi-ethnic empire that had the capacity to absorb and co-opt challengers (Blaydes , ). In the event of challenges to state, slave soldiers remained the backbone of the military elite with an “utter lack of autonomy” (Barkey , ) from the sultan. Comprising the janissary, these soldiers were typically recruited through forced tribute of Christian male children (aged –) in the empire. These children were converted to Islam and underwent extensive military training; often reaching the highest levels of the military and the governing elite (e.g., many served as governors of Ottoman provinces). While janissaries were compensated quite well, they were not permitted to marry. Consequently, because these individuals continued to hold no independent basis of power or wealth – acquiring them only through service to the state – they remained highly loyal to the sultan. For the empire, the janissary were critical repressive agents in safeguarding Ottoman political authority, particularly through their ability to crush “banditry” on the empire’s periphery (Barkey ).
Enslaving Muslims was prohibited. Thus, the Ottomans drew their slave recruits from their Christian subjects. In many instances, Christian families volunteered their male children for military service, as it was viewed as a potential form of social advancement. Troops in the janissary corp were well compensated, especially relative to the masses in the Ottoman Empire.
https://doi.org/10.1017/9781009367509.005 Published online by Cambridge University Press
The Conquest Equilibrium
The Ottoman Empire also instituted a land grant system (timar) like iqta (Wickham , ). Under the timar system, the state’s control of agricultural land permitted it to allot the right to tax revenue from that land in return for military service or other service. As a result, this discouraged the development of a robust land tenancy system and the rise of a feudal aristocracy, unlike in Western Europe. As Barkey (, ) notes, “in marked contrast to the position of the nobility within feudalism, such grants were neither heritable, nor did they connote any rights of jurisdiction over the direct producers, and the revenues attached to them were set by the sultan’s treasury.” In addition to employing these two core mamluk institutions, the multiethnic and religious composition of its empire motivated Ottoman rulers to institute the millet system (Barkey and Gavrilis , Karpat ). This system formally separated subjected people into their own religious communities or millets, with autonomous control over their laws, schools, and general welfare. The legal and often geographic separation of these communities served an additional political function: It helped deter the formation of unified opposition to Ottoman rule from its subjected groups. While Ottoman rule maintained the core institutional features of the Islamic equilibrium, it did not necessarily foster conditions conducive for robust economic development, which translated to political weakness (Kuran ). In response to both internal nationalist movements and external aggression (primarily from European powers), Sultan Malmud II launched a series of reforms, the Tanzimat (re-organization). The scope the Tanzimat era (–) reforms were ambitious, ranging from policies to transform land holdings, the legal code, taxation, welfare regimes, military conscription, and social structure (e.g., education). These reforms were often quite comprehensive and sought to imitate systems in Europe. For example, in the area of taxation, the Ottoman historian Stanford Shaw (, ) avers: The main financial goals of the Tanzimat reformers involved shifting the tax burden from the land to urban wealth, supplanting indirect with direct tax collection by salaried agents of the state, replacing the excise taxes, which were levied mainly on households and land plots regardless of ability to pay, and abolishing many of the historic exemptions which had been granted over the centuries.
For each type of millet, the sultan designated community leaders (e.g., rabbis, priests, ulema) to serve as mediators between Ottoman authorities and their communities. As Karpat () documents the millet system also had lasting legacies for state and identity formation after Ottoman rule. In many modern Middle Eastern societies, particularly those of the Levant where multiple ethnic and religious communities have long resided, politics continue to operate under distinct legal codes. One implication, which is evaluated in Chapter , is whether ethnic and religious “fractionalization” (that may have been stoked from policies of the late Ottoman period) is associated with less democracy in contemporary Muslim societies that experienced Muslim conquest. Similar types of reforms were also undertaken in Qajar Iran and in Morocco (the nizam aljadid, or new order).
https://doi.org/10.1017/9781009367509.005 Published online by Cambridge University Press
The Institutional Legacy of Muslim Conquest
In practice, the implementation of these reforms often lacked enough bureaucrats to effectively survey lands and collect taxes, particularly in the countryside (Shaw , ). Moreover, the process was often arduous. Cammett et al. (, ), describe an episode in which the “Ottoman bey (ruler) of Tunis in introduced a head tax in cash (the majba), which led to a rural revolt in ” and was ultimately rescinded forcing the government to borrow an “unmanageable” amount from abroad. Despite the ambitious aims of the Tanzimat era reforms, they are generally considered to have come up short in changing the trajectory of the Ottoman Empire’s demise and “paradoxically, many of the reforms may have furthered the ability of the European colonial powers to undermine the empire” (Cammett et al. , ). Crucially, with regards to their political implications, the reforms did not change the empire’s underlying political structure. According to Stanford Shaw (, ) “the most important aim of the Tanzimat was autocracy and centralization” (Shaw , ). In summarizing the overall influence of Ottoman rule, Samuel Finer (, ) characterizes it as an “unimproving, ultra-conservative [. . .] blanket on society.” European Colonialism While mamluk institutions had formally disappeared by the nineteenth century, the autocratic trajectory they generated in Muslim territories may have influenced the nature of European colonial rule. Much like their predecessor (the Ottomans), the British and French generally governed through existing institutional structures in a system of indirect rule (Fieldhouse , ). According to Gerring et al. (), indirect rule was relatively efficient as the existence of stable precolonial authority structures provided the British and French with sufficient institutional infrastructure with which to control the territories. Rather than rebuilding or importing their own institutions, it was less costly to govern through existing structures. In doing so, indirect rule helped to “perpetuate the dominance of indigenous elites [leading to] a period of virtual social standstill in the territories” (Fieldhouse , and ). Some scholars speculate that indirect rule may have limited the diffusion of democratic norms and institutions from European colonialism and settlement (e.g., Hariri , Olsson , Woodberry ). In one study, Ola Olsson () probes whether the length (duration) of colonial rule affected
Examples of indirect colonial rule in Muslim-majority states include the British mandates (including modern-day Pakistan and Bangladesh), French governance in Syria, and European rule in Morocco, Tunisia, Algeria, and the emirates along the edges of the Arabian Peninsula. In their analysis, Gerring et al. (, –) contrast their “institutional theory” of direct/ indirect rule with other hypotheses, such as those pertaining to “access,” “power,” “revenue,” “agenda-centered,” “difference/identity” and “normative.” Colonial rule in the Americas also co-opted existing, indigenous structures such as the use of forced labor (Arias and Girod ). This inference follows from studies that find a positive effect of direct colonial rule on contemporary democracy.
https://doi.org/10.1017/9781009367509.005 Published online by Cambridge University Press
The Conquest Equilibrium
contemporary levels of democracy, finding a strikingly robust positive relationship, particularly among former British colonies and countries colonized after during what he deems a “more liberal era.” Robert Woodberry () hones into one potential channel: missionary schools. He finds that missionary schools were a “crucial catalyst” (Woodberry , ) in the initiation and spread of several important antecedents – religious liberty, the printing press, mass education, newspapers, voluntary organizations, most major colonial reforms, and the codification of legal protections for nonwhites – for stable representative democracy in the contemporary period. An alternate channel emphasizes the role of European settlement in colonized territories. Jacob Hariri (), for example, shows that colonies with greater European settlement tend to experience the “exporting” of more (nascent) norms and institutions of democracy from the colonizer. His evidence is consistent with Acemoglu, Johnson, and Robinson’s () finding that “neoEuropes” tended to exhibit greater secure property rights protection and more “inclusive” institutions today. The supposition that European colonizers (e.g., settlers, missionaries) may have been beneficial for democratic governance during the colonial period, however, overlooks instances where colonial powers (governments) actively sought to quell indigenous democratic movements in several Muslim territories. For example, drawing on Arabic-language archives, Elizabeth Thompson () traces how European powers “stole” democracy from many parts of the Arab world. Focusing on the Syrian Congress and the country’s constitution, Thompson reveals a concerted effort by European politicians to thwart the creation of a democratic Syrian Arab state in the aftermath of World War I. In one instance, Thompson (, xiv) notes: France and Britain suppressed a American poll of Syrians’ political preferences, favoring self-rule under a constitutional monarchy. Instead, the French and British declared the Syrian Congress illegal and colluded to destroy the government in Damascus. Fearing the nascent League of Nations might yet uphold Syrian independence, the French prime minister explicitly ordered his generals to destroy every trace of the Syrian Arab Kingdom’s government. French troops consequently ransacked the offices of the Congress and Faisal’s palace. Since then, Syrian historians have searched, to no avail, for an original copy of the constitution, which is known to have been in France’s possession as late as August .
Thompson’s historical account of Syria after World War I is consistent with Adria Lawrence’s () broader analysis of how European powers often opposed, and sometimes violently, indigenous nationalist movements in colonized territories. Lawrence (, ) observes “in the colonial world, the obstacles to collective organizing were immense,” and in “their empire, the French repressed nascent movements, jailed activists, and confronted mass protests with displays of force and often violence” and “fought two major wars in Algeria and Indochina to defeat nationalists, resulting in one of the century’s bloodiest decolonization processes.”
https://doi.org/10.1017/9781009367509.005 Published online by Cambridge University Press
The Institutional Legacy of Muslim Conquest
To the extent that European colonial powers preserved prior governance structures in (conquest) Muslim territories – e.g., by opposing democratic nationalist movements – these accounts do not necessarily negate my central argument that Muslim conquest set conquered territories on an institutional trajectory toward more autocratic governance in the contemporary era. That said, it is plausible that European colonialism may have exerted an independent causal effect on politics in contemporary Muslim societies. I evaluate the veracity of this hypothesis in Chapter . Post-colonial Era Some scholars argue that the purported “modernizing” influences of European colonialism further entrenched existing institutional structures and state-society relations. Writing with Muslim societies in mind, historian Bernard Lewis (, ) claims that “modernization in the nineteenth century, and still more in the twentieth, far from reducing this [historical] autocracy, substantially increased it.” Indeed, upon gaining independence after World War II, the new states inherited and largely maintained their autocratic institutions in which kleptocratic regimes emerged to keep political power concentrated, with a strong military influence (Brownlee ). Many post-colonial and predominantly secularist leaders in Muslim societies, such as Egypt, Turkey, Indonesia, Iran, Iraq, Pakistan, Syria were former military officers who “tried to use Islam to legitimize their regimes” (Kuru , ). Unsurprisingly, in these and other Muslim states, the military continued to be omnipresent in public affairs, where about one in five governments in the s had “active” military involvement. As a consequence, in many post-colonial Muslim countries, a new ulema-state alliance emerged. While of course mamluk institutions had long disappeared, some contemporary observers do note similarities to the past and have called “their [Muslim] rulers Mamelukes, alluding to the slave soldiers who exercised unrestrained and arbitrary power in those countries” (Kedourie , ). For example, the emergence of a distinct military class in Egypt has been described as “neo-Mamluks” (Bulliet ). Thus, in many contemporary countries that experienced Muslim conquest, the concentration of political power and prevailing institutions bear a striking resemblance to their past: centralized autocracy. D Sequential Implications The preceding sections describe a trajectory (“historical legacy”) of centralized autocracy in territories conquered by Muslim armies. To summarize, following the death of Prophet Muhammad in CE, Muslim armies helped spread the
These leaders were influenced by socialist and fascist ideologies, and especially authoritarian modernist ideas, in general. This percentage is higher with lower thresholds of military involvement in politics. These calculations are based on data from Svolik ().
https://doi.org/10.1017/9781009367509.005 Published online by Cambridge University Press
The Conquest Equilibrium Muslim conquest (632–1100)
“Autocratization”
Propagating agents and institutions (i.e., mamluks, ulema, iqta)
European colonial “strategy” (e.g., indirect rule)
Contemporary institutions FIGURE
. From Muslim conquest to contemporary political institutions
religion through conquest. Starting from the Arabian Peninsula, by CE Arab armies had conquered territory in the Iberian Peninsula, large swathes of land in North Africa, up into the Caucuses and Central Asia, and eastward to India. By , Muslims had consolidated political authority in the borders of forty-five contemporary states. This centralization of authority featured several distinctive institutions. Conquest introduced two important institutions: absolutist rule backed by elite slave soldiers (mamluks) and the temporary land grants to compensate these soldiers (iqta). Aided and legitimized by propagating agents – the sultan, his mamluks, and clerics – this “classical Islamic equilibrium” persisted through subsequent foreign interventions (e.g., Ottoman rule and European colonialism) to the contemporary period. In general, the Ottomans employed mamluks and adapted iqta, while the use of indirect rule by European powers did not change the underlying institutional structures in colonized Muslim societies. Upon independence, these autocratic structures largely remained. Thus, Muslim conquest generated a trajectory that: () centralized political authority in absolutist rule in conquered territories during the initial phase of conquest with () autocratic structures that persisted. Figure . summarizes this institutional trajectory.
. - Whereas military conquest played a critical role in the spread of Islam in the Middle East, North Africa, Spain, Anatolia, and Central and South Asia, elsewhere it played a marginal role. In many societies that are Muslim-majority today, particularly in sub-Saharan Africa and Southeast Asia (SSA), Islam
https://doi.org/10.1017/9781009367509.005 Published online by Cambridge University Press
The Institutional Legacy of Muslim Conquest
spread primarily through trade. Critically, the expansion of Islam in these territories had a very light institutional footprint and did not introduce mamluk institutions nor its trajectory of centralized autocracy thereafter. Understanding these “negative cases” is important in () placing scope conditions on my argument and () properly evaluating how the institutional legacy of Muslim conquest matters for the prevalence (stickiness) of autocracy in conquered Muslim territories today. Below, I briefly discuss these cases, identifying how societies in these territories initially adopted Islam and their prevailing structure of political order/governance at that time. A Islam in Sub-Saharan Africa Starting in the seventh century, Muslim traders, teachers, and merchants introduced Islam into the Sahara and “under their influence the veiled Sanhaja of the desert entered the process of Islamization” (Levtzion , ). Military conquest did not play a role. Rather, “linked by trading networks, family connections, teacher-student relationships, and Sufi fellowship, Muslim communities were established within small-scale regional states and in stateless societies” (Lapidus , ). In some communities, Muslims converted local rulers and elites, while in others they formed peaceful minorities in non-Muslim societies. In West Africa, Islam was essentially a “royal cult” and religion of trading communities, while East Africa’s closer proximity to the Middle East also made Islam a “popular religion” among a broader swath of the population (Lapidus , ). Gaining access to lucrative trade networks propelled the earliest converts to Islam, particularly as Berber and Arab merchants were strongly discouraged from trading with non-Muslims. For example, writing in tenth century, an authoritative jurist of Kairouan (in modern-day Tunisia), Ibn Abi Zayd issued an injunction: “Trade to the territory of the enemy and to the Land of the
Admittedly, my description below is brief and glosses over the rich history of Islam in Africa and Southeast Asia. For example, there is significant variation in how Islam is practiced in these regions, how the religion affected local political economies, and the contribution of these Muslim populations to Islamic knowledge. Moreover, I will not describe the trajectory of governance in these non-conquered Muslim territories after their initial “exposure” to Islam. This means, for example, I will not discuss the impact of European colonial rule in these societies. Lapidus () provides a thorough treatment with several chapters on Muslim societies and their history (through to the present) in Africa and Southeast Asia. For instance, in the Sahara, the main trade routes were north–south, linking the Sahel with Morocco, the Niger bend with the central Maghreb, and Lake Chad with Tripoli. As Levtzion (, ) notes, “the dynamics of this trade system . . . induced a more complex pattern with diverse ramifications. Merchants from Ifriqiya were prominent in Awdaghust; Tadmekka traded with Ghana, and Ghadames with Gao. There were also important routes from west to east.” African slaves and gold were the primary “exports” to North Africa, the Middle East, and beyond.
https://doi.org/10.1017/9781009367509.005 Published online by Cambridge University Press
The Conquest Equilibrium
Sudan is reprehensible.” This provided a strong incentive for African merchants to convert to Islam and join the Dar al-Islam (House of Islam). Indeed, in many instances, major non-Muslim urban centers had both Muslim and nonMuslim towns (side-by-side) to ensure trading opportunities. Accessing this Muslim trade network was transformative for premodern African societies: “Through trade and Islam, Africa south of the Sahara became irreversibly linked to the wider world” (Levtzion , ). Politically, despite the conversion of the governing elite to Islam – primarily in West Africa – the rulers did not seek to re-order politics to mimic other Muslim societies of the time. In particular, they did not institute mamluk institutions and the clerics did not align with the state to implement sharia. Rather, these newly converted Muslim rulers continued to govern based on clan and kinship alliances (Levtzion ). Consider, for example, the Mali Kingdom, which from the early thirteenth to the end of the sixteenth centuries was the dominant regime in West Africa and the main center of Islam. The kingdom was a “typical West African Islamic regime” (Lapidus , ) in which politics was organized around the family and village units, with the head of the family serving as both the priest and chieftain. Groups of villages (called kafs) would align and coalesce primarily along shared kinship and would “choose” a ruler (and associated royal clan). This political organization would aggregate further through a consortium of related clans. Functionally, revenues were directly extracted from taxes and tributes from dependent communities. Administratively, there was a central territory (i.e., a capital with surrounding village networks), directly controlled by governors, and tribute-paying territories on the periphery ruled by vassal chieftains. The political structure of these Western African Muslim kingdoms has been associated with patterns of economic development in the long run. Recognizing that Muslim kingdoms were organized via alliances among clans, rather than through mamluk institutions, Vincent Bauer, Melina Platas, and Jeremy Weinstein () find that areas in sub-Saharan Africa that were ruled by Muslim kings or chiefs tend to be underdeveloped relative to areas not governed by Muslim rulers. They attribute this is to the legacy of greater “premodern state centralization” that limited the penetration of confluence influence, particularly of Christian ministries. Their argument aligns with mine: Political centralization in premodern Muslim societies may have stunted development in the long run.
Quoted in Levtzion (), . Rather when Islam became a royal cult, rulers built mosques, brought Muslim scholars from Cairo and Fez to establish centers of Islamic learning, and the king and his entire court (who also converted to Islam) engaged in public prayers. With support from the commercial classes (who became Muslim to participate in international trade), Islam served as “bond of empire” and gave “cultural support to the dynasty” (Lapidus , ). The mass of the agricultural population remained pagan.
https://doi.org/10.1017/9781009367509.005 Published online by Cambridge University Press
The Institutional Legacy of Muslim Conquest
In summary, governance in the most centralized African Muslim states/ kingdoms was based on alliances between clans and not with a cohesive military apparatus. Revenues came from local, predominantly communal based farming in which a system like iqta did not emerge. Initially, Islam was practiced primarily by the governing elite, merchants, and scholars, while the masses (who primarily engaged in agricultural production) remained pagan. To the extent that political organization in Africa had an Islamic character, it did not resemble the equilibrium political configuration in other territories led by Muslim rulers.
B Islam in Southeast Asia Trade also played an integral part in the spread of Islam in Southeast Asia. Islam spread throughout Southeast via maritime trade like a “slow, giant wave, starting from northern Sumatra as early as the thirteenth century and reaching Borneo and the southern Philippines in the sixteenth century” (Houbens , ). Early trading communities comprised of merchants and scholars (predominantly Sufi clerics) introduced Islam in various coastal cities, and over time, indigenous populations inland converted to the religion. This process, however, differed from those in other Muslim societies, particularly from the experience in sub-Saharan Africa. First, whereas the earliest Muslim converts in Africa were the rulers, this was not the case in Southeast Asia. Here it “appears probable, moreover, that the first significant Islamic communities were important functional groups within the city or state, which eventually gained sufficient power to overthrow a pagan king” (Johns , ). This meant that Muslim rule was not initiated by the state/dictator. Second, Sufi preachers were important propagating agents in transmitting Islam and their more spiritual based interpretation of the faith. Importantly, rather than allying with the state/government, Sufi clerics frequently elevated their prominence and reach through “orders” with traders and craft guild affiliations called tarikahs (Gibb , ; Johns , ).
Gradually over time, significant shares of the agricultural population converted to Islam. However, many converts continued to practice their pre-Islamic (pagan) rituals (Levtzion ). A ruler might convert to Islam, “either because a wandering derwish presented it to him with a charismatic conviction, or because of its emergence as a power among his people” (Johns , ). In Malaysia and Indonesia, Islam was able to take roots “thanks to the devoted labours of the Sufi missionaries – peripatetic preachers ranging from the whole known world, sometimes voluntarily espousing poverty, sometimes engaging in trade; they were frequently associated with trade or craft guilds, according to the order (tarkah) to which they belonged; they taught a complex syncretic theosophy largely familiar to the Indonesians [and Malaysians], but which was subject to, although an enlargement on the fundamental dogmas of Islam” (Johns , ).
https://doi.org/10.1017/9781009367509.005 Published online by Cambridge University Press
The Conquest Equilibrium
In the premodern era, clerics and merchants in Southeast Asia worked together to spread the religion; an ulema-state alliance was absent. As a result, these two conditions – in sharp contrast to the Middle East and other conquest territories – did not re-order society in Southeast Asia. Culturally, the mystical version of Islam preached by Sufi clerics permitted the religion to place “only a thin veneer upon indigenous traditional culture of animis and Indian influence” (Houbens , ). Sufis were “consciously or unconsciously [. . .] prepared to preserve a certain measure of continuity with the past, and to use the terms and elements of pre-Islamic culture in an Islamic context” (Johns , ). Politically, while many rulers ultimately converted to Islam, they did not seek to centralize their political authority along ‘Abbasid lines.’ Rather, they largely continued their existing governance practices and did not implement any institutions that mimicked mamluk ones prevalent in other Muslim societies. At the local level, however, Sufi clerics did influence political decision-making by fusing Islamic principles with local customs and practices. However, these did not fundamentally alter the underlying organization of politics across various Malay-Indonesian societies (e.g., Acheh, Malaya, Minangkabau, Java), which at the village level remained relatively egalitarian with collective decision-making (Lapidus , –). In short, the adoption of Islam in Southeast Asia did not put these societies on a path of centralized autocracy along the lines of those in territories conquered by Muslim armies. In Chapters and , I will provide statistical evidence corroborating this characterization of Islam and politics in nonconquered territories.
. This chapter argued that Muslim military conquest may comprise a critical juncture in the political development across conquered territories: Muslim conquest introduced a trajectory of institutionalized centralized autocracy. After the initial period of military conquest, the widespread adoption of mamluk institutions across conquered territories helped consolidate the caliphate’s absolutist rule. By , conquered territories featured a unique political institutional equilibrium of centralized autocracy (dictatorship). Aided by an additional propagating agent – clerics (ulema) – this equilibrium persisted, despite successive foreign rule after from the Turkics, Mongols, Ottomans, and European colonial powers. While formal mamluk institutions no longer exist in
Hamilton Gibb () notes that after the demise of the Abbasid Caliphate in the thirteenth century, Sufi clerics played a critical role in preserving the unity of the Islamic world. The Sufi orders gradually became unified and disciplined foundations and developed affiliations with trade and craft guilds in Muslim societies. In Western Asia, Gibb emphasizes the importance of the Sufi’s more mystical interpretation of the religion as being practically synonymous with the profession of Islam.
https://doi.org/10.1017/9781009367509.005 Published online by Cambridge University Press
The Institutional Legacy of Muslim Conquest
contemporary Muslim societies, the institutional legacy of centralized autocracy largely remains in territories conquered by Muslim armies. This emphasized qualification is an important scope condition: My argument pertains to territories that were exposed to Muslim military conquest. That is, contemporary states whose territory was conquered by Muslim armies should exhibit a “stock” of more autocratic institutions relative to nonconquered territories. Clarifying this scope condition generates two testable implications. First, it implies that conquered territories should be, on average, less democratic today compared to non-conquered territories; thus, offering a historical explanation for the democratic deficit in Chapter (see Figure .). Second, among Muslim-majority states today, those that did not experience Muslim military conquest (e.g., Malaysia) should not exhibit a stock of autocratic institutions. I evaluate the veracity of these implications in Chapter . Before that in Chapter , I use statistical analysis to establish that Muslim conquest caused conquered territories to become centralized autocracies and trace its persistence. Before turning to empirical evaluations, I should highlight what this chapter has not argued. I have not discussed whether the political institutional legacy of Muslim conquest was “efficient.” Several recent studies argue that off-shoots (or correlates) of the political equilibrium discussed in this chapter (e.g., sharia law, the ulema-state alliance) were inefficient (detrimental) for long-run socioeconomic development (Kuran , Kuru , Rubin ). Much of this analysis has centered on comparing the long-run divergence of the Middle East to Western Europe. With the exception of the discussion of absolutist rule in Europe and China, I have not compared the ‘conquest equilibrium’ in relation to the institutional configuration in other societies. Such a comparison is not necessary for my argument, as my primary aim in this chapter was to describe how Muslim military conquest generated a trajectory of centralized autocracy in conquered territories.
As I argue in Chapter , if this second conjecture is true it implies that the Islamic faith is not inherently anti-democratic. Rather, the correlation between Islam and autocracy is driven primarily by the institutional legacy of Muslim conquest. Drawing such an inference requires having a benchmark or reference group to compare Muslim societies to. In this regard, the effect of the ulema-state alliance and sharia are “post-treatment” effects of Muslim conquest. Timur Kuran () argues that aspects of sharia law held back economic innovation and development in the Middle East and contributed to the region’s “long divergence” relative to Western Europe. Focusing on the ulema, both Jared Rubin () and Ahmet Kuru () argue that the alliance between clerics and the state stunted indigenous innovation and the adoption of new ideas from abroad, which, in turn may have contributed to lower levels of socioeconomic development. The central implication from these scholars is that Islamic law and their propagating agents (clerics) have been inefficient for economic development in the contemporary Middle East.
https://doi.org/10.1017/9781009367509.005 Published online by Cambridge University Press
The Conquest Equilibrium
Drawing inferences about whether the institutional trajectory in conquered societies has been efficient is not so clear and is largely based on who has benefited from the conquest equilibrium. Normatively, if one considers dictatorship negatively (i.e., from the perspective of aggregate political welfare), then the autocratic trajectory stemming from Muslim military conquest may be deemed inefficient (non-welfare maximizing). Yet, from the perspective of rulers, the persistence of the conquest equilibrium implies that it has been efficient in preserving the political power of a small group of elites (e.g., the dictator, his military, the clerics) for a long time. In Chapter , I provide one potential evaluation for the political efficiency of the conquest equilibrium as it pertains to political survival of sovereigns during the medieval period.
https://doi.org/10.1017/9781009367509.005 Published online by Cambridge University Press
Conquest Fostered Autocratization
This chapter seeks to quantify the causal effect of Muslim conquest on the emergence of a classical Islamic equilibrium of centralized dictatorship in conquered territories during the premodern era (i.e., prior to CE). While the process of Muslim conquest varied across societies – as described in the previous chapter – the empirical analysis in this chapter probes the common institutional features associated with Muslim conquest across the various types of political regimes and empires during and after the conquest period. This is implemented by estimating the “average” effect associated with the territorial expansion of Islam via military conquest on political centralization in conquered societies (relative to non-conquered societies). The empirics in this chapter evaluate the historical narrative from the previous chapter. That account generated a sequence of testable implications linking Muslim military conquest to various political outcomes in conquered territories (as depicted in Figure .). First, conquest introduced institutions that consolidated political (absolutist) authority in subjugated territories during the initial period of Muslim conquest. Second, these more “autocratized” structures remained intact thereafter, particularly under Mongol/Turkic and Ottoman rule. Third, the prevalence of centralized autocratic institutions affected the European colonial strategy: It incentivized the use of indirect rule by colonial powers and concomitantly reduced European migration to territories conquered by Muslim armies (relative to non-conquered territories). Fourth, upon their independence, Muslim countries largely maintained their autocratic structures, institutions that remain today. In this chapter, I provide quantitative evidence examining the initial two steps. Chapter will examine the remaining two steps related to the persistence of autocratic institutions during and after European colonialism.
Some of this chapter’s analysis is drawn from Ahmed ().
https://doi.org/10.1017/9781009367509.006 Published online by Cambridge University Press
Conquest Fostered Autocratization
Evaluating whether Muslim conquest consolidated absolutist political authority in conquered territories – a process I call autocratization – is no easy task. It confronts two challenges: measurement and causal identification. The first necessitates delineating an objective and transparent way to “measure” military conquest, particularly across territories that did not correspond to modern state boundaries. A related concern centers on measuring (quantifying) the governance structure – and its evolution – for modern states that did not exist over a years ago. Assuming one can overcome these issues with measurement, the second challenge concerns establishing robust cause and effect relationships (i.e., causal identification). Here, issues with endogeneity are worrisome. This includes the possibility of reverse causality (e.g., certain preconquest conditions might favor Muslim conquest) or omitted variables that affect both the likelihood of conquest and the trajectory of political development (e.g., agricultural conditions). Failure to address these concerns with endogeneity can bias statistical inferences. With these challenges in mind, I proceed as follows. The next section describes how I objectively measure the key independent and dependent variables. For the former, I measure Muslim conquest on two dimensions: its geographic scope in terms of modern state boundaries and its duration. For the latter, I use a set of objective criteria to measure the evolution of political centralization in conquered and non-conquered territories from before Muslim conquest through to the present. In Section ., I describe a research design to identify causal effects. In particular, the panel structure of the data on political centralization (i.e., country by time period) and the timing of Muslim conquest allows me to employ a difference-in-differences (DD) research design. Doing so, allows me to estimate the causal effect of Muslim conquest on political centralization in conquered territories relative to non-conquered territories. This empirical strategy also allows me to evaluate and discount other competing historical explanations, such as the role of agriculture, international trade, and the level of state centralization prior to the onset of Muslim conquest. I complement these findings by investigating a correlate of autocratization: the political survival of reigning autocrats. Here, I examine the effect of Muslim conquest on the political survival of Muslim monarchs, particularly in relation to their counterparts in medieval (Christian) Europe. The analysis reveals that Muslim monarchs enjoyed longer tenures in power during the period of Muslim expansion. Together, the statistical evidence provides robust evidence that Muslim conquest helped autocratize conquered territories during the conquest period.
Econometrically, endogeneity between two variables (e.g., conquest and state centralization) can arise from measurement error, reverse causality and/or omitted variables. For the latter, Stephen Haber and Victor Menaldo (b) argue that territories with rain-fed agriculture were more conducive to developing institutions that protect property rights.
https://doi.org/10.1017/9781009367509.006 Published online by Cambridge University Press
The Institutional Legacy of Muslim Conquest
I then evaluate the underlying channel associated with mamluk institutions. First, I evaluate how the presence of mamluk institutions affected political centralization. Here, I show that territories where Islam spread by but where mamluk institutions were not present (i.e., in sub-Saharan Africa) did not autocratize. This is in sharp contrast to the main findings in this chapter: Territories exposed to mamluk institutions (e.g., the Middle East, Central Asia) autocratized during the period of Muslim conquest. Second, I present evidence consistent with the argument that agriculturally productive territories improved the efficacy of the iqta system of land-grants to compensate mamluk soldiers; and this in turn is associated with a greater consolidation of political authority. Finally, the flexibility of the DD research design also allows me to trace the effect of Muslim conquest on political centralization in relation to subsequent “interventions” in conquered territories (i.e., Mongol/Turkic, Ottoman, and European colonial rule). This analysis reveals that subsequent interventions did not undo the Islamic equilibrium in conquered territories.
. Evaluating the impact of Muslim conquest on political authority requires an objective and transparent methodology. In this section, I describe how I conceptualize and measure this chapter’s key independent and dependent variables: Muslim conquest and political authority, respectively.
A Muslim Conquest Attempting to measure Muslim conquest – an event that occurred nearly years ago – must address two issues: () defining its geographic scope and () bounding its duration. For the first issue, Chapter described the expansion of the Islamic empire through military conquest and referred to the conquered territories using contemporary national boundaries. These boundaries, of course, did not exist in the seventh and eighth centuries. While all of modern Saudi Arabia was conquered by Muslim armies, this was not necessarily the case in other areas of the Islamic empire, such as modern-day India which was partially conquered. Resolving this issue also requires addressing the second about the duration of Muslim conquest. While the start date of Muslim conquest can be pinned down to the first few years after the death of Prophet Muhammad (circa ), the end date is less clear. For instance, the reconquista effectively ended Muslim rule in Spain in , yet in other parts of the Islamic empire the historical record is less accurate for the effective termination (if at all) of Muslim rule. As
For example, some reigning monarchies in the Middle East (e.g., Jordan, Saudi Arabia) assert a hereditary connection to Prophet Muhammad and a continuation of the caliphate.
https://doi.org/10.1017/9781009367509.006 Published online by Cambridge University Press
Conquest Fostered Autocratization
I described in the previous chapter, in most cases, subsequent rulers (e.g., Sejluks, Mongols, European colonizers) built on the existing institutions and maintained the prevailing ruling coalitions (e.g., the prominence of the ulema and military elite) to govern Muslim territories. For my purposes, I use the year to bound (end) the duration of the Muslim rule. By , mamluk institutions – the use of mamluks and iqta (as described in the previous chapter) – had been implemented across conquered territories. Moreover, after many territories experienced reconquest (e.g., by Christians in Spain and during the Crusades) and foreign invasions (e.g., Mongol/Turkic incursions from Central Asia). Thus, to avoid introducing the influence of other empires/ rulers, I cleanly bound the period of Muslim conquest from to . By demarking as a clear endpoint, I am in a stronger position to accurately measure the geographic expansion of Muslim military conquest. I do so by leveraging maps describing the spread of Islam over time compiled by Hugh Kennedy (). From his compilation, I use the map that details the geographic exposure of Islam by (see Figure .) and then digitize it on a map with the boundaries of contemporary nation states. I then calculate the share of a contemporary country’s landmass that was exposed to Islam by 1100 ðCi Þ. Ci ranges from (not exposed to Islam) to (fully exposed to Islam). Critically, I then differentiate territories where Islam spread via military conquest (e.g., the Middle East, North Africa, Spain, Central and South Asia) versus those where Islam spread through other means, such as trade and clerics (e.g., Sub-Saharan Africa). Territories in the former category experienced mamluk institutions, while territories in the latter did not. I will refer to these countries in the first category as mamluk territories ðMi Þ. Table . lists the countries by their exposure to Islam by . Panel A shows all countries where Islam had spread to some fraction of their territory by via military conquest, Mi or though non-conquest, Ni. Panel B lists all countries where Islam had not reached its territory by . In both panels, I denote all contemporary Muslim majority countries with an asterisk (*). Panel A shows by , Muslim armies had conquered territory in thirty-three contemporary states, ranging from a trivial share in China (. percent) and Kazakhstan ( percent) to significant shares in Azerbaijan ( percent), Pakistan ( percent), Spain ( percent), Turkey ( percent), and all the Arabian Peninsula and Mesopotamia (e.g., Iraq, Iran). Within these mamluk territories, the Islamic empire controlled around percent of their modern country’s landmass, on average.
This second step differentiates my measure of conquest from Chaney (). Chaney’s measure of “Arab conquest” refers to Muslim majority countries (in the contemporary period) that were exposed to Islam. This therefore excludes Spain. However, it does include many African countries where Islam spread via trade and clerics but not military conquest. I am careful to only include territories where Islam spread through military conquest and where mamluk institutions were subsequently implemented.
https://doi.org/10.1017/9781009367509.006 Published online by Cambridge University Press
FIGURE
The Institutional Legacy of Muslim Conquest
. The spread of Islam by .
Source: Kennedy ()
Panel A also shows that in twelve countries, Islam had spread via nonmilitary means (denoted with N i > 0), covering about percent of a modern country’s territory. Unsurprisingly, the overwhelming share of countries with mamluk institutions are Muslim majority today (twenty-four countries). However, a few countries are not, such as Nepal, Portugal, and Spain. Examining the sample of non-exposed countries, panel B shows that (out of ) are not Muslim-majority; although there are some notable exceptions, such as Bangladesh, Indonesia, and Malaysia. These countries became majority Muslim in the contemporary period either through internal migration prior to their independence (e.g., Bangladesh) or the adoption of Islam through trade and clerics (e.g., Indonesia, Malaysia). B Autocratization: Premodern State Centralization I now turn to the chapter’s main dependent variable. To evaluate whether Muslim conquest – and its associated mamluk institutions – consolidated political authority requires an objective measure quantifying governing structures across both conquered and non-conquered territories. This necessitates articulating what political authority entails, particularly in the premodern period
https://doi.org/10.1017/9781009367509.006 Published online by Cambridge University Press
https://doi.org/10.1017/9781009367509.006 Published online by Cambridge University Press
TABLE
. Territories “exposed” to Islam by Panel A: Percent of territory exposed to Islam by Via military conquest (Mi) Ci
Algeria* Armenia Azerbaijan* Bahrain* China Egypt* Georgia India Iran* Iraq* Jordan* Kazakhstan Kuwait* Kyrgyzstan* Lebanon* Libya* Morocco*
Via non-conquest (Ni) Ci
. . . . .E- . . . . . . . . . . . .
Nepal Oman* Pakistan* Portugal Qatar* Russia Saudi Arabia* Spain Syria* Tajikistan* Tunisia* Turkey* Turkmenistan* UAE* Uzbekistan* Yemen*
. . . . . .E- . . . . . . . . . .
Ci Chad* Eritrea* Ethiopia Kenya Mali* Mauritania* Mozambique Niger* Senegal* Sudan* Somalia* Tanzania
. . . . . . . . . . . .
Nicaragua Nigeria Norway
Switzerland Thailand Togo
Panel B: Countries not exposed to Islam by 1100 ðCi ¼ 0Þ Albania* Angola Argentina
Central Afr. Rep. Chile Colombia
Gambia* Germany Ghana
Korea North Laos Latvia
(continued)
https://doi.org/10.1017/9781009367509.006 Published online by Cambridge University Press
TABLE
. (continued) Panel B: Countries not exposed to Islam by 1100 ðCi ¼ 0Þ
Australia Austria Bangladesh* Belarus Belgium Benin* Bolivia Botswana Brazil Bulgaria Burkina Faso* Burundi Cambodia Cameroon Canada
Comoros* Rep. of Congo Costa Rica Croatia Cuba Cyprus Czech Rep. Denmark Ecuador El Salvador Estonia Fiji Finland France Gabon
Greece Guatemala Guinea* G. Bissau* Guyana Haiti Honduras Hungary Indonesia* Ireland Italy Ivory Coast Jamaica Japan Korea South
Lesotho Liberia Lithuania Macedonia* Madagascar Malawi Malaysia* Mauritius Mexico Moldova Mongolia Myanmar Namibia Netherlands New Zealand
Notes: By definition, for all countries in panel B Ci ¼ 0. * = denotes Muslim-majority territory by .
Panama P.New Guinea Paraguay Peru Philippines Poland Rwanda Sierra Leone Singapore Slovak Rep. Slovenia South Africa Sri Lanka Swaziland Sweden
Trinidad Uganda Ukraine UK USA Uruguay Venezuela Vietnam Zambia Zimbabwe
Conquest Fostered Autocratization
(i.e., before ). To the best of my knowledge, there are no direct measures of autocracy in the premodern era for a large set of territories, and that correspond to modern-day boundaries. Instead, I draw on existing measures of state centralization as a proxy for autocratic governance in the premodern period. Among contemporary social scientists, discussions about political authority most always relate to the concept of state centralization (e.g., Dincecco , Mann , Tilly ). While conceptually quite broad, state centralization – often synonymous with state capacity – refers to the “government’s ability to accomplish its intended policy goals” (Dincecco , ). Importantly, in the premodern period this corresponded to “despotic power” (Mann ) in which a dominant class exerted its will over society. Citing Max Weber, Michael Mann (, –) makes explicit that despotic power necessitates the existence of a “state” with some degree of authority/control by a dominant political actor(s) within its territory to carry out its activities. To capture this essential feature of despotic power, I leverage the State Antiquity Index (Putterman ). The index has been widely utilized by scholars to probe a variety of research agendas. For instance, Valerie Brockstette, Areendam Chanda, and Louis Putterman () employ the index to document the co-evolution of “states and markets.” Motivated by a different research question, Andreas Wimmer () leverages the index to evaluate the relationship between ethnic fragmentation, public goods provision, and the historical legacy of weak states over time. In a couple of prominent works, Jacob Hariri (, ) uses the index to measure premodern state capacity and its impact on contemporary political institutions. For my purposes, the index is attractive on three key dimensions. First, it is panel data from year onward. Second, it uses countries with their contemporary national boundaries as the unit of analysis. Third, the index has an objective and transparent coding criterion. The index is based on three questions (components, z z z) that ascertain the level of a political unit’s organization, its primary location (foreign or locally based), and extent of its geographic control. Using information from the historical accounts of each country in the Encyclopedia Britannica, for each period of fifty years, the coders asked three questions: . Is there a government above the tribal level? (z receives point if yes, . point if the government can be best described as a paramount chiefdom, and points if no government is present); . Is this government foreign or locally based? (z is if the rule is locally based, . if externally based (e.g., the country is a colony), and . for local government with substantial foreign oversight);
This contrasts with “infrastructural power” in which the state exerts it power through society and implies a more cooperative arrangement between the ruler and his subjects. States in the modern era exhibit greater infrastructural rather than despotic power (Mann , –).
https://doi.org/10.1017/9781009367509.006 Published online by Cambridge University Press
Change in state centralization from 600 to 1150 –50 –25 0 25 50
The Institutional Legacy of Muslim Conquest
Algeria Georgia Azerbaijan Uzbekistan Morocco Egypt Pakistan Libya Tunisia Iraq Tajikistan Syria Yemen Turkmenistan Israel Arabia Jordan Saudi
China Russia Kazakhstan Kyrgyzstan IndiaNepal
Spain
Turkey Portugal
0
.2
.4 .6 .8 Percent territory conquered by Muslim armies
Non-Muslim majority Fitted line for Muslim-majority countrie s FIGURE
Iran Armenia
1
Muslim majority
. Change in state centralization during Muslim conquest
. How much of the territory of the modern country was ruled by this government? (z is based on the proportion of territory under some rule: ( percent), . (– percent), and . (under percent)). These three scores are multiplied by one another and then by . Thus, the index ranges from to , where a higher value corresponds to greater centralization of political authority. Increases in the first (z) and third (z) components (i.e., movement from tribal to national government, greater territorial control) increase the overall centralization index, while increase in the second (z; greater control by a foreign power) decreases the index. The latter implies that by construction, the aggregate index does not mechanically code Muslim conquest to necessarily increase the index. This is visible in the raw data. Figure . plots the change in state centralization from to for the cross section of countries that were fully or partially conquered by Muslim armies. While, on average, the typical conquered country became more centralized during this period (as suggested by the fitted line’s positive slope), there were some territories that experienced a decline, such as modern-day Armenia, Iran, and Turkey. For example, after fifteen years of fighting, Muslim armies
The positive association and fitted line support this chapter’s main empirical finding in the cross section of countries. The figure suggests that countries where Muslim armies controlled a greater
https://doi.org/10.1017/9781009367509.006 Published online by Cambridge University Press
Conquest Fostered Autocratization
finally defeated the Sassanid Empire in bringing much of modern Iran and Iraq under Caliph Umar’s rule (at the time, based in Medina). Since this represented a shift from indigenous rule to a mixture of foreign and local based rule, the second component of the index shifts from a value of to ., which also decreases the aggregate index. Thus, the aggregate index does not automatically upward bias the effect of Muslim conquest. Nevertheless, I present results in Section . where I strive to partial out the “foreign vs. local rule” component (z) from the aggregate index by examining the effect of Muslim conquest on the “indigenous” sources of state centralization (z and z). C Interpretation It is worth reiterating that without a democratic alternative in the premodern era, greater state centralization implied the consolidation of absolutist sovereign rule (Anderson , Mann ). Josiah Ober (), for example, characterizes the “premodern normal” as a condition of political domination by an autocrat, who often claimed divine authorization for his rule. Consequently, these rulers were not subject to constraints imposed by ‘mortals.’ In their sweeping account of political order from the earliest recorded human civilizations, Douglass North, John Wallis, and Barry Weingast () describe the “natural state” as autocratic, nepotistic, and prone to generating violence. Thus, I interpret higher values of the state centralization index in the premodern era to imply a strengthening of an autocrat’s political authority.
. A Identification Strategy Specification While the bivariate correlation in Figure . suggests a positive relationship between Muslim conquest and state centralization (autocratization), the relationship is not necessarily causal. Concerns with endogeneity, particularly reverse causality and omitted variables, complicate identifying a causal relationship. The availability of panel data under certain conditions (as I validate shortly) empower me to leverage a quasi-experimental approach to gauge
share of territory by experienced a greater increase in state centralization from . As expected, this positive association is strongest among contemporary Muslim societies (marked with “x”) and less so among non-Muslim ones (marked with “o”). However, on average, the change in state centralization during the conquest period is positive (and statistically significant) among Muslim societies. As both Crone () and Kennedy () describe in detail, in the first few centuries of the conquest, Muslim armies resided in garrison towns outside conquered cities and co-governed with the locals.
https://doi.org/10.1017/9781009367509.006 Published online by Cambridge University Press
The Institutional Legacy of Muslim Conquest
whether Muslim conquest caused autocratization in conquered territories. Specifically, I trace the trajectory of state centralization associated with Muslim conquest using a difference-in-differences (DD) setup of the form: SCit ¼ α þ θðMi Pt Þ þ δXit þ Fi þ Y t þ εit
ð4:1Þ
where SCit is an increasing measure of state centralization for country i in a -year time interval, t. Mi measures the percentage of a country’s territory conquered by Muslim armies, Pt is a dummy variable equal to during the period of Muslim conquest and otherwise. Xit is a vector of controls (preconquest state centralization, year trends, etc.), which will be described as the analysis proceeds, and εit is the error term. Given the panel structure of the data (i.e., country by -year time period), I conservatively cluster the standard errors at the country level to account for any serial correlation. In Equation ., the variable of interest is the interaction of Muslim conquest ðMi Þ and the period of Muslim conquest ðPt Þ. In the analysis to follow, I will refer to this interaction term as “Muslim expansion,” where a positive and statistically significant coefficient estimate implies Muslim conquest helped increase a conquered territory’s political centralization during the conquest period relative to countries that were not conquered by Muslim armies. An attractive feature of leveraging panel data is the possibility of controlling for unobserved unit-level and temporal heterogeneity. As such, the preferred DD specification includes a vector of country ðFi Þ and period fixed effects (for each half-century, Y t ). The inclusion of country fixed effects is noteworthy in two regards. First, they capture all time-invariant country characteristics (e.g., geography, suitability to rain-fed agriculture), especially those that predate Muslim conquest (timing since the Neolithic Revolution, preconquest state centralization, etc.) that might affect state centralization. Second, the coefficient estimates from Equation . will capture the within-country variation explained by the independent variables. Thus, the effect of Muslim conquest can be traced within the same country over time. Sample An important consideration is the appropriate counterfactual sample of territories. In the DD setup, treated units are those that experienced different “intensities” of Muslim conquest (as measured by the percentage of their
DD is a quasi-experimental design used to estimate the effect of a specific intervention or treatment (i.e., Muslim military conquest) by comparing the changes in outcomes over time between a population that is “treated” and a population that is not (the control group). The coefficients on the constituent terms are informative as well. The coefficient on Islamic conquest ðMi Þ estimates the average effect of conquest on state centralization across all time periods. Thus, it measures the persistent effect of conquest on state centralization. In contrast, the coefficient on conquest period ðPt Þ estimates the trajectory of state centralization across nonconquered countries during the period of Islamic conquest. A negative coefficient implies that non-conquered countries became less centralized between and .
https://doi.org/10.1017/9781009367509.006 Published online by Cambridge University Press
Conquest Fostered Autocratization
modern-day territory conquered by Muslim armies). The counterfactual (or control) group of countries comprise territories that did not experience any Muslim conquest (i.e., N i > 0, Ci ¼ 0). I consider three sets of counterfactual countries. First, I begin my analysis with a global sample of countries. I then use history as a guide: Muslim conquest entailed land-based military expansion (starting from the Arabian Peninsula), although these forces were able to traverse short distances by sea, such as the invasion of the Iberian Peninsula from Morocco (Kennedy ). Thus, the second sample of counterfactual countries span Eurasia and Africa. Finally, I consider a narrower set of “matched” counterfactuals comprised of those countries that () had similar levels of preconquest state centralization (as conquered territories) and () could have been conquered by Muslim armies but had not been. This includes territories that were on the periphery of the Islamic empire that Muslim armies could have feasibly been invaded and possibly conquered, but were not (e.g., Cyprus, Greece, Italy, Serbia). B Parallel Trends While DD is a powerful research design in non-experimental settings, several conditions need to be satisfied for a causal interpretation. In particular, the causal interpretation of Equation . is bolstered if the parallel trends assumption is not violated: In the absence of the treatment (Muslim military conquest), the difference in state centralization between the treatment (conquered Muslim territories) and control (non-conquered territories) is constant over time. While there are no formal tests per se for this assumption, one strategy is to plot the underlying trends in the dependent variable (state centralization) across the treatment and control groups. In Figure ., I do so by plotting the average within-country level of state centralization in countries completely or partially conquered by Muslim armies (“Conquest”) and those not conquered at all (“Non-conquest”). Figure . highlights three important patterns that support the use of DD. First, the average levels of state centralization are near identical between conquest and non-conquest countries prior to the start of Muslim expansion (circa ). Second, the trends in state centralization between conquest and non-conquest countries are parallel prior to the start of expansion. This interpretation is supported with more formal statistical analysis which controls for pre- and post-treatment trends (Autor ) and country-specific time trends (Earle and Gehlbach ). This analysis reveals that the pre- and posttreatment terms are not statistically significant, while the treatment effect
The within-country level of state centralization purges the contribution of time-invariant country characteristics (e.g., geography, climate), especially those that predate (e.g., role of Roman civilization, transition to the Neolithic Revolution) and corresponds to the preferred fixed effects DD specification.
https://doi.org/10.1017/9781009367509.006 Published online by Cambridge University Press
0
State centralization (within country) 5 10 15
20
The Institutional Legacy of Muslim Conquest
0
300
600
900
1200
1500
Year Conquest FIGURE
Non-conquest
. State centralization in the premodern period
ðMi Pt Þ is. Notably, the statistical insignificance of the pre-treatment term suggests that the underlying trends between conquered and non-conquered societies did not statistically differ prior to the onset of Muslim conquest. Moreover, the treatment effect remains robust in specifications that plausibly account for counterfactual country-period trends (Earle and Gehlbach ). Together, these findings suggest the parallel trends assumption is unlikely to be violated. Third, starting around , the average level of state centralization in conquest countries starts to increase at a faster rate than in non-conquest countries. This growing divergence – particularly between and – is consistent with a positive interaction term ðθÞ in Equation ..
. A Main Results Table . presents corroborating statistical evidence for the divergence in state centralization during Muslim conquest (corresponding to Equation .). I start with the broadest possible sample of counterfactual territories. In columns –, the counterfactual is a global sample of countries that did not experience any Muslim conquest. For example, this includes countries in the Americas (e.g., Canada, Peru) and East Asia (e.g., Japan) where Muslim armies did not reach.
https://doi.org/10.1017/9781009367509.006 Published online by Cambridge University Press
https://doi.org/10.1017/9781009367509.006 Published online by Cambridge University Press
TABLE
. State centralization during Muslim expansion
Dependent variable Sample
Muslim expansion ðMi Pt Þ Conquered territory (Mi) Conquest period (Pt)
State centralization index Global
Country fixed effects Period fixed effects No. observations Countries R-squared
Global
Eurasia
Global:
Global:
Exclude MENA
Exclude Mi ¼ 1
()
()
()
()
()
()
. (.)*** . (.)*** −. (.)***
. (.)*** −. (.) −. (.)*** . (.)*** . (.) . (.)***
. (.)***
. (.)**
. (.)***
. (.)***
. (.)***
. (.)***
. (.)***
. (.)***
Yes Yes .
Yes Yes .
Yes Yes .
Yes Yes .
State centralization in – CE MENA Constant
Global
. (.)***
.
.
Notes: Estimation via OLS. Robust standard errors, clustered by country reported in parentheses. *, **, *** = significant at , , and percent respectively.
The Institutional Legacy of Muslim Conquest
Columns and report estimates from specifications that do not include country and period fixed effects ðFi , Y t Þ, but do appropriately control for the interaction term’s constituent variables ðMi , Pt Þ. Column shows that Muslim expansion increased political centralization (autocratization) during the conquest period. For the typical conquered country (with Mi ¼ 0:77), Muslim expansion consolidated political authority in conquered territories by . index points relative to non-conquest countries. This is equivalent to moving . standard deviations in state centralization. Consistent with the fragmentation of political authority during the early Middle Ages (Strayer ), particularly in Europe, the coefficient on Pt ð¼ −4:52Þ implies that nonconquered countries experienced a decline in state centralization during the period of Muslim conquest. The effect of Muslim expansion remains robust and positive when controlling for a country’s preconquest level of state centralization and a dummy variable for the Middle and North Africa (MENA) region (column ), as well as in a specification with country and period fixed effects (column ). In column , the inclusion of country fixed effects accounts for the MENA dummy and level of preconquest state centralization, as well as other time-invariant country characteristics, such as a country’s timing since the Neolithic Revolution, suitability for rain-fed agriculture, preconquest trade routes, and presence of Roman settlements. Moreover, this specification traces the effect of Muslim conquest within the same country over the conquest period (relative to the nonexpansion period). Across these specifications, the effect of Muslim expansion on state centralization is quite stable (with coefficient estimates equal to .) and remains highly statistically significant. While there is no a priori reason to limit the sample of counterfactual countries, skeptics may contend that including territories that were implausible to be conquered by Muslim armies might bias the statistical inferences. To allay this concern, in column , I trim the counterfactual sample to include countries in Eurasia that were (plausibly) within the geographic reach of Muslim armies but were not invaded (e.g., Germany, Burma). In this specification, the coefficient on Muslim expansion declines slightly in magnitude but remains positive and statistically significant. Together, the estimated effects in columns – suggest that relative to non-conquest countries, the modal conquered territory (with Ci ¼ 0:68) experienced around a index point increase in state centralization per -year interval. These estimated effects imply Muslim expansion helped transition the modal conquered society from a condition of “no governance” to one with a more centralized political entity that governed a larger area of territory. In some conquered societies, this entailed direct rule by the
This is the total marginal effect for the modal country: 0:77 ð13:75 þ 8:75Þ. Table . shows this more directly. The estimated effects imply that Muslim conquest increased a conquered society’s “political organization” (panel A) and “geographic control” (panel B).
https://doi.org/10.1017/9781009367509.006 Published online by Cambridge University Press
Conquest Fostered Autocratization
caliphate (e.g., Abbassids in modern-day Iraq), while in others this was indirect rule over local tribes (e.g., in parts of Central Asia). These results are not driven by countries from any specific region. Columns and show that Muslim expansion exhibits a positive, robust effect on state centralization in samples that separately drop territories from the MENA region and those that were fully conquered by Muslim armies ðMi ¼ 1Þ. Interestingly, excluding these countries increases the estimated effect: The coefficient estimates are nearly double in magnitude relative to the baseline estimates in columns –. The finding in column is noteworthy as it implies that Muslim expansion helped centralize absolutist rule in territories outside of the MENA region (modern-day Pakistan, Uzbekistan, etc.) and is therefore neither an artifact of arguments related to Middle Eastern exceptionalism nor of the higher level of bureaucratic capacity (e.g., tax collection) in some preconquest territories (e.g., areas closer to the Roman Empire or within Mesopotamian empires from antiquity). B Robustness Alternate Measures of Muslim Conquest A benefit to digitizing Hugh Kennedy’s () maps of Muslim conquest is the ability to precisely quantify the share of a territory conquered by Muslim armies from to . That said, if the map’s geographic demarcations are imprecise, my measure of Muslim expansion ðMi Pt Þ could bias my statistical inferences. While this possible type of measurement error is likely to be “at random” – and is likely therefore to attenuate the estimated effects toward zero – skeptics may worry about potential bias nevertheless. To allay this concern, I consider three alternate measures of Muslim conquest that do not rely on the precision gleaned from the digitization process: Dummy variables equal to if any part of territory was conquered by Muslim armies (“any Mamluk”), at least percent of territory was conquered by Muslim armies (“Half Mamluk”), and percent of territory was conquered by Muslim armies (“Full Mamluk”). I then interact these dummy variables with conquest period (Pt). Columns – in Table . show that with any of these alternate measures of conquest, Muslim expansion exhibits a robust, positive effect on state centralization. Interestingly, the estimated effect associated with territories fully conquered by Muslim armies (column ) is smaller in magnitude than those associated with partially conquered territories (columns and ). This finding is consistent with the estimated effects in columns and in Table . and
These dummy variables equal zero for country-period observations that do meet the criteria (for coding values at ). Territories that were fully conquered by Muslim armies tend to be in the modern Middle East, particularly the Arabian Peninsula. In contrast, “partially” conquered territories tend to be located in Central and South Asia, such as modern-day Pakistan and Uzbekistan.
https://doi.org/10.1017/9781009367509.006 Published online by Cambridge University Press
https://doi.org/10.1017/9781009367509.006 Published online by Cambridge University Press
TABLE
. Robustness of state centralization during Muslim expansion
Dependent variable
State centralization index Incidence of conquest ()
Any Mamluk Conquest
Region and country trends
()
()
()
()
()
. (.)*** . (.)*** −. (.)***
. (.)*** . (.)*** −. (.)*** . (.)*** −. (.)
. (.)***
. (.)***
Half Mamluk Conquest
. (.)***
Full Mamluk Conquest
. (.)*
Muslim expansion (Ci Mt) Conquered territory (Mi) Conquest period (Pt) Period trend Constant
. (.)***
. (.)***
. (.)***
Country fixed effects Period fixed effects Continent fixed effects Continent Trend Country Trend No. observations Countries R-squared
Yes Yes
Yes Yes
Yes Yes
.
.
.
. (.)**
−. (.)*** . (.)*** . (.)*** Yes
Yes
Yes Yes
.
.
Yes .
Notes: Estimation via OLS for a global sample of countries. Robust standard errors, clustered by country reported in parentheses. *, **, *** = significant at , , and percent respectively.
Conquest Fostered Autocratization
implies that Muslim conquest significantly autocratized societies outside of the Middle East. Furthermore, it reinforces the inference that autocratization was not an artifact of stronger preexisting institutional structures prior to conquest (that was prevalent in several Middle Eastern territories but not in Central and South Asia). Unobserved Heterogeneity and Alternate Specifications The results also remain robust in specifications that control for additional spatial and temporal heterogeneity with continent fixed effects, continent period time trends, and most stringently with country-specific time trends (columns – in Table .). The inclusion of continent fixed effects, for example, accounts for potential geographic conditions (e.g., climate) that might affect a country’s subsequent political and economic development (Diamond ). In contrast, interactions with a time trend account for different trajectories of state centralization that vary across regions (column ) and countries (column ), such as those between the Middle Eastern and European states. The key finding is also robust across alternate econometric specifications, including those that control for lagged dependent variables (to mitigate concerns with autocorrelation and dynamic persistence), moving averages of state centralization (over and years) and are estimated via Tobit regression (since state centralization is truncated at and ).
. Skeptics may argue the findings are spurious and are driven by both observable and unobservable characteristics that made subsequently conquered territories more predisposed to be autocratized. Controlling for region and particularly country-specific time trends is one strategy to address this worry with unobservables (see Table ., columns and ). Nevertheless, in the next three subsections I present additional evidence to allay these concerns. First, I restrict my analysis to a matched group of counterfactual territories. I then evaluate (and discount) three prominent competing historical narratives: the role of agriculture, trade, and preconquest state centralization (e.g., from previous empires). Finally, I present evidence that focuses on the ‘indigenous’ components of state centralization. A Matched Counterfactuals While Table . estimates the effect of Muslim conquest on state centralization relative to different samples of non-treated (counterfactual) territories – e.g., spanning the entire globe, Eurasia only, and excluding countries in the MENA region – skeptics may worry the samples of counterfactuals territories are too broad and are substantively different from conquered territories. To allay this particular concern, Table . reports the impact of Muslim expansion on state centralization in a sample with a matched counterfactual (non-conquest) group
https://doi.org/10.1017/9781009367509.006 Published online by Cambridge University Press
https://doi.org/10.1017/9781009367509.006 Published online by Cambridge University Press
TABLE
. Muslim conquest and autocratization: Matched counterfactuals and competing explanations
Dependent variable
Muslim expansion (Mi Pt) Conquest (Mi) Conquest period (Pt)
State centralization index ()
()
()
()
()
()
()
. (.)*** . (.) −. (.)***
. (.)***
. (.)***
. (.)***
. (.)***
. (.)***
. (..)**
. (.) −. (.) . (.) . (.) . (.)***
Log rain Pt
. (.) . (.)
Cereal production Pt Trade route Pt
−. (.) . (.)***
. (.)***
. (.)***
. (.)***
. (.) . (.)***
Yes Yes
Yes Yes
Yes Yes
Yes Yes
Yes Yes
Yes Yes
.
.
.
.
.
.
SC in Pt Constant
. (.)***
Country fixed effects Period fixed effects Observations Countries R-squared
.
Notes: Estimation via OLS for a global sample of countries. *, **, *** = significant at , , and percent respectively.
Conquest Fostered Autocratization
of countries based on preconquest characteristics. Specifically, in these regressions the sample of counterfactual countries are limited to those that fall within a -point band of the median conquered country on its level of state centralization directly prior to the onset of Muslim conquest. Many of these counterfactual territories were on the immediate periphery of the Islamic empire and were feasible targets of Muslim armies but were not actually conquered (e.g., Albania, Cyprus, Greece, Italy, and Serbia). With this matched sample, I reestimate the baseline specification with and without fixed effects. Columns and in Table . show Muslim expansion remains a positive and statistically significant determinant of state centralization. The estimated effect ð¼ 8:12Þ is similar in magnitude to the main results in Table .. An alternative matching approach is to construct a synthetic control (Abadie, Diamond, and Hainmueller ). Figure . contrasts the trajectory of the typical (average) conquered country (i.e., the “treated unit”) against a synthetic control compiled from a global sample of non-conquest territories. Given the fifty-year interval structure of the data, the treatment period begins in (even though conquest commenced on the Arabian Peninsula a couple of years after , thus accounting for the slight uptick in centralization prior to ). The figure suggests that during the treatment (conquest) period, the typical conquered territory became more politically centralized, while the synthetic control did not. Reassuringly, each counterfactual exercise helps answer the question what would have been the level of state centralization in countries “similar” to conquered territories had conquest not occurred. The results suggest state centralization would not have increased.
B Evaluating Competing Historical Explanations It is plausible that Muslim expansion might be capturing the effects of other historical factors that made conquered territories more prone to autocratization. If this is the case, then the causal inferences associated with Muslim expansion may be misleading. Agriculture One such competing explanation emphasizes the role of rain-fed agriculture. Stephen Haber and Victor Menaldo () posit that certain types of rain-fed
Thus, the sample includes all countries that experienced Muslim conquest plus those that were not conquered but had a “similar” level of state centralization (SC) in the year . The median (subsequently) conquered country had SC ¼ 25 in the year . The synthetic control is “predicted” from values of state centralization in the preconquest period. The treated unit consists of the average value (per period) across all conquered territories. A related explanation stems from a country’s transition to settled agriculture (associated with the Neolithic Revolution). The main results remain robust when controlling for a country’s timing since the Neolithic Revolution (from Hariri ) interacted with the conquest period.
https://doi.org/10.1017/9781009367509.006 Published online by Cambridge University Press
30
The Institutional Legacy of Muslim Conquest
15
State centralization 20 25
Conquest begins in 650 -->
0
200
400
600
800
1000
1200
1400
Year Conquered
Synthetic conquered
. State centralization, with a “typical” (treated) and synthetic conquered territory Notes: This figure follows the procedure described in Abadie, Diamond and Hainmueller (). It plots the level of state centralization for the average (“typical”) level of state centralization in conquered territories (the treated unit) and a synthetic control comprised by a weighted sum of non-conquered territories. The synthetic control is based on values of state centralization prior to (i.e., before the onset of the treatment period in ). FIGURE
agriculture affected their societies’ historical paths of institutional development; some proved favorable to stable democracy. They argue that in zones of moderate rainfall, unlike arid and tropical zones, it was possible to grow crops that were highly storable and that exhibited modest scale economies in production, which created broad-based incentives to trade, protect property rights, and make inter-generational investments in human capital. This led to the emergence of stable democracies in these moderate zones. Haber and Menaldo’s conjecture suggests that many territories conquered by Muslim armies were not suitable for the development of rain-fed agriculture and thus more likely to follow nondemocratic trajectories. To evaluate whether rain-fed agriculture differentially affected the trajectory of state centralization, I interact a country’s average rainfall and cereal production (which is an alternate measure for a country’s capacity for settled agriculture that is not related to rainfall) with the conquest period ðPt Þ and include it
https://doi.org/10.1017/9781009367509.006 Published online by Cambridge University Press
Conquest Fostered Autocratization
as an additional covariate in the baseline DD specification. Columns and in Table . show that conditional on these additional interaction terms, Muslim expansion remains a robust determinant of state autocratization. While territories with greater rainfall and cereal production experienced slight increases in state centralization from to , the estimated effects are not statistically significant. Trade Another plausible historical explanation is the role of preexisting trade routes in facilitating the spread of Islam through both Muslim conquest and the potential gains from trade. Stelio Michalopoulos, Alireza Naghavi, and Giovanni Prarolo () provide statistical evidence that Islam spread to regions closer to preexisting trade routes in . Their argument centers on the potential economic gains that individuals experienced by converting to Islam, such as access to trade networks. As such, the trade explanation does not directly explain why new converts to Islam would adopt specific governance practices (e.g., mamluks, iqta) except through outright conquest which directly transplanted these institutions. Nevertheless, failure to account for preexisting trade routes may bias the estimates in Table .. The specification in column of Table . address this concern by controlling for the interaction of Pt with a country’s minimal distance to trade routes in (from Michalopoulos et al. ). The specification suggests societies farther from existing trade routes (e.g., those in sub-Saharan Africa) tend to experience less centralization during the conquest period (coefficient=−.), although the effect is not statistically significant. Importantly, Muslim expansion remains a robust determinant of state autocratization. Preconquest State Centralization More generally, I evaluate the argument that more centralized territories prior to the onset of Muslim conquest were more likely to continue centralizing irrespective of the institutions introduced via Muslim conquest (Stasavage ). The panel structure of the data allows me to directly account for each society’s level of preconquest state centralization. To do so, I include a country’s level of state centralization in ðSC600 Þ interacted with Pt as an additional covariate (column in Table .). In this specification, Muslim expansion remains positive and statistically, while SC600 Pt is not a robust determinant. Finally, column reports a race-horse specification that pits all the competing explanations against each other. This specification reveals that Muslim expansion is the “winner” as it (only) exhibits a positive and statistically significant
The former – rainfall – is Haber and Menaldo’s preferred measure. Both rainfall and cereal production are conditions prior to the onset of conquest and are thus “pre-treatment.”
https://doi.org/10.1017/9781009367509.006 Published online by Cambridge University Press
The Institutional Legacy of Muslim Conquest
effect on state centralization. Substantively, this implies that Muslim conquest was an important determinant of state centralization from to while accounting for a territory’s climatic conditions, agricultural production, distance to preexisting trade routes, and preconquest state centralization. C Indigenous State Centralization Even though Figure . shows the coding procedure does not automatically associate Muslim conquest to an increase in state centralization, skeptics may still worry the DD estimates are upward biased. To allay this concern, in this sub-section I examine the indigenous sources of political centralization. Specifically, I focus on the first and third components of the state centralization index. The first component ðz1Þ measures the level of political organization (i.e., above a tribal level), while the third component ðz3Þ measures the percentage of territory under the control of the government. Higher values of each component correspond to greater political authority. While z1, z2, and z3 are positively correlated, focusing on z1 and z3 provides an objective way to minimize the effect of foreign imposed rule (that follows from conquest) that might bias the aggregate index. Table . presents DD estimates from this exercise. Panels A and B report the effect of Muslim expansion on each separate component (z1 and z3, respectively), while panel C examines a combined measure (by multiplying the components). Across all six specifications, Muslim expansion exhibits a robust, positive effect on a territory’s level of political organization and geographic control by the government. The estimated effects are informative. First, the estimated effects tend to be larger in magnitude relative to those reported in similar specifications/samples in Table .. This suggests the main results are unlikely to be biased upward. Second, the estimated effects tend to be larger in countries outside the Middle East and North Africa (see column ). This suggests the estimated effects are not necessarily driven by preconquest conditions in the MENA region that may have predisposed these territories to greater centralization during the conquest period (lingering bureaucratic structures from previous empires, access to preexisting trade routes to Europe, etc.). On balance, the DD estimates in Table . suggest the baseline results in Tables . and . are unlikely to be upward biased due to the construction of the aggregate index (by Putterman and colleagues).
. The results thus far have focused on the expansion and consolidation of political authority. Did Muslim conquest affect other areas of politics? While data on other dimensions of politics is scant in the premodern period, presumably the process of political centralization in conquered territories is likely to have been beneficial to the political survival of leaders (Svolik ). To
https://doi.org/10.1017/9781009367509.006 Published online by Cambridge University Press
Conquest Fostered Autocratization TABLE
. Examining the components of state centralization
Dependent variable Sample
Muslim expansion
Muslim expansion
Muslim expansion
Country fixed effects Period fixed effects Observations Countries
Components of state centralization Full
Exclude MENA
Panel A: Political organization (a) (a) . . (.)*** (.)*** Panel B: Geographic control (b) . (.)***
(b) . (.)***
Panel C: Political organization & geographic control (c) (c) . . (.)*** (.)*** Across all panels Yes Yes
Yes Yes
Notes: Estimation via OLS for a global sample of territories. Robust standard errors, clustered by country reported in parentheses. *, **, *** = statistically significant at , , and percent respectively. Across all specifications, the R-squared ranges from . to .. In panel A, the dependent variable is the “Z” component of the state centralization index, which measures the level of political authority (i.e., above tribal level, etc.). In panel B, the dependent variable is the “Z” component of the state centralization index, which measures the share of territory controlled by the government. In panel C, the dependent variable is the product of Z and Z (and then multiplied by ). Across all panels, the dependent variable ranges from to, where a higher value corresponds to greater political centralization. In columns a, b, and c, the sample excludes territories from the Middle East and North Africa.
explore this conjecture, I gather information on the duration of political dynasties across both Muslim and non-Muslim societies in the premodern period collected by Edmund Bosworth () and John Morby (). To evaluate whether leaders in Muslim societies enjoyed longer tenures in power during the period of Muslim conquest, I estimate variants of the following specification: DURATION ipt ¼ α þ φZi þ κPt þ λðZi Pt Þ þ ϕp þ νipt
ð4:2Þ
Unfortunately, data on Muslim leader spells is unavailable prior to since Muslim kingdoms did not exist.
https://doi.org/10.1017/9781009367509.006 Published online by Cambridge University Press
The Institutional Legacy of Muslim Conquest
where DURATIONipt is the duration in power of ruler i that assumed power in year t in century p. Zi indicates whether the ruler is the head of an Islamic polity (as identified by Bosworth and/or Morby), Pt is an indicator variable equal to during the period of Muslim conquest, and otherwise, and ϕp is a vector of century fixed effects. The coefficient of interest is λ, which evaluates whether the period of Muslim conquest differentially affected the duration of Muslim rulers (relative to non-Muslim rulers). If state centralization associated with Muslim expansion benefited monarchial rule, then λ should be positive and statistically significant. Table . reports the estimated effect of Muslim conquest on leader duration. I begin by examining the effect among leaders in Muslim societies only. Column shows that during the period of Muslim conquest, leaders in Muslim societies enjoyed . years more in power, which for the modal leader represents a percent extension in his political tenure. This effect is precisely estimated and suggests the strengthening of political authority during the conquest period was also politically salubrious for Muslim sovereigns, on average. This inference, however, may be at odds with recent scholarship. In a comparative analysis of political survival during the medieval period, Lisa Blaydes and Eric Chaney () show that European monarchs survived in power longer relative to their counterparts in the Islamic world. They attribute this divergence to differences in the institutional arrangements underlying absolutist rule. In Western Europe, greater bargaining with feudal lords compelled the monarch to grant political concessions. While this placed greater constraints on monarchial authority, it lowered the incentive of local elites to overthrow the leader. In contrast, Muslim monarchs faced fewer constraints on their authority but an elevated risk of deposition. This suggests that the period of Muslim conquest – which helped expand and consolidate sovereign authority – may have been highly politically unstable for Muslim rulers. To evaluate this hypothesis, I expand the sample to also include non-Muslim (Christian) leaders from Western Europe. Column in Table . suggests nuanced effects of Muslim conquest on political survival in a pooled sample of Muslim and non-Muslim sovereigns. While Muslim leaders survive for shorter durations on average (e.g., column , coefficient ¼ −7:2) relative to European rulers (consistent with Blaydes and
Since this regression is limited to a Muslim-only sample, it therefore does not control for Zi and its interaction with Mt . This finding is broadly consistent with recent evidence in Allen et al. (). Examining the specific case of Iraq, they find leader duration fell significantly (to around two years) after the Muslim conquest. In a related paper, Allen and Heldring (forthcoming) find a decline in state capacity in Iraq (stemming from changes in irrigation) affected local development in the late ninth century. Their findings connecting state capacity to development in ninth and tenth century Iraq suggests Muslim conquest may also explain patterns of economic development. This is potentially a fruitful area of future investigation.
https://doi.org/10.1017/9781009367509.006 Published online by Cambridge University Press
https://doi.org/10.1017/9781009367509.006 Published online by Cambridge University Press
TABLE
. Political survival during Muslim expansion
Dependent variable
Leader duration (years)
Data set
Bosworth/Morby ()
()
()
. (.)*** . (.)***
. (.)*** −. (.)*** −. (.) . (.)***
. (.)*** −. (.)*** −. (.) . (.)***
Muslim expansion (Zi Mt) Islam (Zi) Conquest period (Mt) Constant Century fixed effects Polity controls Sample No. observations
Nüssli ()
()
()
()
. (.)* . (.)***
. (.)*** −. (.)*** −. (.)*** . (.)***
. (.)*** −. (.)*** −. (.) . (.)***
. (.)*** −. (.)*** −. (.)** . (.)
Yes
Yes Yes
IS-WE
IS-WE
Yes Islam
IS-WE
IS-WE
Islam
IS-WE
Notes: Estimation via OLS. Robust standard errors, clustered by dynasty (columns –) and polity (columns –). *, **, *** = significant at , , and percent respectively. In columns and , the sample is restricted to dynasties/polities with a Muslim ruler. Columns , , – pools data from Islamic and Western European dynasties (IS-WE). In column , polity controls include: a polity’s average agricultural suitability, latitude at the polity’s centroid, the polity’s area at the start of the century, and proportion of polity’s territory that was part of the Roman Empire in year .
The Institutional Legacy of Muslim Conquest
Chaney’s “divergence”), Christian monarchs did not experience a significant boon during the period of Islamic conquest as indicated by the statistically insignificant effect on conquest (e.g., column , coefficient ¼ −1:581). However, the period of Muslim conquest did benefit Muslim rulers: They survived about –. years longer in power. The specification in column , which adds century fixed effects, suggests similar estimated effects. The inferences from columns – also hold in specifications with data at the polity level drawn from Christos Nüssli (). In columns –, I replicate the analysis in columns – with this alternate data source. An advantage to using polity-level data is the possibility to control for various polity-level covariates that could affect leader tenure. The specification in column does that by controlling for a polity’s average agricultural suitability, latitude (at its centroid), area, and proportion of territory that was part of the Roman Empire in year . The coefficient estimates in columns – corroborate those with data from Bosworth/Morby: The period of Muslim conquest extended the political tenure of leaders in Muslim societies.
. : A The Presence of Mamluk Institutions The main findings in Table . show that territories exposed to conquest by Muslim armies ðMi > 0Þ experienced a significant increase in state centralization (autocratization) during the conquest period. The historical narrative in the previous chapter attributes this autocratization to the introduction and consolidation of mamluk institutions (mamluks and iqta) in territories conquered by Muslim armies. In this sub-section, I provide statistical evidence in support of this underlying channel by examining patterns of autocratization in territories exposed to Islam by but where mamluk institutions were not important to governance. Whereas military conquest played a critical role in the spread of Islam in the Middle East, North Africa, Spain, Anatolia, Central and South Asia, elsewhere it played a marginal role. In some societies where Islam became prevalent by , particularly in sub-Saharan Africa (SSA), the religion spread primarily through trade. In these territories, conversion to Islam gained adherents access to lucrative trade networks
The estimates in Table . suggest two important inferences about political survival during the early Medieval era. The period of Muslim conquest was not politically detrimental to the survival of Muslim rulers. However, the average effect of being a Muslim ruler is negative over the entire sample period (– ), which is consistent with Blaydes and Chaney’s () narrative that a divergence in the political survival between Muslim and European monarchs emerged in the medieval period due to differences in their institutional configuration (i.e., feudalism in Europe and non-feudal in Muslim societies).
https://doi.org/10.1017/9781009367509.006 Published online by Cambridge University Press
Conquest Fostered Autocratization
(e.g., Lapidus , Kennedy , Michapoulos et al. ). Crucially, based on my reading of Islamic historians, the expansion of Islam in these territories had a very light institutional footprint and did not introduce mamluk institutions nor necessarily its trajectory of centralized autocracy thereafter. For example, as discussed in the previous chapter, in sub-Saharan Africa these newly converted Muslim rulers continued to govern based on clan and kinship alliances (Levtzion ). The Mali Kingdom was a “typical West African Islamic regime” (Lapidus , ) in which politics was organized around the family and village units, with the head of the family serving as both the priest and chieftain. Groups of villages (called kafs) would align and coalesce primarily along shared kinship and would “choose” a ruler (and associated royal clan). This political organization would aggregate further through a consortium of related clans. Functionally, revenues were directly extracted from taxes and tributes from dependent communities. Administratively, there was a central territory (i.e., a capital with surrounding village networks), directly controlled by governors, and tribute-paying territories on the periphery ruled by vassal chieftains. The notable absence of mamluk institutions in sub-Sahara Africa allows me to compare patterns of autocratization in these territories (which correspond to N i > 0 from Table .) with patterns in those territories also exposed to Islam but where mamluk institutions were present (Mi > 0 from Table .). I do so by including the interaction of N i Pt as an additional covariate in Equation .. Thus, this interactive effect – which I will refer to “non-Mamluk Muslim expansion” – traces movements in state centralization in countries exposed to Islam by in sub-Saharan Africa (e.g., modern-day Mali, Nigeria, Sudan) from –. Column in Table . shows that non-Mamluk Muslim expansion ðN i Pt Þ had no effect on state centralization. The estimated effect is small and statistically indistinguishable from zero. In contrast, the effect of Muslim expansion ðMi Pt Þ remains positive, similar in magnitude to the coefficient estimates in Table ., and statistically significant. Substantively, the estimated effects suggest that Muslim expansion autocratized societies in territories where mamluk institutions were present (e.g., Middle East, Central Asia) but had no effect in Muslim territories in sub-Saharan Africa. Column shows this inference effect is not driven by the inclusion of territories in the Middle East and North Africa. Consistent with the main result in this chapter, Muslim expansion remains a robust determinant of autocratization in territories outside of the MENA region. Columns and further demonstrate the importance of mamluk institutions on state centralization when controlling for competing historical explanations (e.g., log rain conquest period) and with the indigenous
As I argued in the previous chapter (Section .), access to trade networks also played a role in the expansion of Islam to Southeast Asia (e.g., Indonesia, Malaysia). Islam reached these regions after and thus are not considered part of the first wave of Muslim conquest.
https://doi.org/10.1017/9781009367509.006 Published online by Cambridge University Press
https://doi.org/10.1017/9781009367509.006 Published online by Cambridge University Press
TABLE
. The importance of mamluk institutions during Muslim expansion
Dependent variable Sample
State centralization index Global
() Muslim expansion (Mi Pt)
Excl. MENA
()
Global
Global
Conquest territories
z and z
More unequal
()
()
()
Less unequal ()
()
. (.)*** −. (.)
. (.)** . (.)
. (.)*** . (.)
. (.)
. (.)**
Non-mamluk Muslim Expansion (Ni Pt)
. (.)*** −. (.)
. (.)* −. (.)
Competing explanations Country fixed effects Period fixed effects No. observations Countries R-squared
Yes Yes .
Yes Yes .
Yes Yes Yes .
Yes Yes .
Yes Yes .
Yes Yes .
Yes Yes .
Notes: Robust standard errors, clustered in countries in parentheses. *** = significant at percent. All specifications include country and year fixed effects. These coefficients and a constant are not reported. Column excludes countries from the MENA region. Column controls for competing historical explanations: log rain conquest period, cereal production conquest period, minimum distance to trade route conquest period and state centralization in conquest period. In column , the dependent variable is the product of the z and z components (and that multiplied by ) of the state centralization index. In columns –, the sample is limited to territories that experienced any conquest (i.e., Mi > 0 or Ni > 0). In column , the sample is limited to territories with land gini levels (inequality) levels above the median. These territories correspond to more unequal territories (where agricultural productivity is likely to be lower). In columns and , the sample is limited to territories with land gini levels (inequality) below the median. These territories are likely to exhibit greater agricultural productivity.
Conquest Fostered Autocratization
determinants of centralization as the dependent variable. Together, the estimates in Table . suggest that the presence of mamluk institutions in territories conquered by Muslim armies played a crucial role in state autocratization during the conquest period. B The Efficacy of Mamluk Institutions In addition to the presence of mamluk institutions, the efficacy of these institutions relied on both the repressive capacity of the elite soldiers (mamluks) and sufficient funds to compensate them. The latter was achieved through iqta, a system of non-hereditary grants that permitted mamluks to tax agricultural lands. Moreover, because these lands could not be inherited or sold, the income generated from this tax scheme gave mamluks a vested interest in the ruler’s political ambitions, including the expansion and consolidation of political authority in conquered territories. This suggests that more agriculturally productive territories would raise the efficacy of mamluk institutions and concomitantly foster state centralization. To explore this conjecture, I draw on findings from Michalopoulos et al. () and employ the distribution of land quality to measure a society’s agricultural productive capabilities during the conquest period. In particular, Michalopoulos et al.’s analysis reveals that within countries, more unequally endowed regions display a larger land allocation toward animal husbandry compared to farming. To the extent that farming was more conducive (lucrative) for taxation, this suggests the efficacy of iqta may have been influenced by the distribution of land quality: More equal distributions would likely generate higher agricultural production, and thus greater tax revenues for mamluks. This in turn may have aided the state centralization process. I test this ‘layered’ hypothesis by re-estimating Equation . across two samples of conquered territories (i.e., Mi > 0 and N i > 0): one above the median level of land inequality in conquered territories in the year (“More unequal”) and one below this group median (“Less unequal”). These results are reported in columns – in Table .. Column shows that Muslim expansion did not have a robust effect on state centralization in territories with a more unequal distribution of land quality. In contrast, column shows that Muslim expansion did have a statistically significant effect in centralizing political authority in societies with a more equal distribution of land quality. Furthermore, this finding holds in a specification that focuses on territories where mamluk institutions were known to be present (column ). Together, the estimated effects in columns – suggest that Muslim expansion strengthened political centralization in
Unfortunately, direct measures of agricultural productivity during the conquest period are unavailable. Rather, Michalopoulous et al. () use contemporary disaggregated data on the suitability of land for agriculture to proxy for regional productive endowments. The median level of land inequality in conquered territories is ..
https://doi.org/10.1017/9781009367509.006 Published online by Cambridge University Press
The Institutional Legacy of Muslim Conquest
territories that were potentially more agriculturally productive and plausibly, a more lucrative source of revenue to compensate mamluks.
. The results so far demonstrate two key findings: () during the period of Muslim conquest, conquered territories became more politically centralized (autocratized) and () Muslim monarchs, on average, survived longer in power. Both findings demonstrate that Muslim conquest – and the introduction of mamluk institutions in conquered territories – effectively centralized political authority in dictatorship. Did this equilibrium of autocratized structures in conquered territories remain intact after the initial period of Muslim conquest? In Chapter , I described how three successive foreign “interventions” by Mongol/Turkic invaders (–), the Ottomans (–), and European colonialists (–) largely built on existing governing structures in conquered territories. If that historical narrative is correct, then state centralization should not decline in conquered territories after . To evaluate this conjecture, I augment the DD specification given by Equation . with additional covariates that interact the mamluk territories under each different foreign intervener with the duration of that intervener’s rule. Specifically, I estimate the following DD specification: SCit ¼ α þ βMi þ ηPt þ θðMi Pt Þ þ ηðTURMi TURt Þ þ φðOTTMi OTT t Þ þ κðEURMi EURt Þ þ δ TURt þ ω OTT t þ πEURt þ μit ð4:3Þ In Equation ., the first three covariates – Mi, Pt, and Mi Pt – are analogous to those in Equation .. The remaining covariates evaluate the effect of subsequent interventions on state centralization (SCit), where TUR, OTT, and EUR refer to rule by Mongols/Turkics, Ottomans, and Europeans respectively. For example, OTTMi measures the percentage of mamluk territories that were ruled by the Ottomans. This means OTTMi is equal to . for Turkey but equal to zero for Spain which was not under Ottoman rule, even though it was conquered by Muslim armies (recall, Mi ¼ 0:37 for Spain). OTTt is equal to for the duration (period) of Ottoman rule and zero otherwise. Thus, OTTMi OTT t identifies the differential effect of Ottoman rule on state centralization in Ottoman controlled mamluk territories during the Ottoman era.
Specifically, OTTMi is equal to Mi for “conquered” territories under Ottoman rule, and zero otherwise. TURMi, TURt, EURMi, and EURt are defined analogously for Mongol/Turkic and European colonial rule.
https://doi.org/10.1017/9781009367509.006 Published online by Cambridge University Press
Conquest Fostered Autocratization TABLE
. The marginal effect of foreign interventions on state centralization Total marginal effect
Muslim conquest Mongol/Turkic Ottoman rule European colonialism
. (.)*** . (.) . (.) . (.)***
Notes: Estimation via OLS using a difference-in-differences specification. Robust standard errors, clustered by country in parentheses. *** = significant at percent.
If my historical narrative in Chapter is accurate, I expect the total marginal effect of Ottoman rule in mamluk territories as well as those associated with Mongol/Turkic and European rule to be non-negative. Accordingly, Table . reports each intervention’s total marginal effect on state centralization. The marginal effect in the first row implies that a one standard deviation increase in Mi corresponds to a . index point increase in state centralization in each half-century during the period of Muslim conquest (or, to a cumulative . index point increase). This marginal effect is larger than that associated with Mongol/Turkic invaders (row ), whose positive coefficient implies that Mongol invasions did not dismantle the political organization in conquered Muslim territories. Ottoman rule does not robustly increase state centralization in conquered territories (row ). This effect likely reflects the use of indirect rule by Ottoman sultans and their failure to introduce institutional innovations in governance that might have further centralized political authority. In contrast, European colonizers seem to have strengthened state capacity in Muslim societies (row ), perhaps through the ‘exporting’ of bureaucratic and administrative reforms. However, the substantive marginal impact associated with European colonialism is smaller than that associated with Muslim conquest. In the next chapter, I will provide additional evidence exploring the institutional legacy of Muslim conquest on the strategies of colonial rule, such as the greater propensity for indirect rule in territories formerly conquered by Muslim armies. Importantly, all three positive marginal effects in Table . imply that subsequent foreign interventions did not weaken the centralized autocratic structure in countries exposed to the initial period of Muslim conquest.
These total marginal effects are the linear combination of Mi and the relevant interaction term, calculated at the means for each covariate. This corresponds to nine half centuries from to CE, so 9 ð0:33 3:542Þ ¼ 10:52.
https://doi.org/10.1017/9781009367509.006 Published online by Cambridge University Press
The Institutional Legacy of Muslim Conquest
. This chapter provided statistical evidence that Muslim conquest and its associated introduction of mamluk institutions helped autocratize conquered territories. As described in the chapter’s introduction, this is no easy task. Any attempt at statistical inference faces challenges with () accurately defining and “measuring” Muslim conquest and state centralization and () mitigating concerns about endogeneity. For the former, I used historical maps to bound the geographic scope of Muslim conquest (between and ) to contemporary national boundaries and employed a transparent definition and construction of state centralization over time (from Putterman ). For the latter, I exploited the panel structure of the data and leveraged a differencein-differences (DD) research design to identify the causal effect of Muslim conquest on state centralization. The DD results demonstrate that Muslim conquest increased state centralization in conquered territories during the conquest period. As there was no democratic alternative in the premodern period, I interpret this robust increase in state centralization as evidence that Muslim conquest helped consolidate political authority in absolutist rule (autocratization). This interpretation of the findings illustrates how the classical Islamic equilibrium emerged in territories conquered by Muslim armies. The chapter’s statistical evidence largely corroborates the historical narrative in Chapter describing how the introduction of mamluk institutions and its propagating agents (clerics, the military) helped centralize political authority and its persistence after . In the next chapter, I trace the autocratic legacy of Muslim conquest through to the contemporary period.
https://doi.org/10.1017/9781009367509.006 Published online by Cambridge University Press
The Autocratic Legacy of Muslim Conquest
This chapter documents the effect of Muslim military conquest on democratic governance after the initial expansion (from to ). Whereas the previous two chapters established how Muslim conquest helped generate a classical Islamic equilibrium of centralized autocracy in conquered territories, this chapter traces the institutional legacy of the Islamic equilibrium through to the present. In particular, the analysis in this chapter evaluates the third and fourth steps in Figure . which described the consequences of Muslim conquest on the European colonial “strategy” (and its effects on democracy) and the effects after European colonial rule. Collectively, the empirical evidence in this chapter and Chapters and show how Muslim conquest and the subsequent trajectory of autocratic institutions it introduced is a substantively important source of less cohesive political institutions (i.e., a ‘stock’ of nondemocratic political structures) in contemporary Muslim societies. This sticky stock of less cohesive institutions in turn forms the basis for whether temporal variation rents has contributed to political violence (dictatorship, civil war) in contemporary Muslim societies (which Chapters – will focus on). I proceed methodically to trace the autocratic legacy of Muslim conquest and so by leveraging various data sources and empirical strategies. I present four key findings. First, I analyze whether the Islamic equilibrium can be “undone” in a historical perspective. I compile original subnational data from
The third step stipulates that the prevalence of centralized autocratic institutions affected the European colonial strategy: It incentivized the use of indirect rule by colonial powers and concomitantly reduced European migration to territories conquered by Muslim armies (relative to non-conquered territories). The fourth step describes politics at the end of European colonial rule. Specifically, at their independence, Muslim countries largely maintained their autocratic structures; institutions that remain today.
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
The Institutional Legacy of Muslim Conquest
medieval Spain to show how Muslim rule (involving mamlukism) stunted the adoption of provincial representative assemblies (“first parliaments”). The introduction of provincial assemblies marked an important first step in the democratic trajectory in Spain and more broadly Europe (Downing , Finer , Stasavage ). Second, I then broaden the analysis beyond Spain and examine the long-run effect of Muslim conquest on contemporary democracy around the world. I establish an extremely robust association: Countries exposed to mamluk institutions (Mi > 0 from the previous chapter) are less democratic today. This analysis of the long-run association reveals an important finding: Muslim majority societies that were not exposed to Muslim conquest (e.g., countries in sub-Saharan Africa, Malaysia, Indonesia) are not statistically more likely to exhibit dictatorship. Essentially, this placebo test implies that Islam is not inherently incompatible with democracy. Third, having established the negative association between Muslim conquest and contemporary democracy, in Section ., I evaluate the intermediate steps: the role of Muslim conquest on politics from the start of the modern period (circa ) to the present. I establish two key results. First, using each country’s -year change in democracy (from Acemoglu et al. ), I show that countries exposed to Muslim conquest were significantly less likely to experience movements toward democracy between and . Second, I show how Muslim conquest affected the European colonial strategy and its effects on democracy: Countries exposed to mamluk institutions experienced less diffusion of democracy through European settlement and a heightened use of indirect rule, which preserved the precolonial autocratic structures and governing coalitions. My analysis also evaluates the effect of colonialism on democracy in Muslim societies. Skeptics may still contend that Muslim conquest captures the effect of the Islamic faith (rather than the institutional legacy of mamluk institutions) on democracy. To dispel this supposition, in Section ., the fourth set of results “decomposes” the effect of Muslim conquest on democracy along two dimensions: religiosity and over time. The first dimension isolates the effect of Islam from Muslim conquest by investigating the relationship between religiosity and democracy at the societal level. Here, I show that more pious Muslim societies tend to have experienced more democracy; and moreover, this relationship is not associated with whether a country was exposed to mamluk institutions. Together this implies that adherence and practice of Islam is not inherently anti
This analysis evaluates the final step in Figure .. This negative association corroborates Eric Chaney’s () original finding and is strikingly strong: It holds across various measures of democracy; specifications that control for a battery of historical and contemporary confounders (e.g., per capita income, coercive capacity, preconquest state centralization, agricultural/climatic conditions), and estimation via instrumental variables. These results are presented in the chapter’s Appendix.
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
The Autocratic Legacy of Muslim Conquest
democratic, countering claims that greater religiosity is an obstacle to democracy (e.g., Huntington , Lewis , Weber []). The second dimension investigates how democracy varies over time in Muslim societies exposed to mamluk institutions. This preliminary exploration shows that rentier structure matters in Muslim societies: Dictatorship has been more stable in oil-producing countries relative to non-oil-producing countries. This finding will be explored in greater detail in Chapters –.
. : The results in Chapter demonstrate that Muslim conquest autocratized political authority with a persistent effect in conquered territories thereafter. Yet not all territories initially conquered by Muslim armies remained that way. And in these territories, the weakening of the Islamic equilibrium may have created an opportunity for the emergence of more representative forms of government. In this section, I probe this conjecture in a historical perspective. I compile novel subnational level data to evaluate the legacy of Muslim rule on the emergence of provincial parliaments in medieval Spain. My aim in this section is to examine whether the Islamic equilibrium can be undone and probe its consequence for the emergence of democratic institutions in a historical perspective. The analysis reveals that Spanish provinces under longer durations of Muslim rule were slower to hold their first parliament. This finding suggests that breaking down the governance structures associated with Muslim conquest can lead to more representative (democratic) politics, albeit with a delay. My analysis of medieval Spain is advantageous in at least two ways. First, it examines the political legacy of Muslim conquest at the subnational level. Second, it employs an alternate measure of conquest exposure: duration of Muslim rule.
A Muslim Rule in Spain Institutional Context The emergence of representative assemblies (first parliaments) is frequently identified as the beginning of a process that placed greater checks and balances on absolutist rule (Downing , Stasavage ). In Europe, this process began in the thirteenth and fourteenth centuries, predominantly at the provincial level. As Samuel Finer (, ) notes “beginning with Spain, next in England, France, and Italy, and then spreading into Germany, Scandinavia, and even Poland and Hungary, there sprang up a multitude of conciliar bodies to give consent to but also – by the same token – to exert some control over their rulers.” In the Iberian Peninsula, these assemblies were called cortes or corts. While these early assemblies varied in their geographic jurisdiction,
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
The Institutional Legacy of Muslim Conquest
composition/structure, degree of mandate, and extent of tax and legislative authority, they all sought to place greater constraints on the sovereign’s powers (Finer , ). In Spain, the introduction of representative assemblies emerged against a backdrop of reconquest of territory controlled by Muslim rulers (O’Callaghan ). In , Muslim armies penetrated the Iberian Peninsula from North Africa, quickly gaining territory up to the eastern Pyrenees within the first decade. Thomas Glick (, –) identifies two distinct phases. The first was a period of “adjustment” in which newly captured territory was organized as a province of the Umayyad caliphate of Damascus. The second phase of “consolidation” – starting in the mid-ninth century – marked the growing power and wealth of Islamic rule, culminating in the establishment of the caliphate of Córdoba by Abd al-Rahman III. Mamluk Institutions Featured under Muslim Rule The caliphate organized the country’s institutions and economy along ‘Abbasid lines’ that “entailed centralizing political power by concentrating it in the person of the emir, ruling through a tightly controlled hierarchical central bureaucracy” (Glick , ). During the tenth century, the caliphate reached its zenith of “fame, power, and wealth” in which the Caliph became “a grand and distant figure” who “ruled through a hierarchy of slaves” (Glick , and ). Reconquest Following the death of Caliph al-Mansúr of Córdoba in , the fortunes of the caliphate turned dour. After a period of internal struggle and shifting power relations (the fitna or anarchy of –), the unitary Islamic state was replaced by a constellation of two dozen or so small states. The caliphate’s fragmentation provided an opportunity for the Christian kingdoms – primarily Castile, León, and Aragon – to intensify reconquering territory controlled by the Moors from different directions on the Iberian Peninsula (starting around ). While the process of reconquest was geographically uneven and occurred at differential “speeds” (Oto-Peralías and Romero-Ávila ), the eleventh century was the “most critical and significant of the Spanish Middle Ages” as it “witnessed a change in the peninsular balance of power” (Glick , –). Following a series of victories by Alfonso IX of León (from the west, e.g., in Badajoz), Ferdinand III of Castile (through the center, e.g., the heart of al-Andalus), and James I of Aragon (from the east, e.g., in the Balearic Islands and Valencia), the caliphate’s territory had been reduced to the region
For example, the composition of Spanish assemblies varied across kingdoms (Finer , ). In León, Castile, Catalonia, and Valencia, individuals from three Estates – the clergy, the nobles, and the towns – were permitted to participate in assemblies. This was not the case in Aragon.
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
The Autocratic Legacy of Muslim Conquest
of Andalusia by the . Confined to Granada, Muslims would continue to rule until , when Caliph Muhammad XII surrendered to Spanish forces. By , a unified Spain had effectively expelled Moorish rule. Because Muslim rule increased centralization, it also contributed to a greater ability to resist reconquest by various fragmented provinces. As a result, the gradual and dispersed process of reconquest meant a geographically non-uniform end to Islamic rule and its institutional legacy across Spain. In reconquered areas, local populations experienced profound social reorganization in which Moorish institutions (e.g., religious, political) were expunged (O’Callaghan ). Critically, centuries of Moorish rule meant that removing Islamic institutional features was neither automatic nor costless in reconquered provinces. Moreover, the height of this societal transformation unfolded around the same time representative assemblies started to emerge in Spanish provinces. Thus, the concurrent timing of reconquest and the emergence of representative assemblies in Spain suggests that geographic variation in the legacy of Moorish institutions – measured with the duration of Muslim rule – is likely to have affected whether and when a province held its first parliament. Since Muslim rule comprised autocratic institutions (with supportive elites), I hypothesize that provinces that experienced a longer duration of Muslim rule were “slower” to hold their first parliament. B Data To investigate this conjecture, I gather economic, geographic, and political data from the Middle Ages (– CE) for each of Spain’s contemporary provinces. From Scott Abramson and Carles Boix (), I identify when each Spanish province held its first representative assembly. Individuals in Barcelona assembled the first parliament on the Iberian Peninsula in , with parliaments in Benavente (in the province of Zamora) and Puigcerdá (in the province of Girona) emerging two and six years later. By , half of Spain’s provinces had held a parliament. Turning to the independent and control variables, I glean provincial level data on the duration of Muslim rule, the speed of reconquest, and various important pre-reconquest geographic and economic controls including area, soil quality, rainfall, urbanization in , and Roman road density. The modal province experienced four centuries of Muslim rule (with a mean of . and standard deviation of . centuries). With this subnational data, I estimate two regression models. The first model examines whether the duration of Muslim rule in each province affected its
Glick (, –) summarizes the progression of battles/victories during the Reconquest. In this regard, the timing of reconquest is itself endogenous to the Muslim conquest in the previous period. A complete set of data is available for forty-six of Spain’s fifty provinces. Much of this data is drawn from Oto-Peralías and Romero-Ávila ().
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
The Institutional Legacy of Muslim Conquest
probability of having its first assembly by , , and . The second model probes whether the duration of Muslim rule affects when (i.e., the year) a province holds its first parliament. For the first model, I expect a negative relationship: Provinces under a longer period of Muslim rule should be less likely to hold a first parliament. In contrast, for the second model – which uses the year of first parliament as the dependent variable – I expect a positive relationship: Provinces with a longer duration of Muslim rule should hold their first parliament in a later year. C Muslim Rule Delays First Parliaments Figure . provides evidence in the raw data that provinces under longer duration of Muslim rule tended to hold their parliament in a later year. The fitted line illustrates this positive bivariate association. The figure also shows this association is not “mechanical.” Specifically, it was possible for a province under Muslim rule to experience some nascent representative assemblies, for example, the province of Cadiz. Moreover, there are provinces where parliaments emerged quite early despite experiencing several centuries of Muslim rule. For example, Muslims ruled in Zaragoza and Lleida for four centuries but both provinces held their first parliaments by . There are also provinces where Muslims did not rule but held their first parliaments quite late. For
1500
Coruña Lugo (A) Pontevedra Orense
Albacete(Illes) Balears Badajoz Cáceres Cuenca Jaén Murcia Huelva Alicante
Year of first parliament 1300 1400
Salamanca
Córdoba
Soria Alava
Castellón de la Plana Guadalajara Madrid Palencia
Rioja (La) Avila Segovia
Navarra
2
1200
Ciudad RealCádiz Sevilla Teruela Valenci
Zaragoza Tarragona Huesca Toledo Lleida
Valladolid Burgos León Girona Barcelona Zamora
0
4
6
Centuries under Muslim rule Year of first parliament FIGURE
Al Gr Má Pal Sa
Fitted values
. Muslim rule and provincial first parliaments in Medieval Spain
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
8
The Autocratic Legacy of Muslim Conquest
example, the northern province of Alava never experienced any Muslim rule yet held its first parliament over a century after Barcelona. Similarly, Muslims never ruled in Coruna and Lugo; yet they did not hold their first representative assemblies until the sixteenth century. Table . reports corroborating statistical evidence for a positive association in specifications that control for confounding factors. Columns – show that provinces with a longer duration of Muslim rule were less likely to hold a parliament between and . Examining variation in the incidence of parliaments by , column reports a rather conservative bivariate relationship: An additional century of Muslim rule reduces that province’s probability of holding its first parliament by . percentage points. This implies that for the typical province, which experienced four centuries of Muslim rule, the institutional legacy of Muslim conquest reduced its likelihood of holding its first parliament by percentage points. Accounting for important economic and geographic confounders does not attenuate the robust negative effect of Muslim rule (column ). Contrary to existing scholarship, geographically smaller provinces (Stasavage ) are not more likely to hold a first parliament by and nor are provinces with greater urbanization (Abramson and Boix ). Column introduces two potentially important political factors: the speed of reconquest which affected the initial concentration of political and economic power in reconquered territories (Oto-Peralías and Romero-Ávila ) and the Kingdom of Aragon’s important role in reconquering territory (Glick , –) and being a realm with early parliamentary activity. The rate of reconquest has no effect on the probability of a province’s first parliament by , while provinces in Aragon were slightly more likely to have done so by then (though the effect is not statistically significant). In this specification, the negative effect of Muslim rule remains statistically significant. Columns – expand the temporal window for a province’s first parliament. These specifications show that provinces under a longer duration of Muslim rule were less likely to have held a parliament by and , respectively. Interestingly, by some of the control variables gain statistical significance (column ). Provinces with greater preconquest economic development (e.g., urbanization, soil quality) are more likely to have held a representative assembly, while provinces with a slower rate of reconquest did so as well. Finally, columns and show that provinces with a longer history of Muslim rule held their first parliament later. In column , the coefficient on Muslim rule implies that an additional century of Islamic control delays a province’s first parliament by forty-three years. While the inclusion of economic, geographic, and political controls in column reduces the magnitude of Muslim rule by almost a decade (to . years), the estimated effect remains
In the data, many provinces in Aragon held their first parliaments by .
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
TABLE
. Muslim rule and first parliaments in Spanish provinces Probability of parliament by …
Dep. variable:
Centuries under Muslim rule
Year of first parliament
()
()
()
()
()
−. (.)**
−. (.)***
−. (.)***
−. (.)***
−. (.)***
. (.)***
. (.)***
. (.) . (.) −. (.) −. (.)* −. (.) (.)
. (.)** . (.)* −. (.) −. (.) . (.) . (.)
. (.) . (.)*** −. (.) −. (.) −. (.) . (.)
. (.) . (.)** −. (.)** −. (.) . (.)*** . (.)
−. (.) −. (.)** . (.) . (.)** . (.) −. (.)
−. (.)** −. (.)** . (.) . (.) −. (.) −. (.)
−. (.) . (.) −. (.)
−. (.)** . (.) . (.)
. (.)***
. (.)* −. (.) . (.)***
.
.
.
.
Initial conditions Log area Soil quality Rainfall Latitude Urban pop. density in Roman road Density Politics Reconquest rate
Constant
. (.)***
−. (.)
−. (.) . (.) −. (.)
Provinces R-squared
.
.
.
Crown Aragon
()
()
Notes: Estimation via ordinary least squares (OLS). Robust standard errors reported in parentheses. *, **, *** = significant at , , and percent respectively
The Autocratic Legacy of Muslim Conquest
statistically significant. Interestingly, unlike the null effects in columns –, some of the confounding variables in column exhibit relationships consistent with related research. For instance, geographically smaller provinces held their first parliaments at an earlier date, corroborating a finding that “distance mattered” with early parliaments (Stasavage ). Moreover, the positive coefficient on the rate of reconquest (=.) supports Oto-Peralías and Romero-Ávila’s () conjecture that territories reconquered at a faster rate tended to have lower levels of settler migration and subsequently higher initial levels of economic and political inequality; such factors would make these provinces more inclined to hold their first parliament later.
. A Conquest Harms Contemporary Democracy The analysis in the previous section shows how Muslim rule stunted the adoption of early democracy (representative assemblies) in medieval Spain. In this section, I broaden the analysis and fast-forward to the contemporary era to study the long-run effect of Muslim conquest on democracy for a global sample of countries. The findings show that Muslim conquest harms democracy. To evaluate the effect of Muslim conquest on contemporary democracy, I follow the standard empirical approach in the historical legacies literature which regresses the main outcome variable(s) on the “critical juncture” (independent variable) plus a vector of exogenous variables that are causally prior to the critical juncture (i.e., “pretreatment” covariates). Thus, I estimate variants of the following regression specification: Di ¼ α þ βMi þ φXi þ εi
ð5:1Þ
where Di is country i’s level of democracy in the contemporary era, Mi is the percentage of country i’s territory that was conquered by Muslim armies and governed with mamluk institutions. Xi is a vector of exogenous variables that are causally prior to Mi that might affect contemporary democracy, such as geography, and εi is an error term. To account for potential heteroskedasticity in the cross section of countries, the standard errors are robustly estimated. Given the historic nature of the argument advanced in this book, my preferred measure of democracy in the contemporary era is a country’s accumulated democratic experience since (“accumulated democracy”) from
This conclusion corroborates Chaney’s () finding that areas exposed to greater Muslim conquest have a lower level of democracy in . As I describe below, my preferred variable of democracy – “accumulated democratic experience” – represents a more attuned measure of democracy in a historical perspective. See Nunn () for an overview.
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
The Institutional Legacy of Muslim Conquest
Gerring et al. (). As these authors note, “democracy and authoritarianism . . . construct deep legacies, extending back several decades, perhaps even centuries” (). Since my argument centers on the historical roots of autocracy, this cumulative measure is more relevant than democracy scores in any individual year (although, the findings hold with these contemporaneous measures, see Table C in the chapter’s Appendix). Accumulated democracy is continuous and lies on a to interval, where zero is least democratic and one is the most. Thus, if Muslim conquest has a pernicious long-run effect on democracy, the coefficient on accumulated democracy ðβÞ should be negative and statistically significant. There is a robust negative relationship between a country’s exposure to Muslim conquest ðMi Þ and its experience with democracy ðDi Þ in a cross section of countries. Using the raw data, Figure . plots this basic relationship with a fitted line across the entire sample of conquered countries (solid line). This negative correlation is also present when excluding countries from the Middle East and North Africa (dotted line). Together, these bivariate associations provide initial evidence that Muslim conquest exhibits a deleterious relationship with contemporary democracy, even in societies outside of the MENA region. Table . provides additional statistical corroboration (based on estimating Equation .) for the negative association in Figure .. Column shows that the accumulated democratic experience for countries whose modern-day territory was fully conquered by Muslim armies (and was exposed to mamluk institutions) is . point lower than that for countries which were not conquered. For the average conquered country (i.e., Mi ¼ 0:74) – which includes many states outside the MENA region – the accumulated democratic experience is . point lower ð¼ 0:74∗ −0:26Þ than non-conquered countries. This
This “stock” variable sums a country’s composite POLITY score from to the present, with a percent annual depreciation rate (Gerring et al. , ). I then use a country’s accumulated democracy score in and normalize it on a [,] scale, where a higher value corresponds to greater accumulated democratic experience. That said, accumulated democracy is highly correlated with alternate, contemporaneous measures, including those from POLITY, Freedom House, and Boix, Miller, and Rosato (). Table C shows that Muslim conquest is negatively associated with these alternate measures of democracy. I am not the first scholar to use this measure. For instance, Jacob Hariri () also uses accumulated democracy in this study of the democratic deficit in Middle Eastern and North African societies. It is worth noting that accumulated democracy underestimates the democratic deficit in Muslim societies in the contemporary period, as the measure includes periods when the lack of democracy in Muslim societies was not as great (relative to non-Muslim societies). For example, this deficit was smaller in the early twentieth century when there were relatively few democratic nonMuslim countries. The negative association holds when excluding countries less vulnerable to Muslim conquest, such as in the Americas and Oceania.
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
The Autocratic Legacy of Muslim Conquest . Muslim conquest harms a country’s contemporary democratic experience
TABLE
Dependent variable:
Accumulated democracy Exclude MENA
Conquest (Mi)
()
()
−.
−.
() −.
()
()
()
−.
−.
−.
(.)*** (.)*** (.)** (.)*** (.)*** (.)*** . . . (.) (.) (.) Muslim −. −. non-conquest (.) (.)
MENA
Continent FE Countries R-squared .
Yes .
.
Yes .
Yes .
.
Note: Estimation via OLS. Robust standard errors reported in parentheses. *** = significant at percent. All specifications include a constant which is not reported. Columns and include fixed effects for the following continents: Africa, Asia, Europe, Americas, and Oceania. These coefficients are not reported. In columns and , the sample excludes countries from the MENA region.
0.8 0.6 0.4 0.2
Accumulated democracy
1
USA COS NEW SWZ AUL CAN UKG NOR DEN BEL SWD NTH IRE FRN FIN JPN AUS GMY JAM ITA GRC CYP SAF TRI MAS SRI COL PNG BOT URU MAL FJI GAM VEN CHL ECU HON PHI PER NAM BOL BNG BRA GUY ROK MAG ARG SIN ZIM COM ERI PAN CZR LES BEN SLO NIG MZM SOM SEN GNB ZAM UGA MLI ANG GUA SIE SAL POL SUD GHA CEN NIR HUN LAO THI KEN MYA CON BFO MAW BUI CHA TOG TAZ MAC CAM CUB EST SLV INS RWA MEX CAO MAA LAT NIC HAI SWA GUI GAB LIT PAR IVO MLD ETI BUL MON VIE CRO LBR UKR RUS ALB PRK BLR KZKKYR CHN
IND
SPN
TUR LEB PAK
POR
UAE ALG EGY BAH TUN QAT LIB KUW JOR SYR MOR YEM GRG IRN ARM IRQ
NEP
TAJ
AZE TKM UZB
OMA
BHU
0
SAU
0
0.2 0.4 0.6 0.8 Percent territory conquered by Muslim armies Observations Without MENA
FIGURE
1
With MENA
. Muslim conquest reduces a country’s accumulated democratic experience
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
The Institutional Legacy of Muslim Conquest
negative association between Mi and a country’s accumulated democratic experience corroborates Eric Chaney’s () original finding that countries with a greater share of their territory conquered by Arab armies had a lower POLITY score in . The robust negative relationship in column holds in specifications that account for the potential effect of Arab/Middle Eastern culture (so-called “Middle Eastern exceptionalism”) and Islam in fostering autocracy in many contemporary Muslim societies. However, appropriately controlling for these potential effects in Equation (.) is challenging since their occurrence is largely a consequence of Muslim conquest (i.e., these factors are “posttreatment”). Accounting for these effects therefore requires a variable that is highly correlated with these factors but is causally prior to Muslim conquest. Since the purported effect of culture and Islam tends to be geographically concentrated in the MENA region, I use a country’s location in the MENA region – with a MENA dummy variable – as a clean and causally prior measure. Column shows that accounting for this MENA effect does not attenuate the deleterious effect of Muslim conquest on accumulated democracy. The estimated effect of conquest increases in magnitude and remains statistically significant. Conditional on conquest, a country’s location in the MENA region is positively associated with democratic experience, but the effect is not statistically significant. More generally, as Jared Diamond () argues, geographic effects at the continent level might affect the trajectory of economic and political development. To account for this, column controls for continent fixed effects. In this specification the association between Mi and accumulated democracy remains negative and statistically significant. The Role of Islam Rather than capturing the effects of Arab/Muslim culture, skeptics may argue that Muslim conquest is proxying for the potential anti-democratic effects of the Islamic faith. Separating Muslim conquest from Islam is inherently
As I described in the previous chapter, Chaney’s measure of Arab conquest includes territories in sub-Saharan Africa (SSA). My measure of Muslim conquest excludes countries in SSA as they were not exposed to mamluk institutions. See Bellin () for a detailed discussion of the “Middle Eastern exceptionalism” argument. MENA is equal to for all countries in the MENA region and zero otherwise. For instance, Diamond’s () continental axis hypothesis contends that crops, animals, and innovations spread more easily along the East–West Eurasia axis (e.g., due to similar climatic conditions on the east–west axis) than on the North–South axis in Africa and the Americas. Consequently, the earliest civilizations and subsequent political developments first occurred in Asia, Europe, and the Middle East. For example, in describing the Third Wave of democracy in the late twentieth century, Samuel Huntington () observes that the Islamic “civilization” has served as an obstacle to the adoption of democracy. Several Muslim scholars have also argued that the Islamic faith can hinder democracy (e.g., Lewis , , Tessler ).
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
The Autocratic Legacy of Muslim Conquest
challenging since conquest preceded the adoption of Islam in many contemporary Muslim societies. However, if Islam does exert an independent antidemocratic effect on democracy, Muslim majority countries that were not exposed to military conquest (e.g., Muslim countries in sub-Saharan Africa, Southeast Asia) should also be less democratic. Column evaluates this proposition and reports an important placebo test. In this specification, the null effect on “Muslim non-conquest” implies that Muslim majority countries that were not conquered by Muslim armies (e.g., Malaysia) have not experienced less democracy in the contemporary era. Rather, Muslim societies that were exposed to the institutional legacy of Muslim military conquest have. Thus, column reveals that adherence to the Islamic faith is not inherently antidemocratic. Finally, to directly purge the effects of the MENA region altogether (any anti-democratic effects associated with its culture, geography, etc.), columns and reports the effect of Muslim conquest on accumulated democracy in a sample that excludes countries from the MENA region. In both specifications, conquest continues to exert a robust negative effect on accumulated democracy. Furthermore, column shows that Muslim societies outside the MENA region that were not exposed to conquest are not inherently anti-democratic as the coefficient on Muslim non-conquest is statistically non-distinguishable from zero. Robustness The findings in Table . are robust. Muslim conquest exhibits a negative effect on alternate measures of democracy from different data sources, including those from Carles Boix, Michael Miller and Sebastian Rosato (), Freedom House (), as well various components of democracy that comprise the aggregate POLITY index, such as constraints on the executive. The main findings are also not unduly influenced by correlates of democracy, such as fiscal capacity (Besley and Persson a, Dincecco ) and hold in specifications that use alternate measures of Muslim conquest (e.g., “Half mamluk” and “Full mamluk” from Chapter ) and instrument for Muslim conquest using a conquered territory’s distance from Mecca. These results are discussed and presented in the chapter’s Appendix. B Institutional Cohesiveness As discussed in previous chapters and emphasized in prior scholarship (e.g., Besley and Persson a, b), democracy is positively associated with
Hence, controlling for a Muslim majority dummy or percentage share of Muslim population in Equation . would not be appropriate since these measures are “posttreatment” (to Muslim military conquest). Recall from Chapter that the Islamic faith was gradually adopted in conquered territories, occurring up to centuries after the initial military conquest campaign.
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
TABLE
The Institutional Legacy of Muslim Conquest . Muslim conquest lowers institutional cohesiveness
Dependent variable:
Government transfers (% expenditure) ()
Accumulated democracy
()
. (.)**
()
−. (.)*
. (.)** −. (.)
()
Constant
. (.)***
. (.)***
. (.)***
−. (.)** . (.)*** −. (.) −. (.)
Countries R-squared
.
.
.
.
Conquest (Mi) Log GDP per capita Ethnic fractionalization
Notes: Estimation via OLS. Robust standard errors reported in parentheses. *, **, *** = significant at , , and percent respectively. “Accumulated democracy” is a country’s accumulated democratic experience on a [,] scale. “Log GDP per capita” is a country’s per capita GDP measured in real purchasing power parity dollars, averaged from to . Ethnic fractionalization is the probability of meeting someone from a different group and is drawn from James Fearon and David Laitin (). It ranges from to , where a higher value implies greater ethnic fragmentation.
institutional cohesiveness (e.g., see Figures ., ., and .). That is, governments in polities with more democratically oriented institutions tend to disburse a greater share of state resources with groups that do not necessarily support them. Thus, measures of democracy can serve as a proxy for institutional cohesiveness. Moreover, from an empirical standpoint, employing measures of democracy are advantageous since there is wide country and temporal coverage of democratic institutions (and practices) relative to measures of government finances, especially for developing countries. Despite this benefit, skeptics may wonder how Muslim conquest affects institutional cohesiveness in the long run. Table . presents evidence that the legacy of Muslim conquest – using the percentage of territory conquered by Muslim armies – lowers a country’s institutional cohesiveness in the long run. As explained in Chapters and , one can measure a country’s institutional cohesiveness with a government’s spending on welfare spending: A government that allocates a greater share of its expenditures on transfers (to the masses) is deemed to exhibit more institutional cohesiveness. With this interpretation in mind, column confirms that countries with longer experiences with democracy do indeed exhibit more cohesive institutions (measured with a government’s spending on transfers and subsidies as share of government expenditures, averaged from to ). Column
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
The Autocratic Legacy of Muslim Conquest
shows that Muslim conquest is negatively associated with institutional cohesion. The coefficient estimate implies that a government in the typical conquest society (i.e., with Mi ¼ 0:75) allocates about percent of its expenditures on transfers compared to non-conquest societies (that allocate about percent). This four-percentage point difference is substantive, as it corresponds to about a . standard deviation change in government transfers. A crucial feature of my argument – as diagrammed in Figure . – is that Muslim conquest introduced an institutional trajectory that set conquered societies to accumulate a “stock” of less democratic institutions and in turn exhibit less institutional cohesiveness today. This implies that if conquest affects institutional cohesiveness through a society’s stock of democracy, then controlling for a country’s accumulated democratic experience should reduce the effect of Muslim conquest on government transfers (my measure of institutional cohesiveness). Column shows this to be the case: The coefficient on conquest is no longer statistically significant and is smaller in magnitude compared to column . In this specification, accumulated democracy remains positively and robustly associated with government transfers. The estimated effects on accumulated democracy and conquest implies that a country’s stock of democracy “mediates” the effect of conquest on government transfers in the long run. Finally, it is plausible that other factors that are associated with both Muslim conquest and institutional cohesiveness might influence the relationships in columns and . To allay this concern with possible omitted variables, I control for a country’s per capita wealth and ethnic fragmentation in column . Prior studies have linked these factors to a government’s spending decisions on public goods (e.g., Acemoglu and Robinson , Alesina, Baqir, and Easterly ) and they could be plausibly associated with Muslim conquest.23 In this specification, Muslim conquest retains its negative and statistically significant association with government transfers. Consistent with existing studies, government transfers are higher in countries with higher per capita income, while they tend to be lower in more ethnically diverse societies.
. A Explanations from the Premodern Period The statistical evidence thus far provides robust evidence of a pernicious association between Muslim military conquest and a society’s subsequent
A similar mediation approach is employed in Chapter . For example, Kuran () shows the standard of living in Muslim countries tends to be lower, on average, compared to non-Muslim countries on many social and economic indicators. Moreover, many Muslim societies, especially in Africa and the Middle East, are quite ethnically diverse. This latter effect is not statistically significant.
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
The Institutional Legacy of Muslim Conquest
accumulated democratic experience. Yet, it is plausible that conquest might be capturing the effects of other premodern historical factors on democracy. Failure to account for these competing premodern narratives may constitute omitted variable bias. In this subsection, I evaluate three prominent historical narratives (as described in Chapter ) and show the negative association between Muslim conquest and contemporary democracy remains robust in specifications that account for these competing explanations. Agriculture The first competing narrative posits that certain types of rain-fed agriculture affected societies’ historical paths of institutional development; some proved favorable to stable democracy. For example, as I discussed in Chapter , Stephen Haber and Victor Menaldo (b) argue that in zones of moderate rainfall, unlike arid and tropical zones, it was possible to grow crops that were highly storable and that exhibited modest scale economies in production, which created broad-based incentives to trade, protect property rights, and make inter-generational investments in human capital. They argue that this led to the emergence of stable democracies in these “moderate” zones. Haber and Menaldo’s conjecture suggests that the results in Table . reflect the fact that many areas conquered by Muslim armies were not suitable for the development of rain-fed agriculture. Columns – in Table . evaluate this possibility. In column , I evaluate the effect of Islamic conquest on accumulated democracy while controlling for a cubic polynomial of a country’s average rainfall (as suggested by Haber and Menaldo). Column substitutes this polynomial with the percentage of a country’s landmass covered by desert terrain, while column controls for a country’s suitability for grain production (which is an alternate measure for a country’s capacity for settled agriculture that is not related to rainfall). Across these specifications, Muslim conquest retains its robust, long-run detrimental effect on democracy, while none of the measures for a country’s suitability to rain-fed agriculture has a robust effect. Precolonial State Development A second, related competing premodern explanation is precolonial state development that may be traced back to a country’s transition to the Neolithic Revolution (Hariri , ). In an effort to unpack the Middle East’s “exceptionalism” to democracy, Jacob Hariri () argues the region’s early transition to the Neolithic Revolution contributed to stronger precolonial state capacity, which in turn permitted these countries to resist the transplanting of democratic institutions through European colonialism. Thus, in exhibiting this greater resilience to colonial influence, countries with stronger precolonial state
Cereal suitability is based on Nunn and Qian (). The underlying country-level data on suitability for growing cereals is from the Food and Agriculture Organization’s Global AgroEcological Zones database.
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
TABLE
. Accounting for explanations from the premodern period
Dependent variable:
Conquest (Mi) MENA ln(rainfall) ln(rainfall) ln(rainfall)
Accumulated democracy ()
()
()
()
()
()
−. (.)*** . (.)*** −. (.) . (.) −. (.)
−. (.)*** . (.)**
−. (.)*** . (.)
−. (.)*** . (.)
−. (.)*** . (.)
−. (.)*** . (.)
−. (.)**
Desert
−. (.)
Cereal suitability
. (.)
State history Agricultural Transition Minimum distance to trade route in Constant No. observations R-squared
. (.) . (.)
. (.)***
. (.)***
. (.)***
. (.)***
−. (.) . (.)***
.
.
.
.
.
.
Notes: Estimation via OLS. Robust standard errors reported in parentheses. *, **, *** = significant at , , and percent respectively.
The Institutional Legacy of Muslim Conquest
development have accumulated less democratic experience. As Hariri (, ) notes, “this general historical argument informs the question of Muslim or Middle Eastern exceptionalism because the Middle Eastern core of the Muslim world had comparatively developed precolonial state institutions.” Columns and in Table . control for the key independent variables from Hariri’s analysis. I first control for his measure of precolonial development, “state history.” Column confirms his main finding that countries with stronger precolonial state institutions tend to exhibit less experience with democracy (although the effect is not statistically significant). Yet, as Hariri () acknowledges state history may be endogenous with Muslim conquest and uses the timing of a country’s transition to the Neolithic Revolution (i.e., when the transition from hunter-gathering to sedentary agriculture took place within a given territory) to identify the exogenous component of state history that is uncorrelated with subsequent Muslim conquest. Using this variable (“Neolithic transition”) as an instrumental variable, Hariri subsequently shows the long-run effect of state history on democracy. Cognizant of this concern, in column , I control for this exogenous component of state history (which predates Muslim conquest). In this specification, conquest retains its negative ðcoefficient ¼ −0:33Þ and highly statistically significant effect ðp-value ¼ 0:00Þ on accumulated democracy. In contrast, Neolithic transition has no discernible effect. Substantively, the results in columns and show that Muslim conquest remains an important determinant of autocracy in the long run while conditioning on a country’s state precolonial history and transition to sedentary agriculture. Moreover, these findings suggest that Muslim conquest, rather than confounders that predate it, remains an important determinant of long-run institutional dynamics in conquered countries. Preconquest Trade A third historical explanation factor is the role of preexisting trade routes in facilitating the spread of Islam through both Muslim conquest and the potential gains from trade (Michalopoulos et al. ). This argument centers on the potential economic gains that individuals experienced by converting to Islam, such as access to trade networks. As such, the trade explanation does not directly explain why new converts to Islam would adopt specific governance practices (e.g., mamluks, iqta) except through outright conquest which directly transplanted these institutions. Nevertheless, failure to account for preexisting trade routes may bias the long-run effect of Muslim conquest on democracy (e.g., through the channel of economic development). Column controls for a
As Hariri (, ) notes, using the timing of the Neolithic Revolution is “based on the simple idea that an early transition to settled agriculture predicts an early development of statehood (see, e.g., Gamble ): Sedentary agriculture permitted increased food production and shortened birth intervals. This increased population density and brought specialization, political organization, and social stratification.”
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
The Autocratic Legacy of Muslim Conquest
country’s minimal distance to trade routes in CE from Michalopoulos et al. (). In these specifications, Muslim conquest remains a robust determinant of autocracy in the long run, while distance to preexisting trade routes has no effect.
B The Subsequent Impact of European Colonialism Recent scholarship in political economy has identified European colonialism as an important determinant of long-run patterns of economic and political development (e.g., Acemoglu et al. , Hariri , Nunn , Olsson ). Since many conquest territories experienced European rule (e.g., France controlled Algeria, Great Britain ruled modern-day Pakistan), these studies suggest that European colonialism may have affected the long-run trajectory of democracy in conquest territories. Moreover, some recent scholarship contends that European colonialism may have strengthened the anti-democratic effects in several conquest societies. Adria Lawrence () and Elizabeth Thompson (), for example, argue how the French colonial administration actively undermined democratic movements in several North African and Middle Eastern colonies during the early twentieth century. From a statistical standpoint, these two factors suggest that failure to account for European colonialism in the empirical analysis may comprise omitted variable bias. Yet, formally controlling for the impact of European colonialism on democracy in Equation (.) is problematic as it introduces posttreatment bias. European states embarked on their colonial endeavors starting in the late fifteenth century, ultimately controlling territories in six continents. This domination occurred after the Muslim conquest; and, as I document in Section ., the institutional legacy of Muslim conquest may have shaped the European governing strategies in colonized conquest societies (e.g., use of indirect rule). Thus, measures of European colonialism may be deemed “dirty” and are typically not included as valid controls (Angrist and Pischke ). Cognizant of this concern, I nevertheless probe whether and how European colonialism may have affected the long-run effect of Muslim conquest on a country’s accumulated democratic experience. To do so, I draw on data compiled by Ola Olsson () that meticulously measures the occurrence (incidence) and duration of Western colonial rule for a global sample of countries. The former measure identifies the specific
For example, Acemoglu et al.’s () empirical findings suggest that in environments hostile to mass European migration inhospitable to European migration (e.g., due to a higher mortality), colonists were more likely to establish “extractive institutions” that concentrated political and economic power among a narrow elite with anti-democratic effects. However, in territories more hospitable to European migration, “neo-Europes” emerged with institutions more conducive for democracy in the long run (e.g., the United States, Canada, Australia, New Zealand). As Olsson documents, previous studies tend to document either the incidence or duration for a narrower set of colonies (see table in Olsson ).
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
The Institutional Legacy of Muslim Conquest
colonizer (e.g., Great Britain, France, Spain, Portugal, Italy, Belgium, Germany), while the latter measure (duration) is especially advantageous as a gauge for the “intensity” of each colonizer’s rule in each of its colonial territories. I then merge this data on the incidence and intensity (duration) of Western colonialism with my own. In the resulting sample, countries (out of ) were colonized: by Great Britain, by France, by Spain, and the remaining () by several European states, such as Portugal, Italy, and Belgium. Among territories that experienced Muslim conquest, were colonized: by Britain, by France, and by Italy. Table . shows the robust negative effect of Muslim conquest on a country’s accumulated democratic experience is not diminished when controlling for the effects of European colonialism. Rather, columns and suggest, that on average, neither the presence nor duration of any European colonial rule seems to have affected a country’s accumulated democratic experience. The coefficient estimates (equal to -. and .) are statistically indistinguishable from zero. However, a colonizer’s identity may matter. Column suggests that former French colonial territories accumulated less democratic experience, while those exposed to rule from other European countries were not necessarily any worse off. Consistent with this inference, column shows that colonies under a longer duration of British rule have experienced more democracy. The coefficient estimate implies that an additional century under British rule is associated with a . index point increase in a country’s accumulated democratic experience. This positive effect is consistent with Olsson’s () finding that any pro-democracy effect associated with European colonialism is driven primarily by colonies under a longer duration of British rule. In contrast, an additional century of French rule reduces a country’s accumulated democratic experience by . index points. This association could be problematic if the institutional legacy of Muslim conquest heightened the likelihood of French rule or extended its duration. This does not seem to be the case: Columns and show that conquest is not robustly associated with either the incidence or duration of French colonial rule. C Explanations from the Contemporary (Modern) Period While accounting for the effects of European colonialism does not seem to diminish the robust negative association between Muslim conquest and a
For Olsson (, ), a Western colony is a “new and lasting political organization created outside Europe by Western countries (countries in Europe excluding Russia but including the Western offshoots United States, Australia, New Zealand, and Canada) from the fifteenth to the twentieth centuries through either invasion and conquest, and/or settlement colonization. Its rulers are in sustained dependence on a geographically remote mother country or imperial center that claims exclusive rights of possession of the colony or in other ways strongly dominates politics in the country.” Great Britain colonized Egypt, India, Kuwait, Pakistan, Qatar, UAE, and Yemen. France colonized Algeria, Morocco, and Tunisia, while Italy colonized Libya.
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
TABLE
. Accounting for the impact of European colonialism
Dependent variable:
Conquest (Mi) MENA Colony
Accumulated democracy
()
()
()
()
−. (.)*** . (.) −. (.)
−. (.)*** . (.)
−. (.)*** . (.)*
−. (.)*** . (.)*
French colony Spanish colony Other colony
British rule French rule Spanish rule
Dummy
Duration
()
()
−. (.)
−. (.)
. (.) −. (.)*** . (.) −. (.)
British colony
Duration of . . . All colonial rule
French rule
. (.) . (.)** −. (.)** . (.) (continued)
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
TABLE
. (continued)
Dependent variable:
Accumulated democracy
()
()
French rule Dummy
Duration
()
()
()
()
. (.)***
. (.)***
.
.
Constant
. (.)
. (.)***
. (.)***
−. (.)* . (.)***
Countries R-squared
.
.
.
.
Other rule
Notes: Estimation via OLS. Robust standard errors reported in parentheses. *, **, *** = significant at , , and percent respectively. In columns –, the dependent variable is a country’s accumulated democratic experience. In column , the dependent variable is equal to if the country was a French colony and zero otherwise. In column , the dependent variable is equal to the number of centuries a country was under French colonial rule. In column , “Colony” is equal to if a country was colonized by any European country and zero otherwise. In column , “All colonial rule” is equal to the number of centuries a country was ruled by any European country. In column , each “colony” variable is equal to if a country was colonized by the corresponding European country (i.e., Great Britain, France, Spain, Other). Other refers to Belgium, Germany, Italy, or Portugal. In column , each “rule” variable is equal to the number of centuries a country was ruled by the corresponding European country.
The Autocratic Legacy of Muslim Conquest
country’s subsequent accumulated democratic experience, it is plausible that explanations from the contemporary (modern) period might matter. In this subsection, I systematically probe the factors (confounders) that scholars have attributed to the lack of the democracy in contemporary Muslim societies. To guide my analysis, I categorize these explanations in four perspectives. The first perspective posits that large, powerful armed forces in contemporary Muslim societies supply governments the coercive capacity (and will) to repress democracy, typically measured with the state’s military expenditures (e.g., Albertus and Menaldo , Bellin ). The second perspective argues that stagnant socioeconomic development underlies the lack of democracy in Muslim societies. Since Seymour Lipset’s () seminal work, economic development is viewed as a strong correlate with democracy. In contrast, the third perspective associates economic and political underdevelopment in many Muslim countries to their oil endowments. In these societies, fiscal reliance on these natural resources helps perpetuate authoritarian political regimes (i.e., the so-called political resource curse, see Ross ). The fourth perspective attributes the dearth of democracy to the somewhat ‘arbitrary’ borders of modern Muslim majority countries – drawn by Britain and France as part of the Sykes-Picot Agreement in – after the fall of the Ottoman Empire. The imposition of these European-made borders created “new” states that divided indigenous communities in some cases and in others, forced separate communities into a single political unit (e.g., Lebanon, Syria). And since societal heterogeneity is often inimical to democracy (e.g., Przeworski , ), the historical legacy of the Sykes-Picot Agreement might explain some of the democratic deficit in Muslim countries. To measure border artificiality, I use the fractal measure of the degree to which borders are straight or squiggly as developed by Alberto Alesina, William Easterly, and Janina Matuszeski (). It is worth emphasizing that many of these competing explanations may be a consequence of Muslim conquest and are therefore, posttreatment. For
The willingness and capacity of governments to maintain high levels of repressive capacity is likely to be endogenous with both the institutional legacy of Muslim conquest (autocracy) and unearned income. The former provides the motivation for governments to maintain a large army, while the latter can comprise a source of funds for repression. For example, Timur Kuran () traces the economic underperformance of Muslim countries to inefficiencies generated by Islamic law (sharia). Relatedly, Jared Rubin () and Ahmet Kuru () attribute these economic inefficiencies to the obstacles imposed by ulema (religious clerics). The following chapter tests this argument with greater empirical and theoretical scrutiny. Like the measures of European colonialism in the previous subsection (which were plausibly affected by Muslim conquest), these contemporary confounders are “dirty” controls.
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
The Institutional Legacy of Muslim Conquest
example, Muslim conquest preceded the introduction of sharia and the prominence of the ulema in Muslim societies. If sharia and the ulema affected the subsequent trajectory of economic development in these societies, then Muslim conquest is causally prior to low levels of economic development in contemporary Muslim societies. Statistically, this means that controlling for posttreatment confounders in Equation (.) will tend to attenuate the effect of conquest on democracy. Table . analyzes the long-run association of Muslim conquest with contemporary democracy taking these competing explanations into account. To account for the coercive capacity perspective, I follow Eva Bellin (, ) and control for the average level of military spending relative to GDP (where a higher share implies greater coercive capacity). Controlling for this explanatory variable in column does not attenuate the robust and negative effect of Muslim conquest on accumulated democracy. In column , I control for two measures of modernization: income per capita and urbanization. As expected, accumulated democratic experience is higher in richer countries, while urbanization has no effect. While the inclusion of these two dimensions of modernization slightly reduces the coefficient on Muslim conquest, it remains substantial and highly statistically significant. To address any anti-democratic effects associated with the resource curse, the specification in column includes an indicator variable that takes the value of if the value of fuel exports is on average more than one-third of total merchandise exports from to . The specifications in columns and control for consequences commonly associated with the Sykes-Picot Agreement: arbitrary borders (column ) and various measures of social heterogeneity (column ). In column , I use Alberto Alesina, William Easterly, and Janina Matuszeski’s () fractal measure of border straightness to proxy for border arbitrariness. Note, the inclusion of this variable reduces the sample to countries. In column , I control for ethnic, religious, and linguistic fractionalization as defined by James Fearon and David Laitin (). Across both specifications, conquest still exhibits a harmful long-run effect on accumulated democracy. Finally, columns and report estimates from two conservative specifications. In column , I control for all confounders, while column does the same but excludes countries from the MENA region. While the coefficient on Muslim conquest is attenuated in these specifications (relative to columns –), it remains negative and statistically significant. Particularly noteworthy is the specification in column which shows that accumulated democratic experience is . points lower in countries outside of the Middle East and North Africa. Taken together, the estimates in columns and show that even conditional on measures of modernization, coercive capacity, natural resource dependence, and fractionalization, accumulated democratic experience is significantly lower in countries conquered partially or fully by Muslim armies during the period of Muslim military conquest.
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
TABLE
. Accounting for explanations from the contemporary (modern) period
Dependent variable:
Conquest (Mi) MENA Log military spending (share of GDP) Log GDP per capita
Accumulated democracy ()
()
()
()
()
()
−. (.)*** . (.) −. (.)
−. (.)*** −. (.)
−. (.)*** . (.)*
−. (.)*** . (.)
−. (.)*** . (.)
−. (.)*** −. (.) . (.) . (.)*** (.) −. (.)** −. (.) −. (.) −. (.) . (.) . (.) .
. (.)*** −. (.)
Urban population (%)
−. (.)
Fuel exporter dummy
. (.)
Border straightness Ethnic Fractionalization Religious Fractionalization Linguistic Fractionalization Constant
No. countries R-squared
. (.)***
−. (.)***
. (.)***
−. (.)
−. (.)** . (.) −. (.) . (.)***
.
.
.
.
.
() Excl. MENA −. (.)*** . (.) . (.)*** (.) −. (.) −. (.) . (.) −. (.) . (.) . (.) .
Notes: Estimation via OLS. Robust standard errors reported in parentheses. *, **, *** = significant at , , and percent respectively. Column excludes observations from the MENA region.
The Institutional Legacy of Muslim Conquest
. The results from the previous section and those in Chapter show that Muslim conquest autocratized political authority during the conquest period (Chapter ) and is associated with less democracy in the contemporary period (Section .). These two central findings empirically corroborate the “starting” and “finishing” positions in Figure .. In this section, I now trace the political effects of the conquest equilibrium by evaluating the figure’s intermediate steps: the effect of Muslim conquest on institutional change (from the premodern to modern period) and its interplay with European colonialism. A Institutional Change since Building on the cross-national statistical analysis in Chapter , I first examine whether greater state autocratization stemming from Muslim conquest affected the trajectory of autocratic institutions thereafter. Table . shows that Muslim conquest is strongly associated with a rigidity of autocratic institutions in the modern era (i.e., since ). To evaluate this, I draw on a measure for the “longrun” change in democracy, meticulously compiled by Acemoglu et al. (). In an attempt to evaluate the modernization hypothesis over the long run, these scholars extend the composite POLITY data series back to and create a dependent variable for the -year change in democracy from to . This variable (“Change in democracy”) is continuous and ranges from (no change in democratic institutions) to (democratic consolidation). Thus, if Muslim conquest generated a trajectory of centralized autocracy – as argued in Chapters and – Mi should be negatively associated with change in democracy. This seems to be the case. In column of Table ., the statistically significant and negative coefficient implies that Muslim conquest reduced a country’s movement change toward democracy since . While the typical nonconquered country experienced a . index point change in democracy, this was dampened by . index point in the typical conquered country ðMi ¼ 0:74Þ. This implies that Muslim conquest reduced the relative long-run change in democracy by about percent. Interestingly, this robust, negative association increases in magnitude when conditioning on a country’s long-run
Acemoglu et al. () study the relationship between income and democracy. They present two key results. First, in specifications that control for time-invariant characteristics (i.e., country fixed effects) they show that income and democracy is uncorrelated in the short run (in the past years). Second, they argue the positive between income and democracy in the cross section of countries is a consequence of historical critical junctures. They argue these critical junctures emerged before/around and accordingly construct a measure of the changes in democracy and income from to to evaluate this long-run relationship. This interpretation follows from the fact that since there were no democracies in , a higher value of change in democracy corresponds to political liberalization. A value of implies a change to a fully consolidated democracy by .
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
TABLE
. Tracing the conquest equilibrium since
Dependent variable:
Conquest (Mi)
Long-run democracy ()
()
−. (.)***
−. (.)*** .
Long-run development
Accumulated democracy ()
()
European settlement
Indirect rule
Constraints at independence
()
()
()
−. (.)***
−. (.)***
. (.)***
−. (.)***
. (.)*** . (.)***
. (.)***
. (.)**
. (.)***
.
.
.
.
(.)***
Constant
. (.)***
. (.)***
. (.)*** . (.)***
No. countries R-squared
.
.
.
European settlement
Notes: Robust standard errors reported in parentheses. **, *** = significant at and percent respectively. In column , the dependent variable is the year change in democracy from to . This variable ranges from to where zero implies no change in institutions and corresponds to a transition to fully consolidated democracy in . In column , long-run development is the -year change in per capita income (GDP). This variable is drawn from Acemoglu et al. (). In column , European settlement is the percentage of the current population that speaks a European language. In column , the dependent variable ranges from to where a higher value corresponds to a greater degree of indirect rule by European colonizers. These dependent variables in columns and are drawn from Hariri (). In column , the dependent variable measures the level of institutionalized constraints on the chief executive’s political authority at the time of the country’s independence. This variable ranges from to , where a higher value corresponds to greater constraints on executive authority. The dependent variables in columns , , and are drawn from Acemoglu et al. ().
The Institutional Legacy of Muslim Conquest
change in per capita income (column ). In this specification, the robust and statistically significant association with income is consistent with modernization theory and Acemoglu et al.’s () contention that economic growth and democratization are correlated in the long run. More importantly, the results in column show that Muslim conquest stunted movements toward democracy over the long run even when accounting for long-run changes in economic development. B European Colonial Strategy Of course, between and the contemporary era, many Muslim societies experienced European colonial rule which may have influenced their institutional trajectory and thus affected their political institutions today. As described in Chapter , whereas Ottoman rule in Muslim societies was absolutist, there was greater variation in the colonial strategy of European powers in the nineteenth and twentieth centuries. In some instances, European colonialism served to “transplant” democratic governance in their colonies, through the emigration of Europeans and/or education of the local populations (Olsson , Woodberry ). Relatedly, other scholars highlight how precolonial institutions shaped the governing strategy of European powers (Fieldhouse , Gerring et al. , Hariri ). First, European colonizers were more likely to govern through indirect rule in countries with strong precolonial institutions. Second, as a result, colonies governed through indirect rule were more likely to maintain their precolonial political (autocratic) institutions. Together, this suggests that the existence of more robust autocratic political institutions in territories conquered by Muslim armies repelled (potentially) the prodemocratic institutional effects of European colonial rule. Specifically, territories conquered by Muslim armies were more prone to indirect European rule and thus, less likely to experience the export of democratic governance, through, for example, European settlement. Columns – evaluate in Table . this conjecture in several ways. Column shows that on average, countries that experienced greater European colonial settlement have experienced more democracy. In this specification, I follow Jacob Hariri () and measure European colonial influence with the fraction of the population that speak as their first language one of the main European colonial powers: English, French, German, Portuguese, or Spanish. Column evaluates whether Muslim conquest mediates this relationship. It demonstrates countervailing effects: Muslim conquest lowers a country’s democratic
I use language to measure settlement instead of descendancy because the former is likely to be measured with less error. As Hariri (, footnote ) notes, “sorting out European descendancy is not only difficult, it also requires a threshold value above which mestizos are counted as European. Such a threshold will always be arbitrary. Simply counting who speaks what language seems to be a more straightforward measure.”
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
The Autocratic Legacy of Muslim Conquest
experience, while European settlement increases it. Importantly, the net effect is negative in countries exposed to Muslim conquest. For the typical conquered country ðMi ¼ 0:74Þ, the net effect is −.. This likely reflects variation in the colonial strategy of European powers. Column shows that countries exposed to greater control by Muslim armies were more likely to experience indirect colonial rule. Concomitantly, these countries experienced lower levels of European settlement (column ). As noted above, since several prominent studies link the intensity of European settlement to the introduction of democratic norms and institutions (e.g., Hariri , Olsson ), columns and imply the institutional trajectory of centralized autocracy introduced through Muslim conquest limited the diffusion of democracy thereafter. Finally, at their independence, countries exposed to Muslim conquest were less likely to adopt institutions that did not place constraints on the executive (column ).
. A Muslim Conquest vs. Islam Despite the findings from the placebo test (see Table . column ), a lingering concern is whether Muslim conquest (Mi) is proxying for the political effects associated with Islam. Several prominent scholars maintain that adherence to the Islamic faith has served as an obstacle to democracy (e.g., Huntington , Lewis ). Central to these arguments is the contention that Islam and domestic politics are intertwined. According to the historian, Bernard Lewis (, ), “In Islam . . . there is from the beginning an interpretation, almost an identification, of cult and power, or religion and the state: Mohammad was not only a prophet, but a ruler. In this respect, Islam resembles Old Testament Judaism and looks quite different from Christianity.” More generally, this
Among conquered territories, the average level of European settlement is . and Ci ¼ 0:74. The net effect is: 0:05∗ 0:296 þ 0:74∗ −0:188 ¼ −0:12. My measure of indirect colonial rule draws on Lange () and Hariri (). I follow Lange (, appendix A) in using the number of colonially recognized customary court cases divided by the total number of court cases as a proxy. The logic is based on the idea that under direct rule, colonial powers would administer a uniform legal system based on the metropole’s laws and not recognize customary courts. However, under indirect rule, the colonial legal system would incorporate the indigenous legal structure. As Lange’s measure is only available for thirtythree former British colonies, I use Hariri’s (, footnote ) method of imputation – using information on European language fraction and the fraction of the population of European descent for the missing data. The dependent variable is from Acemoglu et al. (). It lies on [,] where a higher value corresponds to greater constraints on executive authority. This sentiment has been voiced by scholars from different methodological backgrounds. For example, the political scientist, Mark Tessler (, ), notes that “there is a strong historical connection between religion and politics in the Muslim world, reflecting Islam’s character as a
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
The Institutional Legacy of Muslim Conquest
thesis resonates with prior claims that greater religiosity (at the societal level) stunts socioeconomic development and political liberalization (e.g., Weber []). To address this concern, in this section, I examine whether the Islamic faith and its practice is anti-democratic. This requires separating Islamic theology and practice in everyday life from the institutional legacy of Muslim conquest. In a recent book, Steven Fish () leverages surveys to carefully study how self-identified responses to questions on religiosity affect individual attitudes toward democracy and related norms, such as tolerance. Fish provides compelling evidence that Muslims are not that distinctive in their views toward democracy. In summarizing his findings, Fish concludes that any differences in attitudes toward democracy between Muslim and non-Muslims is “minuscule” and “Muslims, like non-Muslims, tend toward support for democratic institutions, at least in the abstract” (Fish , ). Building on Fish’s research design with survey data, I examine the relationship between democracy and religiosity, but at the societal level. To do so, I construct a country-level religiosity index from the Gallup World Poll taken between and . The index ranges from to (where higher values indicate greater religiosity) and aims to capture “the importance of religion in respondents’ lives.” To construct the index, I average scores from the following three questions. First, “is religion an important part of your daily life?” Second, “have you attended a place of worship or religious service within the last seven days?” Third, “in [country name], do you have confidence in each of the following or not? How about religious organizations (churches, mosques, temples, etc.)?” For each question, positive responses are scored with a value of and other answers receive a score of zero. I create an individual level index by taking the mean of the three scores and multiplying by . I then calculate the mean of this measure for each country and divide by to derive a country (societal) level measure of religiosity. Thus, religiosity varies across country and ranges from (least) to (greatest). In general, the average level of religiosity tends to be higher in Muslim majority societies ð¼ 0:74Þ compared to non-Muslim societies ð¼ 0:61Þ, although there is considerable cross-national variation among Muslim societies ðstandard deviation ¼ 0:14Þ. As noted above, existing scholarship suggests that greater religiosity should be associated with less democratic experience. Figure . provides empirical corroboration in a global cross section of countries. The figure plots each country’s level of religiosity against its level of
religion of law pertaining to societal organization, as well as individual mortality.” Relatedly, in now a famous quip, Samuel Huntington (, ) states, “God and Caesar, church and state, spiritual and temporal authority, have been a prevailing dualism in Western culture. In Islam God is Caesar.” This data source and method of deriving societal metrics has been used in related studies (e.g., Chaney ).
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
The Autocratic Legacy of Muslim Conquest
Accumulated democracy 0.2 0.4 0.6 0.8
1
NEW SWZ CAN NOR UKG AUL BEL SWDDEN NTH FRN FIN JPN AUS GMY GRC
USA IRE
JAM ITA
IND CYPTRI
EST
SAF COL BOT
SRI MAL TUR VEN CHL ECU HON SPN PHI LEB PER NAM BOL PAK POR BNG BRA GUY ROK MAGCOM ARG SIN PANZIM BEN SLO NIG SOM MZM ZAM SEN UGA MLI ANG GUA SI SAL POL SUD GHA CEN NIR HUN LAO THIKEN CON BFO MAW BUI CHA TOG TAZ CAM CUB INS RWA UAE MEX HAI NICNEP CAO MAA LAT SLV ALG GUI TUN LIT PAR KUW QAT MLD ETI IVO BUL MON JOR SYR VIE YEM MOR LBR RUSUKR ARM IRNGRG IRQ BLR KZK AZE KYR TAJ UZB URU
CZR
COS
0
SAU
0.2
0.4
0.6
0.8
1
Religiosity Non-Muslim majority Fitted line for all countries FIGURE
Muslim majority
. Religiosity and democratic experience
accumulated democratic experience, with Muslim societies denoted with an “x” and non-Muslim societies with an “o.” The line of best fit shows a robust negative association: Societies exhibiting greater religiosity tend to have experienced less democracy. Further decomposition of Figure . reveals that the negative association between religiosity and accumulated democracy is not driven by the greater average of religiosity in Muslim societies. This is evident from Figure . which separates this association by whether a country is Muslim-majority or not. The left panel in the figure shows that greater religiosity in non-Muslim societies is associated with less democratic experience. However, the opposite pattern exists in Muslim societies (right panel): Societies exhibiting greater selfidentified religious adherence/practice tend to have experienced more democracy. While this association does not reveal the direction of causality, the positive correlation does strongly suggest that more pious Muslim societies are not inherently less democratic. This is further evidence that Islam and democracy are compatible. Of course, it is possible that the positive association between religiosity and accumulated democracy in Muslim societies may be capturing the institutional effects of the conquest equilibrium. For instance, in the right graph in Figure ., the least democratic societies are those where significant fractions of their modern-day territory were conquered by Muslim armies (e.g.,
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
The Institutional Legacy of Muslim Conquest
Non-Muslim
Muslim
Accumulated democracy 0.5
1
NEW SWZ USA COS AUL CAN NORUKG DENBEL SWD NTH IRE FRN FIN JPN AUS GMY IND ITA GRCJAM CYP TRI SAF SRI COL BOT URU TUR
CHL VEN ECU HON PHI PER BOL NAM POR BRA GUY ROK ARG SIN ZIMMAG PAN CZR SLO MZM NIG ZAM UGA ANG GUA SIE POL SAL GHA CEN HUN LAO THI KEN CON MAW BUI TOG TAZ CAM NEP RWA CUB EST SLV MEX LAT HAINIC CAO LIT PAR MLD ETIIVO BUL MON VIE LBR UKR RUS GRG ARM BLR KZK SPN
MAL
LEB PAK BNG COM BEN SEN SOM MLI SUD NIR BFO CHA INS UAE MAA TUNALG GUI QAT KUW JOR MOR SYR YEM IRN IRQ AZEKYRTAJ UZB
0
SAU
0.2
0.4
0.6
0.8
1 0.2
0.4
0.6
0.8
1
Religiosity Graphs by Muslim majority
FIGURE
. Religiosity and accumulated democracy, by Muslim-majority status
Azerbaijan, Iran, Iraq, Saudi Arabia, Uzbekistan). In contrast, societies exhibiting greater experience with democracy were less likely to be conquered by Muslim armies (e.g., Bangladesh, Comoros, Mali, Senegal). To evaluate whether the Islamic equilibrium is unduly influencing the association (positive) between religiosity and democracy in Muslim societies, I decompose the association by whether a society was exposed or not to Muslim conquest. Figure . plots the association between religiosity and accumulated democracy in Muslim societies by their conquest status. The left graph plots the association across Muslim societies that were not conquered by Muslim armies (and thus not subsequently exposed to mamluk institutions). The right graph plots the association across societies conquered by Muslim armies. These two graphs reveal three important patterns. First, societies not exposed to Muslim conquest tend to exhibit higher societal levels of religiosity ðaverage ¼ 0:83Þ compared to those exposed to Muslim conquest ðaverage ¼ 0:68Þ. The difference in group means ð¼ 0:16Þ is statistically significant, with a p-value less than .. Second, the average level of democratic experience is higher in nonconquest societies compared to conquest societies (i.e., . versus .). This difference in group means ð¼ 0:13Þ is also statistically significant, with a pvalue less than .. The third pattern combines the first two to reveal a critical relationship: The flat line of best fit in each graph shows no apparent association between societal religiosity and democracy across conquest and non-
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
The Autocratic Legacy of Muslim Conquest
Non-conquest
Conquest
MAL
0.6
TUR LEB PAK
Accumulated democracy 0.2 0.4
BNG COM BEN
SEN SOM MLI SUD NIR BFO CHA INS MAA GUI
IRN
UAE ALG TUN QAT KUW JOR SYR MOR YEM IRQ
KYR TAJ AZE UZB
0
SAU
0.4
0.6
0.8
1 0.4 Religiosity
0.6
0.8
1
Graphs by Conquest status in Muslim majority countries FIGURE
. Religiosity and democratic experience in Muslim societies, by conquest
status
conquest Muslim societies. This suggests the institutional legacy of Muslim conquest does not necessarily condition the association between religiosity and democratic experience in Muslim societies. In summary, the first two facts imply that Muslim societies exhibiting greater religiosity are not more predisposed to experiencing less democracy, while the third fact suggests that the institutional legacy of Muslim conquest does not necessarily affect this association.
B Democracy in Muslim Societies over Time The results in the preceding sections show that Muslim conquest is an extremely robust correlate of less democracy in the cross section of countries. That is, Muslim conquest is strongly associated with a stock of less cohesive political institutions (less democracy) in the modern era. In this final subsection, I further decompose this by providing a preliminary exploration of how the institutional legacy of Muslim conquest varies with democracy over time. The
By design, the analysis in this subsection is preliminary as it seeks to provide motivation for more rigorous empirical evaluation in subsequent chapters.
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
0
0.2
POLITY (normalized) 0.4 0.6 0.8
1
The Institutional Legacy of Muslim Conquest
1970
1975
1980
1985
1990
1995
2000
2005
2010
Year Mamluk
Non-Mamluk
FIGURE . Democracy over time in Muslim societies, by their exposure to mamluk institutions
analysis below identifies two main patterns, which subsequent chapters will evaluate with greater empirical scrutiny. The first pattern shows that countries exposed to Muslim conquest (mamluk territories) have been persistently less democratic than non-mamluk territories in the contemporary period. Figure . vividly illustrates this pattern. The figure plots the average annual normalized POLITY score for Muslim majority societies that were exposed to mamluk institutions and those that were not. Using . as a cutoff between autocratic and democratic governance, during the s and s, the typical Muslim society was a dictatorship with an average normalized POLITY score below .. However, this changed after . While Muslim societies exposed to mamluk institutions remained nondemocratic, societies not exposed to mamluk institutions became increasingly more democratic. Focusing on societies exposed to mamluk institutions only, the second pattern identifies differences in the relative political stability of dictatorship by rentier structure. Figure . provides an initial snapshot, comparing conquest societies that produce oil (as defined by British Petroleum) and those that do not. The figure reveals two key patterns. First, despite a slight upward trend
The patterns/facts in Figure . hold when accounting for temporal effects (e.g., the Cold War, common shocks such as changes in oil prices); for example, with year fixed effects in a regression framework.
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
0
0.2
POLITY (normalized) 0.4 0.6 0.8
1
The Autocratic Legacy of Muslim Conquest
1970
1975
1980
1985
1990
1995
2000
2005
2010
Year Oil
Nonoil
FIGURE . Democracy in oil and non-oil-producing Muslim societies exposed to mamluk institutions
in the direction of democracy since the s, oil producers have remained robustly nondemocratic. Second, in contrast, the typical non-oil-producing conquest society has exhibited greater variability in dictatorship. In the s and s, the typical non-oil producer was firmly a dictatorship. Since the s, non-oil producers have become more democratic, although there are periods of transitions back to dictatorship (e.g., –). These two patterns are consistent with Figure . as they show () that conquest societies that produce oil are less democratic than those that do not and () there is greater variation in dictatorship (and political instability) in non-oil producers. In Chapters –, I unpack these dynamics in oil and non-oil-producing conquest societies with greater empirical rigor, attributing much of it to variation in two sources of economic rents: oil revenue and foreign aid.
. In this chapter, I carefully traced the autocratic legacy of the conquest equilibrium in the modern era using statistical analysis. I presented four main results. First, medieval Spanish provinces that experienced longer durations of Muslim
For example, their average normalized POLITY score has never exceeded the conservative threshold for democracy (of .).
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
The Institutional Legacy of Muslim Conquest
rule (that featured mamluk institutions) were slower to move toward representative assemblies. Second, fast forwarding to the modern era, countries exposed to mamluk institutions have experienced less democracy. This finding is very robust. Third, European colonial rule did not fundamentally alter this institutional equilibrium. In colonized Muslim territories, limited European settlement and the use of indirect rule largely preserved existing autocratic institutions and governing coalitions. Fourth, greater religiosity in Muslim societies is unlikely to be the reason why they exhibit dictatorship today. Rather, the institutional legacy of Muslim conquest – not the practice of Islam in these societies – is a more compelling reason for why these societies tend to exhibit less cohesive (democratic) institutions today. Together, the four main results trace the institutional legacy of Muslim conquest: Societies exposed to mamluk institutions (via Muslim military conquest) tend to exhibit less cohesive institutions today. From a methodological standpoint, the empirical approach in this chapter follows that of most quantitative studies of critical junctures and historical legacies: investigating the relationship between the critical juncture and various outcomes from the contemporary era in a cross section of countries (or across other political units, such as subnational provinces). A limitation of this crosssectional approach, however, is its masking of potentially important temporal variation in these contemporaneous outcomes. This was evident in my preliminary exploration of how democracy varies over time in conquered territories (in Section .) and how this can differ by a country’s rentier structure. Specifically, conquest countries that produce oil seem to have experienced greater authoritarian stability relative to non-oil-producing conquest societies. Why has this been the case? Recalling my motivational plots from Chapter (Figures . and .), these different temporal dynamics in dictatorship – and as I will also show, varying intensities of political violence (repression, civil war) – are largely affected by temporal variation in different types of economic rents. The analysis in this chapter and the preceding two collectively argue and empirically demonstrate how societies exposed to mamluk institutions (via Muslim conquest) tend to exhibit a “stock” of less cohesive political institutions today (as proxied using measures of democracy). In Part III of this book, I will argue that variation in rents explains why these societies have been prone to dictatorship and political violence over time.
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
Appendix C Additional Results
The findings in Table . are robust with alternate measures of democracy. Table C demonstrates a pernicious long-run effect of Muslim conquest on democracy using measures from Boix, Miller, and Rosato (), Freedom House (), and POLITY IV (Marshall and Jaggers ). Whereas accumulated democracy is a “weighted” democracy score of the POLITY index (Gerring et al. ), these alternate measures are based on different coding schemes. For example, the Boix et al. () measure is dichotomous and based on two conditions: contestation and participation. Specifically, democracies feature political leaders chosen through free and fair elections and satisfy a threshold value of suffrage. A country that does not satisfy any of these conditions is not democratic. Unlike the POLITY measure of democracy, the TABLE C The long-run effect of Muslim conquest on alternate measures of democracy
Dep. variable:
Conquest MENA Countries R-squared
Boix
FH
POLITY
XCONST
PARCOMP
()
()
()
()
()
−. (.)*** −. (.)
−. (.)*** . (.)
−. (.)** −. (.)
−. (.)*** . (.)
−. (.)** −. (.)
.
.
.
.
.
Note: Estimation via OLS. Robust standard errors reported in parentheses. **, *** = significant at and percent, respectively. All specifications include a constant which is not reported.
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
The Institutional Legacy of Muslim Conquest
concentration of power held by the executive (XCONST) is not explicitly relevant for Boix’s categorization. The Freedom House measure takes three values – free, partially free, and not free – based on the quality of political rights and civil liberties. For each of these series, I average each country’s annual value over the period – and then scale them on a [,] interval where a higher value corresponds to more democratic governance/institutions. In columns – of Table C, the coefficient on conquest is statistically significant and ranges in magnitude from −. to −., while the effect on the MENA dummy remains statistically insignificant. Using two sub-indices of the overall POLITY measure (in column ), columns and show that Muslim conquest lowers checks on the executive (XCONST) and reduces political competition (PARCOMP).
A large body of research finds that fiscal capacity is strongly correlated with contemporary democracy; and this capacity often has deep historical origins. For example, the strengthening of fiscal capacity lies at the heart of Charles Tilly’s () theory of state formation, particularly for European states from the eleventh century onward. In order to preserve (and often expand) their territorial integrity, European sovereigns had to develop the capacity to extract financial resources to thwart both internal and external threats. Thus, fiscal capacity co-evolved with statehood. This insight underlies more recent theoretically refined and empirically rigorous scholarship on state formation/capacity, including treatments by Timothy Besley and Torsten Persson (a), Mark Dincecco (), Margaret Levi () and Kenneth Scheve and David Stasavage (). Fiscal revenues also feature in state development in Muslim societies. However, as I described in Chapter , the effective control of land rents via iqta in conquered territories diminished the necessity of developing an effective taxation system, particularly from non-land based economic activity (i.e., “capital” in Tilly’s framework). This suggests that Muslim conquest is unlikely to be associated with indicators of effective fiscal capacity in the contemporary era. I explore this conjecture in two ways. First, I examine whether Muslim conquest is a strong predictor of a state’s tax effort and fiscal transfers. Second, I evaluate whether controlling for tax effort and fiscal transfers diminishes the long-run association between Muslim conquest and accumulated democratic experience. Table C reports the findings. Following convention (as described in Chapter ), I measure a state’s tax effort with the share of government revenues collected from taxes. A higher share implies a greater exertion of tax effort. To measure a state’s fiscal transfers, I use a government’s spending on social transfers as a share of its total expenditures. As described in Chapter , a higher share is associated with more cohesive institutions (democratic governance). Columns and in Table C show that
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
The Autocratic Legacy of Muslim Conquest TABLE C
Muslim conquest and fiscal capacity
Dependent variable:
Conquest (Mi)
Tax revenue
Fiscal transfers
()
()
−. (.)
−. (.)
MENA Tax revenue
Accumulated democracy ()
()
−. (.)*** . (.) . (.)
−. (.)*** . (.)
Constant
. (.)***
. (.)***
. (.)***
. (.)** . (.)***
No. countries R-squared
.
.
.
.
Fiscal transfers
Notes: Estimation via OLS. Robust standard errors in parentheses. ** and *** = significant at and percent respectively. Tax revenue is the average between and of tax revenue (% total government revenue). Fiscal transfers is the average between and of social transfers (% total government expenditures). These data are from the World Development Indicators ().
Muslim conquest is negatively correlated with tax effort and fiscal transfers, but the association is statistically indistinguishable from zero. These null effects are consistent with the conjecture that the legacy of mamluk institutions (proxied with Mi) has not affected a state’s contemporary fiscal capacity. Moreover, columns and reveal that controlling for fiscal capacity has not attenuated the robust, negative association between conquest and accumulated democratic experience. In column , tax effort has no effect on democracy, while in column , greater fiscal redistribution is positively and robustly correlated with democracy (as existing theories would suggest).
As discussed in Chapter , skeptics may worry that Hugh Kennedy’s () maps of Muslim conquest may be mis-measured and my measure of conquest (based on digitizing those maps) may introduce measurement error. I addressed those concerns by creating several indicator variables that capture the intensity of Muslim conquest. The first, “Any conquest” is equal to if any share of a country’s territory was conquered by Muslim armies and zero otherwise. The second, “Half mamluk” is equal to if more than percent of a country’s territory was conquered by Muslim armies and zero otherwise. The third, “Full mamluk” is equal to if a country’s entire territory ( percent) was conquered by Muslim armies and zero otherwise. Table C shows that Muslim
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
TABLE C
The Institutional Legacy of Muslim Conquest Muslim conquest (alternate measures) and democracy
Dependent variable:
Accumulated democracy ()
()
()
Any −. conquest (.)*** Half −. mamluk (.)*** Full −. mamluk (.)*** MENA −. . . (.) (.) (.) Constant . . . (.)*** (.)*** (.)*** Countries R-squared
.
.
.
()
() Exclude MENA
()
−. (.)*** −. (.)*** −. (.)*** . . . (.)*** (.)*** (.)*** .
.
.
Notes: Estimation via OLS. Robust standard errors reported in parentheses. *, **, *** = significant at , , and percent respectively. “Any conquest” is equal to if any percentage of a country’s territory was conquered by Muslim armies and zero otherwise. “Half mamluk” is equal to if more than percent of a country’ s territory was conquered by Muslim armies and zero otherwise. “Full mamluk” is equal to if a country’s entire territory ( percent) was conquered by Muslim armies and zero otherwise. In columns –, the sample excludes countries from the Middle East and North Africa.
conquest exhibits a negative and highly statistically significant effect on a country’s accumulated democratic experience across a global sample (columns –) and a sample that excludes countries from the Middle East and North Africa (columns –). Moreover, the coefficient estimates are similar in magnitude, ranging from −. to −..
As Chapter demonstrated, the harmful effect of Muslim conquest on contemporary democracy remains robust in specifications that account for three prominent competing historical explanations (agriculture, precolonial state capacity, trade). Nevertheless, if Muslim conquest is correlated with these preconquest conditions, the estimated effects in this chapter’s tables may be biased. To allay this potential concern with endogeneity, I employ a country’s distance to Mecca as an instrumental variable for the percentage of territory conquered by Muslim armies and subsequently governed by mamluk institutions (Mi).
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
The Autocratic Legacy of Muslim Conquest
The instrumental variable exploits geographic variation in how conquest unfolded and satisfies the “validity” and “exogeneity” requirements for a strong instrument. With respect to instrument validity, military conquest started first by unifying the Arabian Peninsula – beginning from Mecca – and then by conquering territories in Mesopotamia, North Africa, the Iberian Peninsula, and parts of Central and South Asia in a largely radial fashion (Lapidus , –). Consequently, territories geographically closer to Mecca tended to experience greater penetration/control by Muslim armies. With respect to exogeneity, distance to Mecca is plausibly uncorrelated with the competing historical factors discussed in the previous subsection, as the location of Mecca is exogenous to preconquest political and economic conditions in other territories. With regards to the exclusion restriction, conditional on Muslim conquest – and its economic and political legacy – a country’s distance to Mecca should not have an a priori effect on its level of democracy in the contemporary period. Armed with this instrument, I estimate variants of the following set of twostage least (SLS) regressions: First stage : Mi ¼ α þ θDIST i þ κXi þ εi Second stage : Di ¼ α þ βMi þ φXi þ εi where in the first stage regression, DISTi is country i’s distance from Mecca (in , kilometers) and all other variables are the same from Equation .. Given the geographic terrain and prevailing modes/technologies of warfare in the seventh and eighth centuries (as described in Chapter ), the ability of Muslim armies to conquer territories declined as they moved farther from the Arabian Peninsula. Consequently, the percentage of conquered territory ðMi Þ is expected to decline with DISTi. This implies θ should be negative in the first stage regression. In the second stage regression, I insert “predicted values” for Mi (from the first stage regression) to generate an unbiased causal estimate of Muslim conquest on contemporary democracy. Table C reports the SLS results corresponding to first and second stage equations reported above. For each specification (in the columns), the effect of instrumented conquest on accumulated democracy is reported in Panel A, while Panel B reports the estimated effect of DISTi on Mi in the corresponding first stage regression. Column reports a sparse bivariate regression. The corresponding first stage regression (column b) shows that Muslim armies conquered less territory in countries farther away from Mecca. The coefficient estimate ð¼ −0:45Þ implies that, on average, Muslim countries conquered percent less territory in a country , kilometers from Mecca relative to countries very close to Mecca. Reassuringly, the instrumental variable is a “strong” predictor of Muslim conquest as the F-statistic passes the appropriate threshold value of (Stock and Staiger ). Using this instrument, the corresponding second stage estimate in column a shows that countries fully
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
TABLE C
SLS estimates with distance to Mecca as an instrumental variable Panel A: Second stage regression
Dep. variable:
Conquest (Mi)
Accumulated democracy ()
()
()
()
()
()
()
−. (.)***
−. (.)*** −. (.)**
−. (.)***
−. (.)***
−. (.)***
−. (.)*** . (.) −. (.)** −. (.)** . (.)*** −. (.)***
−. (.)** . (.) −. (.)** −. (.)*** . (.)*** −. (.)***
.
Yes .
Religiosity
−. (.)
Alcohol
. (.)
Fuel exporter Log GDP per capita Constant Continent FE No. countries R-squared
. (.)***
. (.)***
. (.)***
. (.)***
. (.)*** −. (.)***
.
.
.
.
.
Panel B: First stage regression Dep. variable:
Distance to Mecca (, KM) F-statistic
Conquest (b)
(b)
(b)
(b)
(b)
(b)
(b)
−. (.)*** .
−. (.)*** .
−. (.)*** .
−. (.)*** .
−. (.)*** .
−. (.)*** .
−. (.)*** .
Notes: Estimation via SLS. Panel A reports the second stage regression. Panel B reports the corresponding first stage regression with the same controls as the first stage. The controls in Panel B are not reported. Robust standard errors reported in parentheses. *, **, *** = significant at , , and percent respectively.
The Autocratic Legacy of Muslim Conquest
conquered by Muslim armies ðMi ¼ 1Þ have accumulated . point less of democratic experience compared to a non-conquered country. This instrumented effect is larger than the corresponding OLS estimate in Table ., column , which suggests the OLS estimates are downward biased. As with all instrumental variable analysis, potential violations of the exclusion restriction can bias the SLS estimates. In this case, might distance to Mecca affect contemporary democracy through other variables, independent of Muslim conquest? Skeptics may argue that distance to Mecca proxies for Middle East (or Muslim) culture, and that these variables may be driving the anti-democratic effect. Columns – account for two proxies for Arab/Muslim culture: religiosity and alcohol consumption. Accounting for these measures of Arab/Muslim culture does not reduce the robust negative effect of instrumented conquest on accumulated democracy. In these specifications, countries with higher per capita consumption of alcohol are not more (or less) likely to be democratic, while greater religiosity is negatively associated with less accumulated democracy. As I demonstrated in Section ., this effect is unlikely to be driven by greater religiosity in Muslim societies. Of course, it is plausible that distance to Mecca may be correlated with other variables that affect contemporary democracy. For instance, since many Muslim oil-producing countries are located near the Arabian Peninsula, the instrument may be identifying the effect of natural resource wealth on democracy (i.e., the political “resource curse”). Column shows that controlling for natural resource wealth does not diminish the negative effect of instrumented conquest on democracy. More generally, instrumented conquest remains robust when controlling for a country’s level of economic development in column . Consistent with the modernization hypothesis, countries with greater per capita income tend to exhibit greater democracy. Finally, column controls for all these covariates and with continent fixed effects as well in column . In both specifications, instrumented Muslim conquest exhibits a robust, negative effect on accumulated democracy. Taken together, the results in Table C show that distance to Mecca is a valid instrumental variable for Muslim conquest and its deleterious effect on democracy in the long run.
Religiosity varies from (least) to (greatest) and is based on the aggregation of survey data from the Gallup World Poll. I describe this variable’s construction in greater detail in Section .. Alcohol consumption is the country’s total average alcohol consumption per capita in liters of pure alcohol. This information is drawn from the World Health Organization Global Status Report on Alcohol and Health.
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
https://doi.org/10.1017/9781009367509.007 Published online by Cambridge University Press
PART III
THE IMPACT OF CONTEMPORARY RENTS ON DICTATORSHIP AND VIOLENCE
https://doi.org/10.1017/9781009367509.008 Published online by Cambridge University Press
https://doi.org/10.1017/9781009367509.008 Published online by Cambridge University Press
Fueling Authoritarian Resilience
Cumulatively, the analysis in Part II argued and provided evidence that Muslim conquest fostered a trajectory of centralized autocratic political structures in “conquest territories,” which in turn has contributed to persistently less cohesive political institutions (i.e., less democratic) in many contemporary Muslim societies. This exercise served two central purposes that followed from the analytical framework developed in Chapter . First, the autocratic legacy of Muslim conquest provides a plausible explanation for the persistent democratic deficit in the cross section of countries between Muslim and nonMuslim societies identified in Figure .. Second, in documenting this relationship, Muslim conquest (and its institutional legacy) can serve as a plausibly exogenous source of less cohesive institutions in the contemporary Muslim societies corresponding to a low value of θ in the formal model advanced in Chapter . In that model, θ was exogenous – or casually prior – to a society’s prevailing level of rents. Moreover, as that model demonstrated, in societies with less cohesive institutions, variation in the levels of rents can affect temporal patterns of political violence. As suggested by Figures .–., rents seem to be correlated with temporal patterns of political violence observed in many contemporary Muslim societies: An increase in rents can entrench dictatorship, while a subsequent decline in rents can elevate the incidence of civil war. The aim of Part III of this book is to better substantiate these two claims. Part III strives to show how the institutional legacy of Muslim conquest (corresponding to low θ) has shaped how rents affect political violence in many contemporary Muslim societies over time. The empirical analysis largely corroborates the temporal patterns in dictatorship and civil war depicted in Figures .–.. In this opening chapter of Part III, I examine how oil wealth has fueled authoritarian resilience – dictatorship without significant outbreaks of civil war – in oil producers (petrostates) exposed to Muslim conquest. For scholars
https://doi.org/10.1017/9781009367509.009 Published online by Cambridge University Press
How Rents Affect Dictatorship and Violence
and policymakers, understanding these political effects is substantively important: Petrostates fully conquered by Muslim armies comprise around percent of the world’s total production of oil. While this chapter corroborates existing research that oil rents can hurt democracy, it makes four original contributions relevant for understanding why oil wealth has engendered authoritarian resilience in conquest petrostates. First, I identify a tension in existing scholarship: the negative relationship between oil wealth and democracy but positive relationship between oil wealth and civil war, particularly in dictatorships. My discussion in Section . clarifies the conditions under which greater oil wealth hurts democracy but may not elevate the incidence of civil war. Second, as I empirically document in Section ., this clarification is important in understanding why greater oil wealth in conquest petrostates has hurt democracy but has not fostered civil war. Cognizant that endogeneity may muddle causal inference, in this section I present estimates using an instrumental variable for oil wealth stemming from plausibly exogenous shocks to the world supply of oil. Third, I introduce a type of rent that is intricately tied to oil production in many conquest petrostates. In Section ., I argue that an implied security guarantee from the United States government – over both Democratic and Republican administrations – has comprised an additional geopolitical rent which has augmented the military capacity of conquest petrostates to thwart insurgency. Together, the empirical evidence in Sections . and . show that oil-based and geopolitical rents have been two important drivers of authoritarian resilience in conquest petrostates. However, these rent streams have not necessarily been a source of authoritarian resilience in non-conquest petrostates. The chapter’s fourth contribution shows that oil wealth has not hindered a trajectory toward greater democracy in non-conquest petrostates. In Section ., I motivate this discussion with cases of democratic transitions in Venezuela and Nigeria. For a larger sample of countries, I then show that non-conquest petrostates that started out as dictatorships in the s, on average, became democratic by the early s. This democratic trajectory suggests a pathway from dictatorship to democracy in petrostates where preexisting institutions are more cohesive. In contrast, where preexisting institutions are less cohesive – such as petrostates exposed to Muslim conquest – this pathway seems to have been effectively closed off.
This percentage increases if partially conquered territories are included, such as Azerbaijan, Turkmenistan, and Uzbekistan. All these calculations are based on annual data (–) from the Energy Information Agency at the US Department of Energy. Nigeria is an interesting case as it is a Muslim majority country that did not experience Muslim military conquest (and thus not exposed to mamluk institutions). This example shows that a Muslim majority petrostate can be democratic. Other examples include Indonesia and Malaysia.
https://doi.org/10.1017/9781009367509.009 Published online by Cambridge University Press
Fueling Authoritarian Resilience
. A Prevailing Findings There is a voluminous literature relating natural resource wealth to political violence. While Chapter discussed several prominent works on this topic (e.g., Dunning , Karl , Menaldo , Morrison , Ross ), in this section I focus on two key findings and highlight a tension in existing research; a tension that is important in studying the impact of oil wealth on dictatorship and civil war in Muslim societies. In a recent review article, Michael Ross () distills and summarizes the “hundreds” of academic studies evaluating whether natural resource wealth is a source of political dysfunction. He argues that there is “strong evidence that one type of resource wealth – petroleum – has at least three important effects: It tends to make authoritarian regimes more durable; it leads to heightened corruption; and it helps trigger violent conflict in low- and middle-income countries, particularly when it is located in the territory of marginalized ethnic groups. The effects of authoritarianism and conflict appear to be recent phenomena, emerging after the s” (Ross , ). The first two effects of authoritarian durability and corruption coincide with predictions from prevailing theoretical works that nontax income provides sufficient revenue for an incumbent to “buy” political support through patronage, particularly in dictatorships (Bueno de Mesquita and Smith , Morrison ). More generally, these two effects are byproducts of the rentier effect, whereby ample oil revenues enable incumbents to reduce taxes, increase patronage (and public goods) and thus, empowering them to buy off a larger set of potential challengers and reduce dissent (Mahdavy , Ross ). Many studies have empirically corroborated that greater oil wealth strengthens nondemocratic politics. The earliest quantitative studies established this association cross-nationally (Barro , Ross ), with more recent studies replicating this finding using better data and increasingly sophisticated methods. For instance, the negative relationship between oil and democracy is robust to the inclusion of country fixed effects (Andersen and Ross , Aslasken ) and holds with instrumental variables to predict oil revenues, such as those exploiting natural disasters (Ramsay ) and “random” oil discoveries (Tsui ). A couple of studies explore the importance of oil in relation to other factors associated with democracy. For instance, Martin Gassebner, Michael Lamla, and James Vreeland () use extreme bound analysis to show that oil dependence is one of the few robust correlates of regime types. Anar Ahmadov () employs statistical meta-analysis of empirical estimates (from twenty-nine individual studies) to conclude that oil has a negative, nontrivial, and robust effect on democracy. In contrast, the relationship between oil wealth and democracy is less clear. Some studies argue that oil revenues foster prodemocratic effects in
https://doi.org/10.1017/9781009367509.009 Published online by Cambridge University Press
How Rents Affect Dictatorship and Violence
democracies, either by improving their democracy scores or by stabilizing democratic regimes (e.g., Dunning , Morrison , Tsui ). Other studies, however, contend that oil has no effect on stabilizing democratic regimes (e.g., Andersen and Aslasken , Caselli and Tesei , Ulfelder ). Another set of studies identify conditional effects: Oil can stabilize democracies that are wealthy and have strong ex ante democratic institutions but weaken accountability in democracies that are poor and have weaker institutions (Jensen and Wantchekon , Ross ). Conditional effects have also been identified across dictatorships. For example, Anne Boschini, Jan Pettersson, and Jesper Roine () contend the degree to which oil helps dictatorships depends largely on the ruler’s ability to “capture” the available oil wealth. Greater capture exacerbates the oil-autocracy nexus. Michael Ross () situates this argument in a historical context, arguing that oil became antidemocratic in the late s after most oil-endowed developing countries nationalized their petroleum industries. This granted political leaders far greater access to revenues from the sale of oil. His analysis shows that from to , oil- and non-oil-producing countries were equally likely to transition to democracy. However, after , non-oil-producing countries were almost three times more likely to democratize. The stabilizing role of oil in autocracies, however, is seemingly at odds with the third prevailing finding in the political resource (oil) curse literature: Oil fosters civil conflict in poor countries, especially those with less cohesive institutions. As Ross (, ) summarizes “most published studies identify a harmful effect, albeit a conditional one.” Regarding these conditional factors, scholars consistently identify two: a nonmonotonicity associated with the level of resource wealth and its location. The former posits the relationship between oil wealth and civil war risk resembles an inverted U: As the value of resource wealth increases, the risk of violence rises, then falls (Basedau and Lay , Bjorvatn and Naghavi , Collier and Hoeffler ). In explaining this relationship, Collier and Hoeffler () contend that higher levels of resource wealth stabilize regimes by enabling the government to either buy off or repress potential rebels. The willingness of governments to engage in these spending patterns may be heightened in nondemocratic polities. Several formal models – including the one developed in Chapter – generate this prediction (e.g., Besley and Persson , , a, b, Bueno de Mesquita and Smith , Smith
Relatedly, recent studies at the subnational level find that oil wealth can lengthen the tenure of leaders in democracies and hybrid autocracies, such as Brazil (Brollo et al. ), India (Asher and Novosad ), Iran (Mahdavi ), and the United States (Goldberg et al. ). Paasha Mahdavi () argues the willingness of an autocrat to capture oil wealth (e.g., via nationalization of oil production) is largely driven by consideration for his political survival. A separate literature probes whether natural resource scarcity affects civil war (e.g., Gleditsch , Koubi et al. ).
https://doi.org/10.1017/9781009367509.009 Published online by Cambridge University Press
Fueling Authoritarian Resilience
) and empirically is the case in conquest petrostates (as I demonstrate shortly). The second conditional factor highlights how the location of oil matters. Here the distinction centers on whether the oil is drilled on or offshore. For example, in a cross-national study, Päivi Lujala () finds that when oil is found offshore it does not exhibit a robust effect on a country’s propensity for civil war. However, when it is found onshore, oil heightens the propensity for armed conflict. Subsequent studies have unpacked this finding further by exploring the precise location and confounding local “conditions” of onshore oil. Several studies find that oil is more likely to incite conflict when it is found in regions that are populated by marginalized/excluded ethnic groups (Basedau and Richter , Hunziker and Cederman ) and/or are poor relative to the national average (Østby et al. ). In these settings, which often overlap (i.e., marginalized ethnic groups live in poorer regions), the out of power group has a greater incentive to rebel against the central government; particularly in societies with less cohesive institutions in which the incumbent may not distribute state resources to the opposition.
B A Tension The two prevailing findings discussed above are seemingly in tension with each other. On the one hand, greater oil wealth can be a source of authoritarian durability. Yet, on the other hand, oil revenues can incite civil war, particularly in societies where state resources may not be distributed (equitably) to excluded/opposition groups. I am not the first to recognize this tension. In a recent article, Jeff Colgan () also finds these prevailing findings “puzzling” and argues the relative stability of authoritarian petrostates stems from the incumbent’s ability to better capture more of the benefits associated with oil wealth. According to Colgan (, ), “this is because oil generates financial resources that can be used by both rebels and an incumbent government to fund armed conflict, and an incumbent government typically has greater access to these resources, which confers a significant advantage over rebel challengers. The military incumbency advantage closes one path to democracy.” While Colgan does not formalize his argument in a model, its main implication is broadly consistent with a key proposition from the formal model in Chapter : Rents increase an autocrat’s repressive capacity to effectively thwart violence from opposition groups. Recognizing this tension is more than a theoretical exercise. It is crucial to understanding the greater relative authoritarian stability of petrostates exposed to the institutional legacy of Muslim conquest versus those that were not (i.e., conquest versus non-conquest petrostates). Table . summarizes this tension
In recent years, about percent of global oil production is offshore. The rest is onshore.
https://doi.org/10.1017/9781009367509.009 Published online by Cambridge University Press
TABLE
How Rents Affect Dictatorship and Violence . Political stability in nondemocratic petrostates
Conquest petrostates Non-conquest petrostates
Normalized POLITY
Civil war
. (.) . (.)
. (.) . (.)
Notes: Group means calculated for a sample of petrostates that were nondemocratic in the s. Corresponding standard deviations reported in parentheses. Each row reports the average value (column) for the associated type of petrostate (row).
in the underlying raw data for a sample of nondemocratic petrostates. The group means reported in each row of the table reveal that conquest petrostates are comparatively less democratic and less prone to civil war compared to nonconquest petrostates. The latter finding is particularly surprising since a large share of oil production in conquest petrostates are onshore, and thus, predicted to be more vulnerable to attacks from anti-government groups. Section . provides more stringent empirical evaluations of these patterns, while Section . strives to resolve the tension by identifying an additional form of rent that has helped ensure dictatorship in many Muslim oil producers.
. A Oil Wealth Hurts Democracy Motivated by the patterns in Table ., in this section I examine the relationship between oil wealth and democracy. The analysis will show that oil wealth is negatively associated with democracy, with larger effects larger in conquest petrostates relative to non-conquest petrostates. This finding is consistent with proposition in Chapter that rents in societies where preexisting institutions are less cohesive are more likely to entrench dictatorship. I follow the standard cross-country empirical setup across a sample of oil producers. I estimate the following specification: Dit ¼ α þ βRit þ φXit þ Y t þ εit
ð6:1Þ
where Dit is country i’s level of democracy in year t, Rit measures its annual oil revenue, Xit is a vector of controls (e.g., per capita GDP, economic growth, regional dummies), Y t are year fixed effects, and ε is an error.
Some econometric studies examining the relationship between oil and democracy include country fixed effects to control for unobserved country-specific factors and to focus on within-country variations over time rather than cross-country variations (e.g., Haber and Menaldo a). However, many oil-producing countries – particularly those outside the richest regions – have experienced very little within-country variation in democracy over time. Indeed, as Michael Ross
https://doi.org/10.1017/9781009367509.009 Published online by Cambridge University Press
Fueling Authoritarian Resilience
The dependent variable is a country’s normalized polity score on [,] scale where a higher value corresponds to a greater quality of democracy. The key independent variable, Rit , is a country’s per capita oil income in US dollars (logged). I follow Kris Ramsay () and construct this measure by multiplying a country’s annual production of oil in barrels with the average daily spot price of crude oil (as reported by BP’s statistical review); and then divide by its population. The inclusion of year fixed effects accounts for any temporal shock that confront all oil producers, including annual variation in oil prices and prevailing geopolitical conditions. The data is drawn from POLITY-IV, the World Bank’s World Development Indicators, and BP’s statistical review. Table . reports the relationship between per capita oil wealth and democracy. Column reports a statistically, robust negative relationship: A one standard deviation increase in per capita oil wealth (corresponding to about $ dollars) reduces democracy by −. points. This is equivalent to a . standard deviation reduction in the normalized score among oil producers. Controlling for per capita income and economic growth doubles the estimated negative effect of oil wealth on democracy (column ). Consistent with modernization theory, countries with higher per capita income tend to be more democratic, while annual changes in per capita income (growth) exhibit a negative but imprecisely estimated effect on democracy. Skeptics may ask whether unobserved characteristics associated with Muslim societies – particularly those situated in the Middle East and North Africa – are unduly driving the results. To allay these concerns, column
(, ) notes, “these tasks are desirable, [but] the decision to include country fixed effects creates an even larger problem: It eliminates from the sample all countries that have never transited to democracy, a group that includes most of the oil-rich developing countries. The result is a severe selection bias: Countries that have transited to democracy remain in the sample, while countries that have not transited are dropped.” The results are robust using rents as a share of GDP. While there is no single best measure of oil wealth, normalizing a country’s oil income by population mitigates potential concerns with endogeneity. As Ross (, ) notes: “Some measures are endogenous to the outcomes they purportedly explain. For example, one common measure oil export dependence, meaning petroleum exports as a fraction of GDP – is probably biased upward in poorer and more conflict-prone countries because they are typically too impoverished to consume the fuel they produce, causing them to export more than they otherwise would. On a per capita basis, the United States produces more oil than Nigeria, but Nigeria is an exporter whereas the United States is not because the United States is wealthier and consumes all its oil domestically. This makes it hard to interpret, for example, correlations between oil export dependence and the frequency of violent conflict; both might be independently boosted by a country’s poverty.” In the full sample, the standard deviation for log oil revenue per capita is .. Thus, the estimated effect is .*‒. = ‒.. The standard deviation for the normalized polity score is ..
https://doi.org/10.1017/9781009367509.009 Published online by Cambridge University Press
https://doi.org/10.1017/9781009367509.009 Published online by Cambridge University Press
TABLE
. Oil wealth harms democracy
Dependent variable
Log oil revenue per capita
Democracy ()
()
()
()
()
−.
−.
−.
−.
−.
Non-conquest −.
Conquest −.
−.
(.)***
(.)*** .
(.)*** .
(.)*** .
(.)*** .
(.)*** .
(.)*** .
(.)*** .
(.)*** −. (.)
(.)*** −. (.)* −. (.)***
(.)*** −. (.)**
(.)*** −. (.)** −. (.)* −. (.)***
(.)*** −. (.)***
(.)** −. (.)***
(.)*** −. (.)***
Log GDP per capita Growth MENA
−. (.)***
Conquest
()
()
. (.)*** −.
Log oil rev. per capita x Conquest Year FE No. observations R-squared
()
(.)*** Yes .
Yes .
Yes .
Yes .
Yes .
Yes .
Yes .
Yes .
Notes: Robust standard errors, corrected for autocorrelation in parentheses. *, **, *** = significant at , , and percent respectively. All specifications include year fixed effects. These coefficients and a constant are not reported. In column , the sample is restricted to petrostates that did not experience Muslim conquest. In column , the sample is restricted to petrostates that were partially or fully conquered by Muslim armies.
Fueling Authoritarian Resilience
controls for the catchall measure of “Middle East exceptionalism” with a MENA dummy, while column controls for the percentage of territory conquered by Muslim armies between – (Mi from Chapters and ). Column controls for both measures. The results in columns – show that while countries in the MENA region and those that experienced greater conquest exhibit lower levels of democracy – a finding consistent with the analysis in Part II of this book – the main effect of oil wealth on democracy remains negative and statistically significant. There are two important implications from the specifications in columns –. First, the negative relationship between oil and democracy is a phenomenon that exists beyond Muslim societies and countries in the MENA region. Oil harms democracy in non-Muslim and Muslim societies. Second, the results from column show the estimated negative effect of Muslim conquest is four times larger than that associated with the MENA dummy (−. versus −.). Thus, while geography “matters,” the institutional legacy of Muslim conquest matters more in explaining lower levels of democracy. Columns – investigate whether the institutional legacy of Muslim conquest affects the relationship between oil wealth and democracy. I begin with a split sample approach comparing the effect of oil wealth on democracy in countries that did not experience Muslim conquest (column ) versus those that did (column ). Across both samples, greater per capita oil wealth is associated with a lower level of democracy. However, the estimated negative effect is much larger in conquest countries, suggesting the autocratic institutional legacy of Muslim conquest is heightening the harmful effect of oil wealth on democracy. Column shows this more directly. It introduces the interaction of a country’s per capita oil wealth and the percentage of its territory conquered by Muslim armies (from to ) as an additional predictor for democracy. The interaction term is negative, large in magnitude, and statistically significant. Figure . plots the associated marginal effect on democracy as conquest exposure varies. The figure shows that countries with greater oil wealth, whose modern territory experienced increasingly greater control by Muslim armies, are significantly less democratic. For example, for a log unit change in per capita oil revenue, the average conquered country (with Ci ¼ 0:77) is about . index points less democratic than a non-conquered country. At the extremes, a fully conquered country ðCi ¼ 1Þ is about −. index points less democratic than a non-conquered country.
The estimated effects on the interaction’s constituent terms are informative as well. The negative coefficient on oil wealth ð¼ −0:041Þ implies that oil wealth reduces democracy in non-conquered countries. The positive effect on conquest implies the average level of democracy tends to be higher across conquered territories that do not produce oil. This implication is wholly consistent with the final figure in Chapter .
https://doi.org/10.1017/9781009367509.009 Published online by Cambridge University Press
Marginal Effect of Log oil rev. per capita on Normalized POLITY –0.15 –0.1 –0.05
0
How Rents Affect Dictatorship and Violence
0
FIGURE
0.2 0.4 0.6 0.8 Percentage of territory controlled by Muslim armies
1
. Marginal effect of oil wealth on democracy as conquest varies
B Causal Estimates The results in Table . provide evidence that greater oil wealth is negatively associated with less democracy, particularly in countries exposed to the institutional legacy of Muslim conquest. This association, however, may not be causal due to the inherent challenges of endogeneity. For example, this can stem from reverse causality: A country’s regime type can affect the intensity of oil exploration and subsequent levels of oil wealth. Democracies, for example, tend to ensure stronger property rights and thus attract higher levels of investment in oil exploration (Cust and Harding ). If this is the case, then the effect of oil wealth on democracy will be downward biased; implying that the estimates in Table . are conservative. Addressing this worry requires identifying a plausibly exogenous variation in oil wealth that is plausibly exogenous to prevailing political (and economic) conditions in oil-producing countries. To establish a causal relationship, I use an instrumental variables approach. Here I adopt Kris Ramsay’s () identification strategy which exploits the “out-of-region damage” (DAMAGE) that five types of natural disasters inflict on oil-producing countries. The validity of this approach “comes from the fact that an earthquake in Mexico, for examples, does not influence regime characteristics in Saudi Arabia other than through its effect on the price of oil. Similarly, a mud slide in Colombia has no effect on the political regime in Cameroon, except for its impact on world prices. As such, using the natural
https://doi.org/10.1017/9781009367509.009 Published online by Cambridge University Press
Fueling Authoritarian Resilience
disaster instrument makes it possible to estimate the causal effect of changes in oil revenues on politics” (Ramsay , ). The instrument is constructed based on the damage associated with five types of natural disasters: earthquakes, volcanoes, mudslides, waves and surges, and windstorms (e.g., hurricanes, typhoons). To satisfy the exclusion restriction, I use disaster damage that occurs “far away” from a given country. This is done by dividing the world into five regions: Europe, the Middle East and North Africa, sub-Saharan Africa, Asia, and the Americas. For instance, for oil producers in the Middle East, DAMAGE is equal to the total disaster damage in the other four regions (in that year). Armed with this instrument, I estimate the following set of two-stage least squares (SLS) regressions: First stage : Rit ¼ a þ b∗ DAMAGEit þ c∗ Xit þ uit
ð6:2aÞ
Second stage : Dit ¼ π þ θRit þ κXit þ μit
ð6:2bÞ
where in the first stage regression, Rit measures per capita oil wealth in country i in year t, DAMAGEit corresponds to the total dollar value of out-of-region disaster damage for that country, Xit is a vector of controls, and uit is an error term. DAMAGEit affects Rit through oil prices: Greater out-of-region disaster damage comprises a negative supply shock to oil production; thus, leading to a higher world oil price. Ceteris paribus, higher oil prices increase a country’s per capita oil income. Thus, in Equation .a, I expect the coefficient on DAMAGE (b) to be positive and statistically significant. Intuitively, the predicted value for Rit in Equation .a is used in the second stage regression. In Equation .b, the dependent variable ðDit Þ measures a country’s annual normalized polity score. If greater oil wealth harms democracy, I expect the coefficient on Rit ðθÞ to be
Some scholars are skeptical whether out-of-region disaster damage is exogenous due to possible spatial correlation. As Betz, Cook, and Hollenbach () show, accounting for spatial correlation attenuates the SLS estimates slightly, but does not diminish the robust negative relationship between instrumented oil wealth and democracy. The substantive implication (interpretation) remains unchanged. The choice of disaster types follows Ramsay (, ) who includes these types for two reasons: “First, the damage they inflict are unlikely to be correlated with regime type or quality unlike industrial and transportation accidents, famines, and epidemics. In fact, many of the most severe natural disasters in the five classes considered occur in the developed world (for example, Hurricane Katrina in the United States and the tsunami in Japan). Second, they also have the most plausible link to oil markets because they are physically powerful and have the ability to damage drilling and transportation infrastructure and, thus, prices.” Satisfying the exclusion restriction requires omitting the effects associated with a country’s own disaster damage and those associated with geographically “nearby disasters.” As Ramsay (, ) observes “a disaster in a country might lead to an outflow of refugees. Seeking shelter, food, and water they may enter bordering countries. This could lead neighboring states to tighten border controls, increase internal policing, or activate the military.” These actions might lead to an expansion of executive authority and reduction in democracy.
https://doi.org/10.1017/9781009367509.009 Published online by Cambridge University Press
TABLE
How Rents Affect Dictatorship and Violence . The causal effect of oil wealth on democracy
Dependent variable:
Log oil revenue per capita (a)
Out-of-region disaster
()
()
Growth Year FE No. observations
(b)
Non-conquest
Conquest
−.
.
−.
. (.)*** −. (.)***
(.)*** . (.)*** −. (.)***
(.) . (.) . (.)
(.)** . (.) −. (.)**
Yes
Yes
Yes
Yes
. (.)***
Log oil revenue per capita Log GDP per capita
Democracy
Notes: Estimation via SLS. Robust standard errors, corrected for autocorrelation. *, **, *** = significant at , , and percent respectively.
negative and statistically significant. Following convention, Equations .a and .b are estimated jointly. Table . reports the corresponding two-stage least square (SLS) estimates. Column a reports the first stage regression. It shows that a greater amount of out-of-region disaster damage increases a country’s per capita oil revenue: The average amount of DAMAGE raises oil revenue per capita by . log points (which is about equal to its means value). The instrument is a “strong” predictor of oil revenue as its corresponding F-statistic ð¼ 20:09Þ exceeds Staiger and Stock’s () threshold for a weak instrument ðF-statistic ¼ 10Þ. Turning to the control variables, richer countries (as measured with per capita GDP) tend to generate more per capita oil revenue, presumably due to more efficient and/or higher intensity oil extraction. Per capita oil revenues tend to decline slightly in years of positive economic growth. Using predicted values from the first stage regression (column a), column b reports the corresponding second stage estimate. It shows that instrumented per capita oil revenue causes a significant reduction in democracy. The estimated effect ð¼ −0:36Þ is larger than those in Table ., suggesting the instrumental variable helps correct for the downward bias in the OLS estimates (associated with reverse causality as discussed above). This effect, however, seems to mask the differential effects associated with the institution legacy of Muslim conquest. Columns and show this clearly in separate samples of non-conquest and conquest countries. In countries that were not conquered by
https://doi.org/10.1017/9781009367509.009 Published online by Cambridge University Press
Fueling Authoritarian Resilience
Muslim armies, instrumented per capita oil is positively correlated with democracy (column ). In contrast, in countries that were conquered by Muslim armies, greater instrumented per capita oil revenue has a statistically significant effect in harming democracy (column ). The results in columns and coupled with those in columns – of Table . imply that greater oil rents in countries exposed to the institutional legacy of Muslim conquest are significantly less democratic. This is consistent with proposition in Chapter that greater levels of rents in countries with less cohesive institutions entrenches autocracy. C Oil and Domestic Peace Has the greater authoritarian stability of conquest petrostates also elevated their propensity to experience civil war? As Section . describes, existing scholarship would suggest so. To investigate the veracity of this conjecture, I focus on the sample of conquest petrostates. The results in Table . answer in the negative: Oil wealth has not heightened the propensity for civil war in conquest petrostates. To evaluate this, I regress the incidence of low and high intensity civil war on per capita oil wealth (plus a vector of confounders). Following the convention in the quantitative civil war literature, the threshold for the former dependent variable is twenty-five battle deaths per annum, while TABLE
. Oil wealth does not heighten civil war in conquest petrostates
Dependent variable: Intensity:
Civil war Both
Low
High
Both
Low
High
()
()
()
()
()
()
OLS −.
OLS −.
OLS −.
SLS .
SLS −.
SLS .
(.)*** .
(.)*** .
(.) −.
(.) −.
(.) .
(.) −.
Growth
(.) −. (.)*
(.)* −. (.)
(.) −. (.)
(.) −. (.)
(.) −. (.)
(.) . (.)
Year FE No. observations R-squared
Yes .
Yes .
Yes .
Yes .
Yes .
Yes .
Estimation: Log oil rev. per capita Log GDP per capita
Notes: Estimation via OLS in columns – and SLS in columns –. Robust standard errors, adjusted for autocorrelation. *, **, *** = significant at , , and percent respectively. All specifications include year fixed effects. These coefficients and a constant are not reported.
https://doi.org/10.1017/9781009367509.009 Published online by Cambridge University Press
How Rents Affect Dictatorship and Violence
the latter requires at least , battle deaths per annum. These measures are drawn from ACD/PRIO. First, using a pooled measure, column in Table . shows that a log unit increase in per capita oil revenue is associated with a percentage point reduction in the incidence of civil war. This corresponds to a percent decline in the average baseline rate of conflict in conquest oil producers. Column shows that this is largely associated with a sharp decline in low intensity civil war comprising at least twenty-five battle deaths, while column indicates that oil wealth has no discernible effect on large scale civil war. Columns – reports the causal impact of per capita oil revenue using outof-region disaster damage as an instrumental variable. While the directions of the effect vary across these specifications, in none is the causal effect statistically significant. Rather these null effects imply that greater oil wealth has no discernible effect on low or high intensity civil war in conquest territories. Taken together, the six specifications in Table . coupled with the results in Tables . and ., demonstrate that greater oil wealth in conquest territories has strengthened autocracy without inducing internal political instability. Skeptics may contend that limiting the empirical analysis in Table . to conquest countries conflates other background country characteristics that might affect civil war in oil producers. A prominent candidate is societal fragmentation (e.g., ethnic and linguistic diversity), which several studies have linked to a greater propensity of civil war. I probe this concern in two ways. First, a statistical test of group means reveals that conquest and non-conquest petrostates do not differ on their levels of ethnic and linguistic diversity (using measures from Alesina et al. ). For example, there is about a percent chance of “meeting” someone from a different ethnic group in non-conquest petrostates. In conquest petrostates, this probability is percent. The corresponding percentage point difference is not statistically significant (p-value = .). Similarly, there is a statistically insignificant difference in the group means associated with linguistic diversity across conquest and non-conquest petrostates. Second, I evaluate the effect of per capita oil wealth on civil war in the most diverse societies. Here, I limit the sample of countries to those with levels of ethnic and linguistic diversity above the sample median. In both OLS and SLS regression specifications, the estimated effect of per capita oil wealth on civil war are negative but not significant. The negative association is surprising as existing studies suggest the opposite. More importantly, these null effects are reassuring as they suggest that cross-national variation in societal fragmentation is not a
Blattman and Miguel () review this literature. To be clear, I regress the incidence of civil war on a country’s per capita oil revenue. These results are not reported but available in the book’s replication materials.
https://doi.org/10.1017/9781009367509.009 Published online by Cambridge University Press
Fueling Authoritarian Resilience
particularly robust determinant (contextual setting) for whether oil revenues affect civil war.
D Sources of Political Stability Why has oil wealth been a source of authoritarian peace in conquest petrostates? In Chapter , the formal model demonstrated that greater levels of rents can strengthen a dictator’s repressive capacity (e.g., Intervals , , and in Table .). Table . presents evidence from conquest oil producers consistent with this conjecture. Here, I examine two aspects of an incumbent’s repressive capacity: her ability to suppress political freedoms and expand her dictatorial powers. For the former, I use the “civil liberties” and “political rights” indices from Freedom House. For the latter, I use the executive constraints index from POLITY. I normalize each of these indices on a [, ] scale where a higher value corresponds to more political freedoms or greater constraints on the chief executive’s political authority. Thus, a downward movement in the normalized scores implies an expansion in the state’s level of political repression. Table . reports the effect of per capita oil wealth on political repression in conquest petrostates. Panel A reports the association from specifications estimated via OLS. Panel B reports the corresponding causal effects from SLS
The Freedom House measures are based on an annual survey of the state of global freedom as experienced by individuals. The survey measures freedom according to two broad categories: political rights and civil liberties. Political rights enable people to participate freely in the political process, including the right to vote freely for distinct alternatives in legitimate elections, compete for public office, join political parties and organizations, and elect representatives who have a decisive impact on public policies and are accountable to the electorate. Civil liberties allow for the freedoms of expression and belief, associational and organizational rights, rule of law, and personal autonomy without interference from the state. The political rights and civil liberties measures lie on a -point scale ( to ), with representing the highest and the lowest level of freedom. I re-scale these indices so that a higher value corresponds to greater freedom. I then normalize the index (i.e., divide the re-scaled index by ) on a [,] scale. Operationally, the executive constraints index measures the extent of institutionalized constraints on the decision-making powers of chief executives, whether individuals or collectivities. According to POLITY’s codebook (Marshall and Jaggers , ) “such limitations may be imposed by any “accountability groups.” In Western democracies these are usually legislatures. Other kinds of accountability groups are the ruling party in a one-party state; councils of nobles or powerful advisers in monarchies; the military in coup-prone polities; and in many states a strong, independent judiciary. The concern is therefore with the checks and balances between the various parts of the decision-making process. A seven-category scale is used.” The seven categories (with the point values in parentheses) are: () unlimited authority, () intermediate category, () slight to moderate limitations on executive authority, () intermediate category, () substantial limitations of executive authority, () intermediate category, () executive parity or subordination. Thus, a higher value corresponds to greater constraints on executive authority. I normalize this index (i.e., divide the index by ) on a [,] scale. A value closer to implies greater constraints on executive authority (i.e., more democratic).
https://doi.org/10.1017/9781009367509.009 Published online by Cambridge University Press
TABLE
How Rents Affect Dictatorship and Violence . Repression in conquest oil producers
Dep. variable:
Civil liberties
Political rights
Exec. constraints
Coups
()
()
()
()
Log oil revenue per capita
−. (.)***
A. Estimation via OLS −. −. (.)*** (.)***
. (.)
Log oil revenue per capita
−. (.)**
B. Estimation via SLS −. −. (.)** (.)**
. (.)
Controls Year fixed effects No. observations
Yes Yes
Yes Yes
For both panels Yes Yes
Yes Yes
Notes: Estimation via OLS (in Panel A) and SLS (in Panel B). Robust standard errors, corrected for autocorrelation in parentheses. *, **, *** = significant at , , and percent respectively. All specifications control for log GDP per capita, economic growth, and year fixed effects. These coefficients and a constant are not reported.
regressions, where the instrumental variable is a country’s out-of-region disaster damage. All the specifications control for per capita GDP, economic growth, and a vector of year fixed effects. These coefficients and a constant are not reported. In Panel A, column indicates that greater per capita oil revenue is associated with a reduction in the level (quality) of civil liberties in conquest petrostates. The effect is highly statistically significant. The corresponding causal effect is larger in Panel B, column ðcoefficient ¼ −0:217Þ as it corrects for the downward bias in the OLS estimate. The estimated OLS and SLS effects of per capita oil revenue on political rights (column ) are almost identical in magnitude and statistical significance to those in column . The reduction in the political rights enjoyed by citizens is accompanied with an expansion in the incumbent’s political authority. Column shows that greater per capita oil income reduces the constraints faced by the country’s chief executive. In Panel B, the estimate implies that a one standard deviation increase in per capita oil revenue reduces the normalized executive constraints index by . points (which is equivalent to a standard deviations movement in the normalized executive constraints index). For the modal conquest petrostate, this corresponds to empowering a dictator with “unlimited” political authority. While greater oil wealth tends to expand the state’s power relative to its citizens in conquest oil producers, skeptics may worry this will empower the state’s main repressive agent – the military – to mount a coup (Nordvik ). Using data on the incidence of a coup from Lei and Michaels (), column shows that this worry is overstated. While greater per capita oil wealth slightly raises the incidence of a coup, the effect is statistically
https://doi.org/10.1017/9781009367509.009 Published online by Cambridge University Press
Fueling Authoritarian Resilience
indistinguishable from zero. Reassuringly, this null effect coupled with effects in columns – imply that greater oil wealth increases a government’s repressive capacity without heightening its likelihood of experiencing a coup. More generally, the inferences from Table . are consistent with existing accounts that oil can “buy” political stability, especially when rentier governments experience periods of heightened internal political unrest. For example, faced with widespread protests following the country’s presidential elections in , the Iranian regime employed its Revolutionary Guard and various statesponsored militias to repress political activists. In , Muammar Gaddafi leveraged his government’s immense oil wealth to fund thousands of African mercenaries to combat domestic rebel forces (though Gadaffi ultimately lost, in part because of military intervention by NATO). More generally, these specific episodes fit a broader finding that petrostates – particularly those governed by dictators – spend more on military arms and personnel than non-oil producers (Ross , Colgan ).
. The greater authoritarian resilience of conquest petrostates has also been enhanced by an additional type of rent: geopolitical support from the United States. In this section, I argue and present statistical evidence that greater certainty of external support from the US government – measured using annual US troop deployments in foreign territories – has enhanced the military strength of governments in conquest petrostates and lowered their propensity for experiencing civil war. The willingness of the United States to station its troops in these territories comprises an important part of the US government’s broader foreign policy objectives since World War II to secure a stable supply of oil for itself and its allies (e.g., Japan, Western Europe). A External Support in Civil Wars Support from a foreign power can help mitigate violence between an incumbent government and opposition groups. This follows, for instance, from rationalist explanations for civil war which attribute their occurrence to a bargaining failure stemming primarily from incomplete information and/or commitment problems (Fearon ). In such a framework, the availability of support (or lack thereof ) from a foreign actor can often shape the bargaining dynamics between a government and opposition group(s). External support can fall into
In recent years, US support for these regimes has also served to attract their financial capital to the United States. For example, large holdings of US treasury bills by several Middle Eastern oil producers (e.g., Saudi Arabia, Kuwait, UAE) has helped finance the US government’s budget deficit (Bapat ). A third, and less likely cause, is “issue indivisibility.”
https://doi.org/10.1017/9781009367509.009 Published online by Cambridge University Press
How Rents Affect Dictatorship and Violence
two categories: () future, direct intervention (e.g., “boots on the ground”) and () current and future efforts to prop up the current government or rebel group (e.g., through financial assistance, arms shipments). Building on this rationalist framework, David Cunningham () posits that potential support from a foreign power can affect conflict dynamics. With greater certainty of external support (e.g., through the deployment of a foreign power’s troops in the crisis territory) warring factions are more likely to reach a negotiated settlement that limits violence. This may occur by helping resolve ongoing conflicts or deterring new conflict onsets by clarifying the relative power asymmetries between the incumbent and opposition groups. Cunningham’s theoretical conjectures and empirical analysis emphasize how greater certainty of external support for the government can strengthen it relative to opposition groups (e.g., by increasing the government’s repressive capacity), thus inducing opposition demands to fall within the peaceful bargaining range. This can happen by effectively sending a clear signal about the relative capabilities of warring parties, thus reducing the likelihood of armed combat. Several cases of French support in West Africa – particularly on the side of incumbent governments – provides illustrative examples of how external support for the incumbent government can affect conflict dynamics. For instance, this was the case in Gabon in , in which the military ousted President Leon M’ba in a coup that generated little violence. However, in response, President Charles de Gaulle honored a agreement with M’ba, deploying French paratroopers and restoring M’ba to power. While this intervention had the potential to incite conflict between the M’ba government and the rebellious arm of the military (and its domestic supporters), French support of the incumbent effectively diminished the opposition’s probability of victory to zero; thus, ensuring peace. This is not an isolated example, as France has actively extended support over a dozen times in Africa – primarily in its former colonies – to maintain political stability. This behavior reinforces James Fearon and David Laitin’s () observation that civil wars are less likely in former French colonies than in other African countries, as well as Paul Collier’s () analysis that France’s “over-the-horizon guarantees” to protect the governments of its former colonies has reduced the risk of civil war in Francophone Africa by nearly three-quarters.
In his analysis, Cunningham refers to support as “intervention.” Other foreign powers regularly intervene in civil wars in support of the government. A recent example is the Russian government’s support for the Assad’s regime in the ongoing Syrian civil war (since ). My own analysis of US interventions (using data from the UCDP External Support Dataset) finds that between and , almost percent of US interventions in civil conflicts supported the government. In recent years, as part of its effort to combat Islamic terrorism, the French government has deployed its special forces to repel anti-government Islamic groups in West Africa.
https://doi.org/10.1017/9781009367509.009 Published online by Cambridge University Press
Fueling Authoritarian Resilience
The examples above illustrate how greater certainty of external support can have a pacifying effect on violence between a government and opposition group (s), particularly by enhancing the relative strength of the incumbent vis-à-vis the opposition. B US Foreign Oil Policy The arguments in the previous sub-section can help situate US foreign policy toward many conquest petrostates since World War II. US foreign oil policy has largely been a continuation of British foreign policy of preserving access to a stable supply of oil from the Middle East. In his monumental study of the “prize” to secure stable access to oil, Daniel Yergin () provides a detailed account of the foreign policies of major powers, particularly of Great Britain and the United States during the twentieth century toward petrostates. Yergin identifies two important drivers of British and American foreign oil policy in the Middle East: securing stable oil supplies and Cold War politics. For the first driver, Yergin (, ) writes: The overlapping of politics and economics had, in the second half of the s, created a new strategic focus for the British and American governments. In the case of the British, even as they withdrew, from the far reaches of the empire, they could not turn their backs on the Middle East. The Soviets were putting pressure on the “northern tier” – Greece, Turkey, and in particular, Iran. And Iran, plus Kuwait and Iraq, were Britain’s major sources of oil. Continued access was required for military security, and the dividends from the Anglo-Iranian were a major revenue generator for the Exchequer . . . If the British focus narrowed, the shift in the outlook and commitments greatly expanded for the United States. No longer would an American President say, as Franklin Roosevelt had said in , that Saudi Arabia was a little far afield. The United States was becoming an ever more petroleum-based society, which could no longer supply its own needs with domestic production. The world war, just over, had proved how central and critical oil was to national power.
The Western alliance’s security competition with the Soviet Union also elevated the importance of oil-producing countries, particularly in the Middle East. As Yergin (, ) observes:
My discussion of the US foreign oil policy below is quite brief as it focuses on the main reasons why the United States provides an implicit security guarantee to many Middle Eastern oil producers. My account omits interesting discussions on the role of American oil companies, the nationalization of oil production in many Arab countries in the s and s, rivalries among Middle Eastern oil producers (e.g., Iraq versus Iran, Saudi Arabia versus Iran), and tensions over the Israel-Palestine issue. An important consideration in the US government’s foreign oil policy, particularly in the Middle East, has been safeguarding Israel’s national security. Pursuing the latter has affected US relations with many Middle East, Muslim majority states. For example, US support for Israel played a critical role in the decision of OPEC to severely restrict oil exports to the United States during the Arab oil embargo.
https://doi.org/10.1017/9781009367509.009 Published online by Cambridge University Press
How Rents Affect Dictatorship and Violence
Soviet expansionism – as it was, and as it might be – brought the Middle East to center stage. To the United States, the oil resources of the region constituted an interest no less vital, in its own way, than the independence of Western Europe; and the Middle Eastern oil fields had to be preserved and protected on the Western side of the Iron Curtain to assure the economic survival of the entire Western world . . . In the Cold War, this oil would be great value, and everything possible should be done to prevent its loss.
In pursuit of this objective, Saudi Arabia became a dominant focus of American policymakers. This was made very clear in the early years of the Cold War. For example, in a letter to King Ibn Saud in October , President Harry Truman wrote: “I wish to renew to Your Majesty the assurances which have been made to you several times in the past, that the United States is interested in the preservation of the independence and territorial integrity of Saudi Arabia. No threat to your Kingdom could occur which would not be a matter of immediate concern to the United States.” This letter effectively signaled a US security guarantee to Saudi Arabia (and other pro-Western oil producers) from foreign and domestic political threats. According to Daniel Yergin (and other scholars), this security guarantee formed the basis of a “special relationship” (particularly between the US and Saudi Arabia) that has interweaved public and private interests for over seventy years. This multi-faced relationship has evolved over time, balancing intricate relationships among private oil companies (e.g., Exxon-Mobil, Chevron, BP, Shell) and national governments in petrostates, especially after many governments nationalized their oil industries in the s and s. Moreover, the implicit American security guarantee has been augmented by large arms sales from the United States (and its Western allies) to petrostates in the Middle East (Yetiv and Oskarsson ) and the deployment of US troops to many of these countries (as I document shortly). Why did many Middle Eastern oil producers choose to side with the United States during the Cold War? In addition to the expertise and market power of American oil companies, aligning with the United States represented a more attractive political option, at least relative to the Soviet Union. For example, on several occasions, the Saudi monarchy turned down invitations from the Communist regimes in China and the Soviet Union to improve their bilateral relations. The monarchy proclaimed to do so on grounds of its adherence to and leadership of the Islamic faith. A US diplomatic cable from reports King Saud’s reasoning (based on official written correspondence with US officials):
Not all countries in the Middle East sided with the United States during the Cold War. Most notably, up until the late s, Egypt enjoyed strong ties with the Soviet Union. “Telegram From the Embassy in Saudi Arabia to the Department of State, July ,” Foreign Relations of the United States, –, Near East: Jordan-Yemen, Volume XIII, US Department of State. URL: https://history.state.gov/historicaldocuments/frus–v/d. Accessed: September , .
https://doi.org/10.1017/9781009367509.009 Published online by Cambridge University Press
Fueling Authoritarian Resilience
Our very special attitude towards communism is well-known to US Government and to world. It is to our interest that communism not infiltrate into any area of Middle East. In opposing communism, we do so on basic religious belief and Islamic principle, in which we believe with all our heart, and not to please America or Western states. . . . My position, in particular, of Moslem Arab King, servant to Holy Shrines, looked up to by million Moslems in East and West, is extremely delicate and serious before God, my nation and history.
Furthermore, the Soviet Union’s willingness to finance Communist movements to oust incumbent regimes in the Middle East, compelled many other Middle Eastern governments to turn away from Moscow. For example, a US diplomatic cable in identifies why the Iraqi government – despite accepting foreign aid from the Soviet Union – did not side with the Soviet Union at that time: The weaknesses of the local Communist position are numerous and deep and include: ) Iraqi hatred of local Communists stemming from the latter’s excesses during the Qasim period; ) Iraqi Government resentment of clandestine Soviet support for Kurds and local Communists; ) The failure of some Soviet aid projects; ) The cultural affinity of educated Iraqis for the West; ) Iraqi preference for Western goods, travel, contacts and higher education, and the poor calibre of Soviet products and personnel; ) Iraqi dread of USSR “great power” dominance; and ) Mutual antipathies (Arabism, Islam vs. atheistic communism, etc.
Since the end of the Cold War, the United States has, for the most part, continued to extend an implicit security guarantee to many Middle Eastern petrostates, with Iran being a notable exception. This continuity in policy became strikingly apparent at the outset of the post–Cold War period. The rapid deployment of American forces to thwart Saddam Hussein’s invasion of Kuwait in signaled the United States’ commitment to preserving the territorial integrity of contemporary Middle Eastern states and their forms of governance. Even after the / attacks – in which Osama Bin Laden attributed Al Qaeda’s motivation to the robust presence of American troops in Saudi Arabia – the United States has continued to station troops in Arab/Muslim countries. The security guarantee has also been “costly” for many of its beneficiaries. Many Arab petrostates continue to accept it despite the United
“Memorandum From the Country Director for Israel and Arab-Israel Affairs (Atherton) to the Assistant Secretary of State for Near Eastern and South Asian Affairs, November ,” Foreign Relations of the United States, –, Volume XXI, Near East Region; Arabian Peninsula, US Department of State. URL: https://history.state.gov/historicaldocuments/ frus–v/d. Accessed: September , . Following the Iranian Revolution in , US-Iranian relations have been hostile. Several non-oil-producing Middle Eastern countries – particularly Egypt and Jordan – enjoy strong bilateral relationships with the United States. These countries receive large sums of US economic and military aid and the US government has not actively pressed their autocratic regimes (military rule in Egypt and a monarchy in Jordan) to democratize.
https://doi.org/10.1017/9781009367509.009 Published online by Cambridge University Press
How Rents Affect Dictatorship and Violence
States’ invasion and subsequent occupation of Iraq in , which sparked tremendous anti-American sentiment in the “Arab street.” In short, despite some costs associated with the implicit US security guarantee, it continues to operate for many petrostates in the Middle East. C US Troops Engender Peace The preceding discussion suggests that US foreign policies toward many petrostates – particularly those in conquest territories in the Middle East – can safeguard their regimes from external and importantly internal (domestic) threats. In this sub-section, I statistically evaluate this conjecture using data on US troop deployments as a credible measure of US resolve and capacity to provide external support for the incumbent. For the United States, the allocation of its troops in foreign territories is an important component of its ability to project power abroad and rapidly respond to overseas threats. For the countries hosting these troops, their presence can serve to deter external aggression from hostile neighbors (e.g., US troops stationed in South Korea) and/or signal to domestic opposition groups that the US can credibly aid the incumbent to combat insurgencies. The latter is consistent with David Cunningham’s () argument that greater certainty (and credibility) of external support can strengthen an incumbent’s military actual (or perceived) military capacity vis-à-vis its domestic opposition, thus, incentivizing the opposition to reach a negotiated settlement that reduces the likelihood of civil war. To statistically test whether this in the case, I regress a country’s incidence of civil war on the number of US troops it hosts (plus a vector of controls and year fixed effects). Table . reports the results. Columns and show that US troop deployments do not affect civil war in non-conquest oil producers. The estimated effect is statistically indistinguishable from zero. However, in conquest oil producers – many of whom are in the Middle East and are the beneficiaries of the implicit US security guarantee – the presence of US troops has a pacifying effect. The estimated effect in column implies that a log unit increase in US troops (equal to about troops) reduces the likelihood of civil war by percentage point. This effect increases in magnitude and statistical significance when accounting for a country’s per capita oil revenue (column ). The finding in column is noteworthy in two important ways. First, it shows the presence of a greater number of US troops in conquest oil producers significantly reduces the country’s likelihood of experiencing civil war, presumably by strengthening the government’s (perceived) ability to thwart challenges from domestic opposition groups. Second, a stronger presence of US troops seems to serve as an important “mediating” variable in the relationship between oil
Data on annual US troop deployments abroad are from Kane (). It is available at: www .heritage.org/defense/report/global-us-troop-deployment--. Accessed: September , .
https://doi.org/10.1017/9781009367509.009 Published online by Cambridge University Press
Fueling Authoritarian Resilience TABLE
. The pacifying effect of US troop deployment in conquest petrostates
Sample:
Non-conquest
Dep. variable:
Conquest
Civil war ()
()
Civil war ()
. (.)
Log oil revenue per capita
() −. (.)
Log US troops
. (.)
. (.)
−. (.)*
−. (.)**
Controls Year fixed effects No. observations R-squared
Yes Yes .
Yes Yes .
Yes Yes .
Yes Yes .
Log US troops ()
()
−. (.)***
−. (.) ***
Yes Yes .
Yes Yes .
Notes: Estimation via OLS. Robust standard errors corrected for autocorrelation. *, **, *** = significant at , , and percent respectively. All specifications control for log GDP per capita, annual economic growth, and year fixed effects. These coefficients and a constant are not reported.
wealth and civil war in conquest oil producers. This inference follows through the combination of two relationships. First, column in Table . shows that greater per capita oil wealth is associated with a statistically significant reduction in civil war. Second, however, as column in Table . shows, this statistically significant association is reduced when controlling for US troop deployments. The remaining columns in Table . provide additional statistical evidence verifying that US troop deployment is a mediating variable in conquest oil producers. To do so, I use a four-step regression procedure developed by Rueben Baron and David Kenny () to statistically decompose a potential mediation process. To guide my analysis, consider the diagram: W!T!C where W, T, C refer to per capita oil wealth, US troop deployments, and civil war respectively. US troop deployments ðT Þ is the hypothesized mediating variable. The first step examines the overall relationship (W ! C). It regresses C on W, which corresponds to the main specifications in Table .. The second and third steps probe the intermediate steps. The second step regresses T on W to test the significance of the first intermediate relationship, W ! T. The third step regresses C on T to test the significance of the second intermediate relationship, T ! C (which corresponds to column ). The fourth step conducts a multiple regression with W and T predicting C (which corresponds to column ). The purpose of the steps – is to verify that there exists a zero-order relationship among the variables. If one or more of these relationships are not
https://doi.org/10.1017/9781009367509.009 Published online by Cambridge University Press
How Rents Affect Dictatorship and Violence
statistically significant, one cannot usually conclude whether mediation is not possible or likely. If there are significant relationships from steps through , one proceeds to step . In this fourth step regression model, some form of mediation is present if the effect of the mediator ðT Þ remains significant after controlling for W. If W is not significant when T is controlled, the findings support full mediation. In Table ., columns and report the third and fourth steps. Columns and show the initial two steps. Column shows that oil wealth exhibits a negative and statistically significant effect on civil war in conquest oil producers. Examining the first intermediate relationship ðW ! T Þ, column reports a statistically significant association between per capita oil wealth and US troop deployments. Interestingly, the estimated effect is negative, implying that US troops do not tend to be stationed in countries with greater oil wealth and thus suggest the estimated effects associated with US troops in columns and are not upward biased. Column shows the second intermediate relationship ðT ! W Þ: US troop deployments are a statistically significant predictor for a lower incidence of civil war. Following the reasoning in Baron and Kenny (), since the estimated relationships are statistically significant across columns , , and one can safely infer that mediation is likely. Finally, column provides evidence of full mediation.
. - The empirical analysis thus far supports proposition from Chapter ’s formal model that rents are a source of authoritarian stability in countries whose preexisting institutions are less cohesive, that is, countries exposed to the institutional legacy of Islamic conquest. In these conquest territories, oil wealth has fostered dictatorship without inciting civil war. Yet the formal model also provides a potential pathway for democracy: Dictatorships can transition to democracy provided preexisting institutions are more cohesive. This suggests that nondemocratic petrostates that were not exposed to the institutional legacy of Muslim conquest may be able to democratize. Several historical episodes provide illustrative examples. Consider, political development in oil abundant Venezuela. In , political elites ousted dictator Perez and peacefully transitioned toward a democracy through an explicit “pact” between the country’s three leading political parties. The Punto Fijo Pact sought to ensure an equitable distribution of political and economic power/resources (Dunning ). First, the pact
If W remains statistically significant (i.e., both W and T both significantly predict C), the finding supports partial mediation. Venezuela spearheaded the creation of OPEC and, as of , has the world’s largest proven reserves of oil.
https://doi.org/10.1017/9781009367509.009 Published online by Cambridge University Press
Fueling Authoritarian Resilience
required that each of the parties respect the outcome of the elections but include members of the other parties in positions of power. Second, the government agreed to use its oil revenues to fund expenditures equitably to benefit constituencies favored by each of the major parties. This second feature corresponds to a society with more cohesive institutions, where the incumbent tends to allocate government expenditures equitably. With this pact in place, Venezuela transitioned relatively peacefully from a military dictatorship to a presidential democracy. A similar peaceful transition from dictatorship to democracy unfolded in a Muslim majority petrostate: Nigeria. However, unlike its Muslim counterparts in the Middle East and North Africa, mamluk institutions were not prevalent in the premodern period. While Islam spread to Nigeria by , it did so primarily via trade and the teachings of religious scholars. Thus, Nigeria is a non-conquest Muslim majority petrostate. Like Venezuela, Nigeria’s democratic transition was relatively bloodless and advanced through a coalition of political parties. In , the National Movement joined northern/Muslim and southern/Christian interest groups to oust the military regime. The first elected president shared power (and oil revenues) with a broad coalition based on the National Movement. These two historical episodes are indicative of a more general pattern toward greater democracy in non-conquest petrostates over time. Figure . traces the trajectory of democracy across petrostates that were nondemocratic in the s. Using this sample, the figure plots the average annual normalized polity score across conquest and non-conquest petrostates. The figure reveals two important points. First, prior to the first oil price shock (circa ) the “typical” conquest and non-conquest petrostate exhibited a similar level of dictatorship. Prior to , the average normalized polity hovered around .; well below the cutoff for democracyð¼ 0:5Þ. Second, and more importantly, a sharp divergence in democracy emerges after . While the average level of democracy increased slightly in conquest countries, they remained strongly nondemocratic (with an average normalized polity score well below .). In contrast, the average level of democracy has gradually increased in nonconquest petrostates. Since the early s, the average non-conquest petrostate had successfully transitioned to democracy with a normalized polity exceeding .. Combining these two points clarifies a key takeaway from this chapter: The institutional legacy of Muslim conquest has compounded the antidemocratic effects of oil wealth. With this legacy of less cohesive institutions, oil revenues have fueled dictatorship in conquest petrostates without
The ousted military leader (General Olusegun Obasanjo) was not punished and was later elected after the country’s second democratic transition in . To be clear, the group means exclude countries that were democratic (i.e., during the s, had an average normalized polity greater than .). For example, the United States is excluded from the non-conquest group.
https://doi.org/10.1017/9781009367509.009 Published online by Cambridge University Press
0
Normalized POLITY 0.2 0.4
0.6
How Rents Affect Dictatorship and Violence
1965
1970
1975
1980
1985 Year
1990
Non-conquest
1995
2000
2005
Conquest
FIGURE . Democracy in “initially nondemocratic” petrostates over time, by conquest status Notes: Group means calculated from a sample of petrostates with an average normalized polity score less than . during the s. The data is the same as those used in prior sections.
generating internal warfare. In contrast, there has been a pathway for democracy in petrostates that were not exposed to the institutional legacy of Muslim conquest.
. Existing studies find that oil rents can generate antidemocratic effects but may also be a source of civil war in dictatorships. Recognizing this tension, this chapter demonstrates that economic and geopolitical rents associated with oil have strengthened the nondemocratic institutional apparatus in petrostates that were exposed to Muslim conquest without engendering civil war. In short, oilrelated rents have fueled authoritarian resilience in conquest petrostates. Yet, in non-conquest petrostates oil wealth has not necessarily closed off the path toward democracy. In these societies where preexisting institutions are more cohesive (relative to those exposed Muslim conquest) there has been a gradual march toward more democratic politics over time. Together, these findings imply that the confluence of oil-related rents and the institutional legacy of Muslim conquest has been an important source of
https://doi.org/10.1017/9781009367509.009 Published online by Cambridge University Press
Fueling Authoritarian Resilience
authoritarian stability in many Muslim petrostates, especially in relation to petrostates that were not exposed to Muslim conquest. Is this inference generalizable to other forms of rents? In the next two chapters, I study the impact of foreign transfers (that accrue to government revenue) on democracy and civil war. Here, the greater volatility in the amount of transferred rents has generated considerably more variability in democracy and civil war. As with the evidence in this chapter, the institutional legacy of Muslim conquest has shaped whether foreign transfers have fostered democracy or dictatorship and civil war or peace.
https://doi.org/10.1017/9781009367509.009 Published online by Cambridge University Press
Aiding Political Violence
Whereas rents in Muslim petrostates have been a source of relatively peaceful authoritarianism, this has not been the case in many Muslim countries that do not produce oil. Muslim non-oil producers – many of whom were also exposed to the institutional legacy of Muslim conquest – have tended to experience more frequent and violent political upheaval since the s. This chapter presents evidence that buoyant levels of foreign aid receipts strengthened dictatorship in Muslim non-oil producers, but their subsequent decline led to heightened political instability in the form of civil war. The former is a direct test of corollary of A in Chapter that a decline in rents heightens the prospects of civil war in societies where preexisting institutions are less cohesive. However, empirically evaluating whether foreign aid has a causal impact on patterns of political violence faces several challenges. First is the concern whether foreign aid is “like” unearned income, such as oil. For instance, Sarah Bermeo () aptly discusses why foreign aid, even though it has some features of unearned income (e.g., nontax income where some fraction goes to the central government), differs from oil. She identified donor intent and its composition as two important differentiators. Donors do not allocate aid randomly. Rather, economic and political conditions within recipient countries can shape donor allocation decisions. Donors, for example, may disburse more aid to encourage political liberalization, buy political stability, and/or to limit
In this chapter’s Appendix, Table D lists the sample of non-oil producers. There is a large literature on the political economy of foreign aid that evaluates both why donors give aid and its economic consequences. Some examples of the former include, Alesina and Dollar (), Bermeo (), Bueno de Mesquita and Smith (), Werker (), and Winters and Wright (). Some examples of the latter include Ahmed (, ), Dollar and Burnside (), Nunn and Qian (), and Wright (). This cited scholarship is but a fraction of the existing scholarship on foreign aid and development.
https://doi.org/10.1017/9781009367509.010 Published online by Cambridge University Press
Aiding Political Violence
the transmission of crises from recipient countries (e.g., terrorism, refugees). These allocation decisions can also influence the composition of aid. Aid may be ‘tied’ to donor preferences to limit the recipient government’s ability to channel the aid to areas not preferred by donors (i.e., limit aid fungibility) and/or to reward domestic producers in the donor country. A second and related concern is the endogeneity of aid with observable (and unobservable) conditions within the recipient country. Reverse causality is a prominent source of endogeneity bias: Economic and political conditions within a recipient can affect how much aid donors give. For instance, a donor may want to reward recipients undertaking democratic reforms. If this is the case, estimates of the impact of aid on democracy will be upward biased. Alternatively, if a donor wants to strengthen the longevity of certain dictatorships (e.g., US aid to Mubarak’s Egypt or the Hashemite monarchy in Jordan), this will downward bias the effect of foreign aid on democracy. Non-random measurement error can be an additional source of endogeneity bias. This may potentially arise from the correlation of a country’s domestic politics and its systematic collection and misreporting of its economic data. Mitigating these concerns requires identifying a plausibly exogenous source of untied foreign aid that is uncorrelated with a recipient’s domestic political economy, particularly in Muslim aid recipients. To grapple with these concerns, I draw on collaborative work with Eric Werker that leverages a quasi-natural experiment of largely untied aid disbursements from oil producers in the Persian Gulf to poor, non-oil-producing Muslim recipients (Ahmed and Werker ). As I discuss in the next section, two features of this quasi-natural experiment allow me to draw causal inferences. First, Gulf aid was primarily in the form of block grants to central government ministries and disbursed to poor, non-oil-producing Muslim countries. Second, the amount of aid disbursed in any given year was positively and highly correlated with oil prices. These two features allow me to construct an instrumental variable to explain variation in largely untied aid disbursements to non-oilproducing Muslim recipients in Africa, the Middle East, and South Asia. Using this quasi-natural experimental setting, I present five main results in this chapter. First, I motivate the analysis with a case vignette of foreign aid
An additional source of endogeneity bias is omitted variables. The empirical strategy in this chapter strives to mitigate this worry by controlling for country and year fixed in the baseline specifications. For example, James Hollyer, Peter Rosendorff, and James Vreeland () document how autocracies are more likely to misreport economic data to international organizations, such as the World Bank. The existence of non-random measurement error will bias regression estimates. I refer to this as a quasi-natural experiment as the allocation of aid was not random per se. However, conditional on the relevant covariates (e.g., country fixed effects which account for the non-random allocation to Muslim recipients), aid disbursements can be viewed “as if random.” In prior collaborative work with Eric Werker and Charles Cohen (), we exploit this natural experiment to study how foreign aid can impact a country’s macroeconomy by tracing its effect
https://doi.org/10.1017/9781009367509.010 Published online by Cambridge University Press
How Rents Affect Dictatorship and Violence
and patterns of political violence in Somalia and then illustrate broader patterns in the raw data across all Muslim aid recipients. With this initial motivation, I then show that oil prices are a robust determinant of aid disbursements in Muslim recipients but not in non-Muslim recipients. This forms the basis of a strong instrumental variable explaining predominantly untied foreign aid flows to Muslim recipients. I then provide robust causal evidence that instrumented aid lowers the incidence of civil war in Muslim recipients, with more pronounced effects in Muslim aid recipients that were exposed to mamluk institutions. Critically, this third result is equivalent to testing corollary A in Chapter : A decline in rents in societies heightens the prospect of civil war in societies whose preexisting institutions are less cohesive. Fourth, I discuss potential factors that might violate the exclusion restriction: The role of oilprice induced migrant remittances and the direct “meddling” of oil producers (e.g., Iran, Libya, Saudi Arabia) in the internal politics of other Muslim countries, through for example the “exporting” of religious extremism. I show that accounting for these factors does not affect the chapter’s core results. Finally, I show that despite greater political upheaval (during periods of lower aid inflows), Muslim recipients remained staunchly nondemocratic.
. To mitigate the challenges discussed in the previous section, I use the ArabIsraeli conflict in , popularly known as the Yom Kippur War, and the consequent shock to world oil prices as a quasi-natural experiment for largely untied aid disbursements to predominantly poor, non-oil-producing Muslim recipients in Africa, the Middle East, and Asia. In this section, I discuss the quasi-experimental setting and how I derive a set of regression equations to gauge causal estimates of foreign aid on political violence in Muslim non-oil producers.
A Quasi-Natural Experiment Beginning in , around the same time as oil prices began to surge, the Persian Gulf nations of the Organization of the Petroleum Exporting Countries (OPEC) ramped up their distribution of foreign aid. Eric Neumayer (, ) estimates that Arab petrostates donated about . percent of their GDP between and , which amounted to . percent of all aid disbursed over this period. In explaining which countries were the main beneficiaries, Neumayer () provides compelling econometric evidence that Gulf donors favored oil-importing Muslim countries in their aid disbursements. No across all parts of its national accounts (e.g., government expenditures, household consumption, investment, savings).
https://doi.org/10.1017/9781009367509.010 Published online by Cambridge University Press
Aiding Political Violence
doubt some motivation was political: Arab petrostates sought to subdue unrest due to the huge inequality among their co-religionists (between the oil haves and have-nots), as well as to “assure them[selves] a clear position of dominance within the Muslim world” (Kepel , –). A striking feature of Gulf aid was its composition: It was largely untied and resembled unearned income flowing directly to the state’s coffers. Unlike foreign aid from the World Bank and many Western donors, for example, Gulf petrostates gave their money with comparatively few strings attached through the form of block grants to finance ministries. According to Shireen Hunter (): “The largest part of OPEC aid has still consisted of general balance of payment and budgetary support.” Stuart Hallwood and Paul Sinclair () agree: “Most OPEC aid is given on very favorable terms and conditions from the recipient’s point of view. A large proportion of this aid is given on a grant basis; otherwise, loan terms are with low interest rates and long grace and amortization periods” (–). And while aid from Western donors has often been tied toward contracts with the donor country, “Arab aid has practically never been tied, with the exception of relatively unimportant specific loans and grants for oil purchases” (Neumayer , ). Two facts about the pattern of Gulf states’ largely untied foreign aid disbursement underlies the identification strategy. First, this aid favors other Arab and non-Arab Muslim recipients (Neumayer ). Second, the aid is highly correlated with the world price of oil. In particular, the aid programs only began in earnest after the Yom Kippur War (and the subsequent oil crisis in ), and annual aid outlays closely track oil prices (Hallwood and Sinclair , Hunter ). These two facts form the basis of the quasi-natural experiment, in which non-oil-producing Muslim majority countries were the unique recipients of an aid windfall that only lasted as long as the price of oil remained high. Many of these Muslim recipients had been previously governed with mamluk institutions (e.g., Afghanistan, Lebanon, Jordan, Morocco, Pakistan) while some – particularly those in Africa (e.g., Guinea, Mali, Niger) – had not been. (Table D lists the sample of aid recipients.) The two facts discussed above are visible in the raw data. Figure . plots average official development aid (ODA) receipts across non-oil-producing Muslim and non-Muslim countries since . From to around ,
In an alternate, but not necessarily mutually exclusive account, Ahmed and Werker () argue this generosity stemmed in part to decrease the value of the donor’s “state prize” and thwart domestic rent-seeking. Consistent with prior chapters, a country is categorized as Muslim if at least percent of its population identifies with the Islamic faith. The measure of foreign aid includes disbursements from Development Assistance Committee (DAC) and non-DAC donors; the latter group includes various Arab aid agencies (World Bank ). The measurement of the foreign aid here and throughout this chapter uses total ODA receipts, and, as such, does not require separating out Arab aid from non-Arab aid. The figures and corroborative regression estimates are therefore net of any changes in the disbursements of traditional donors or non-Arab oil-producing donors.
https://doi.org/10.1017/9781009367509.010 Published online by Cambridge University Press
0
5
Foreign aid (% GDP) 10 15
20
How Rents Affect Dictatorship and Violence
1960
1965
1970
1975
1980
Muslim FIGURE
1985 Year
1990
1995
2000
2005
Non-Muslim
. Average foreign aid receipts in Muslim and non-Muslim non-oil producers
average aid inflows in Muslim and non-Muslim countries were remarkably similar. This changed dramatically at the onset of the first oil shock. During the oil price boom from until the mid-s, Muslim countries received substantially higher amounts of aid (as a percent of GDP) compared to nonMuslim recipients. Over this period, on average, Muslim countries received . percentage points of GDP more aid than their non-Muslim counterparts. Aid receipts in Muslim countries declined thereafter and, for the most part, since the s remained largely similar to inflows received in non-Muslim countries. As is apparent in Figure ., Muslim countries experienced a windfall in foreign aid over the period that roughly corresponds to the years of high oil prices. To provide a clearer picture of this correlation, Figure . depicts movements in oil prices (right axis) and the difference in foreign aid (as a percentage of GDP) between non-oil-producing Muslim and non-Muslim countries (left axis). The figure demonstrates a strikingly strong correlation between the Unfortunately, bilateral aid data from specific Arab aid donors (e.g., Saudi Arabia, Kuwait) is unavailable, particularly from the s and s. I restrict my analysis to a sample of developing countries (i.e., not “high income” according to the World Bank’s classification) that do not produce any oil. Rich countries, of course, do not receive development assistance. Rather, they tend to be donors. I exclude oil-producing countries (as defined by British Petroleum), since the impact of high oil prices will have a direct impact on a country’s political economy that dwarfs any increase in foreign aid.
https://doi.org/10.1017/9781009367509.010 Published online by Cambridge University Press
0
–2
20
0
40 60 Oil price
Foreign aid (% GDP) 2 4 6
80
8
100
Aiding Political Violence
1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 Year Aid differential FIGURE
Oil price
. Oil price and the aid windfall to Muslim countries
amount of bonus aid disbursed to Muslim countries from Gulf donors and the world price of oil. More stringently, difference-in-differences estimates from the high oil price period (–) and the surrounding low-price periods (–, –) between Muslim and non-Muslim countries reveal that these differentials are statistically significant, and at . percent of GDP, nearly equal in magnitude to the first difference. The robust statistical relationship between oil price and aid tends to break down after due to a confluence of three potential causes in the aid allocation decisions of Gulf oil producers (Ahmed and Werker ): larger domestic populations requiring greater spending at home (rather than sending the government revenues abroad); the rise of sovereign wealth which encouraged saving; and higher allocations of aid to non-Muslim recipients from the traditional donors after . Thus, I limit my empirical analysis (from this quasi-natural experiment) to the pre- period. (In Section ., I examine the post- period.) B Exogeneity Figures . and . underlie my empirical strategy to identify the causal effect of foreign aid on political violence in Muslim recipients. Such an identification
Results reported in Appendix table A in Ahmed and Werker ().
https://doi.org/10.1017/9781009367509.010 Published online by Cambridge University Press
How Rents Affect Dictatorship and Violence
strategy is attractive since temporal variation in the world price of oil should not be affected by economic and political conditions in poor, non-oil-producing aid recipients. This exogeneity assumption, however, may be weakened if underlying differences between the treatment (Muslim aid recipients) and control (non-Muslim aid recipients) in the pre-treatment (i.e., prior to the first oil price shock in ) period are not properly accounted for. As the analysis in Part II of the book demonstrated, one plausible candidate is the role of history. Here, I emphasized the role of Muslim conquest in shaping the subsequent institutional trajectory of many Muslim countries (i.e., conquest societies) relative to nonMuslim countries. Societies exposed to Muslim conquest tend to exhibit a set of “sticky” nondemocratic political structures in the contemporary period. Since societies can be associated with experiencing Muslim conquest or not – a feature that is country-specific but does not vary over time (in the contemporary period) – this historical process can be controlled for in a regression framework with country fixed effects. More generally, controlling for country fixed effects is a flexible way to account for other time-invariant country characteristics such as geography, resource endowments, climatic conditions, and colonial legacy that might affect a country’s economic and political development. Furthermore, to allay concerns that a country’s level of democracy prior to the first oil shock (the treatment) might differentially affect a country’s propensity to experience civil conflict, I also control for a country’s POLITY score in and interact that with the price of oil, year fixed effects (to account for the direct effect of annual oil price movements), and important time-varying country characteristics (e.g., per capita GDP, annual rate of economic growth) in the baseline specification. Including these controls in the baseline specification allows me to causally attribute differences in political outcomes (e.g., incidence of civil war) in the post-treatment period between Muslim and non-Muslim recipients, conditional on other potential explanations, such as economic growth, global shocks, and country-specific time-invariant characteristics (e.g., colonial legacy, geography). In Section ., I describe these potential confounders in more detail.
. : () A Case Vignette: Somalia The relationship between foreign aid and political violence that I analyze crossnationally in this chapter is starkly apparent in the case of General Said Barre’s Somalia. While ethnically diverse, Somalia has been predominantly Muslimmajority since at least . The country’s proximity to the Arabian Peninsula
Rather, supply decisions in oil producers (e.g., OPEC) and demand conditions in developed countries (i.e., those that tend to be aid donors and not recipients) are the main drivers of world oil prices.
https://doi.org/10.1017/9781009367509.010 Published online by Cambridge University Press
Aiding Political Violence
facilitated its conquest by Muslim armies, where by , percent of its modern territory was “exposed” to Islam and its initial governing institutions (Kennedy , Lapidus ). With regards to foreign aid, since its independence from colonial rule in , Somalia has received considerable sums of foreign assistance. From through , Somalia received, on average, foreign aid equal to . percent of its GDP – from the Soviet Union, then the United States, plus the Arab League and the United Nations, among others (Besteman , ). Following the loss of Soviet support, Barre dropped a scientific socialist platform and adopted clannism as a method for maintaining power. With increasing foreign aid (with a large share coming from Gulf donors), Barre favored the clans in his inner circle and bombed his rival clan’s strongholds. Correspondingly, the regime became increasingly repressive and consolidated its nondemocratic rule. During the s and s, Siad Barre ruled with “unlimited” political authority (based on the executive constraints index from POLITY) where political and civil liberties were ranked at the lowest possible levels of “not free” (according to Freedom House). The average normalized POLITY score was . (out of ), falling well below the conservative threshold for democratic governance of .. As the price of oil fell in the mid-s and Gulf aid declined, Barre’s policy of divide and rule (called “clan-klatura”) would ultimately lead to his destruction as the excluded clans expressed their dissatisfaction through armed insurgency. “Foreign aid,” observed one Somalia scholar, “provided the glue that held the system together in spite of internal waste and corruption” (Adam , ). Similar dynamics unfolded in other Muslim aid recipients, as I document in the empirical analyses below. B Trends in the Raw Data As a preliminary exploration of these broader dynamics, I plot aid and conflict differentials between Muslim and non-Muslim recipients. To be consistent with my analysis in the previous chapter, I measure a country’s experience (incidence) with civil war using data from Armed Conflict Database/PRIO. Figure . depicts the overall trends in oil prices and the annual incidence of civil war in non-oil-producing Muslim countries. At both thresholds of civil war – at least and , battle deaths per annum in Panels A and B of Figure . respectively – violence remained relatively stable and low until the early s; a period that coincided with high oil prices (and therefore higher aid inflows). The propensity for conflict in these countries picked up significantly from through the late s when oil prices declined (and with them, aid inflows). Since the sample of countries in Figure . does not produce any oil, the positive correlation between oil prices and political stability (i.e., lower incidence of civil war) is unlikely to operate directly via oil prices. Rather, as I show shortly, it operates through the oil-price induced foreign aid disbursement by Gulf oil donors to developing, non-oil-producing Muslim countries.
https://doi.org/10.1017/9781009367509.010 Published online by Cambridge University Press
How Rents Affect Dictatorship and Violence
0
0
20
0.1
40 60 Oil price
Incidence 0.2 0.3
80
0.4
0.5
100
Panel A: Incidence of civil conflict (at least battle deaths per annum)
1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 Year Conflict
Oil price
0
0
20
0.05
Incidence 0.1 0.15
40 60 Oil price
0.2
80
0.25
100
Panel B: Incidence of civil war (at least , battle deaths per annum)
1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 Year Civil war
Oil price
7.3 Political violence in Muslim aid recipients Notes: Oil prices are in 2009 US dollars. In each panel, civil war/conflict is the average incidence among nonil-roducing Muslim countries.
FIGURE
https://doi.org/10.1017/9781009367509.010 Published online by Cambridge University Press
–0.1
–2
0
0
Civil war
Foreign aid 2 4
0.1
6
0.2
8
Aiding Political Violence
1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 Year Aid differential
Civil war differential
FIGURE . Differential in foreign aid and civil war incidence between Muslim and non-Muslim recipients Notes: In the figure, the differential is the difference in average aid (or civil war incidence) between Muslim and non-Muslim aid recipients. For example, a positive differential for foreign aid implies the “typical” Muslim country received more aid than the typical non-Muslim country.
Figure . shows this more directly. It examines the relationship between the aid differential and conflict differential across Muslim and non-Muslim non-oil producers. Muslim countries experienced less conflict than non-Muslim countries when they received comparatively more aid. As the aid differential reversed due to lower oil prices from the mid-s until the early s, Muslim countries were substantially more likely to experience civil war. The differential effects of foreign aid on civil war are large and statistically significant. Table . presents difference-in-differences (DD) estimates for the incidence of civil war between periods of high and low prices across Muslim and non-Muslim aid recipients. Prior to the oil crisis, Muslim countries were approximately percentage points more likely than non-Muslim countries to experience civil war (although this difference is not statistically significant). During the period of high oil prices, Muslim countries were percentage points less likely than non-Muslim countries to experience civil strife. As the windfall vanished in the mid-s, the propensity for civil war spiked dramatically in Muslim countries: It was percentage points more likely than in non-Muslim countries.
https://doi.org/10.1017/9781009367509.010 Published online by Cambridge University Press
How Rents Affect Dictatorship and Violence
. Incidence of civil war in Muslim and non-Muslim non-oil producers, by periods of low and high oil prices
TABLE
Before: – During: – After: – Post: –
Non-Muslim
Muslim
Difference
Std. Error
Difference-in Differences
Std. Error
()
()
()
()
()
()
.
.
.
(.)
.
.
−.
(.)
−.
(.)
.
.
.
(.)
.
(.)
.
.
.
(.)
−.
(.)
Notes: Each cell in columns and report the average incidence of civil war for that type of country (indicated on column head). The standard error corresponding to the difference or difference-indifferences are reported in columns and , respectively. These standard errors (in parentheses) are clustered by country.
The DD estimates in Table . imply that the aid windfall between and made Muslim countries percentage points relatively more stable compared with the underlying conflict propensity in non-Muslim countries. The end of the windfall in the mid-s engendered a relative rise in high-intensity violence as Muslim countries became percentage points more likely to engage in a civil war, while non-Muslim countries became slightly (around percentage) more stable. This effect is statistically significant ðp-value ¼ 0:054Þ. For the period after , when higher oil prices prevailed, the propensity for the high-intensity civil war declined in Muslim countries (relative to the prior period), but they were as stable as non-Muslim countries. The validity of the DD strategy assumes the underlying level of civil conflict between Muslim and non-Muslim recipients exhibit “similar” trends prior to the oil price shock (i.e., the so-called parallel trends assumption). Yet this may not be the case, for example, due to unobserved differences between these groups of countries (e.g., due to differences in fertility) or differential trends by geographical regions (e.g., regional “waves” of democratization). Fortunately, the difference-in-differences estimates in Table . are robust to the inclusion of a Muslimi x Year trend and a vector of regional dummies interacted with a year trend as controls (e.g., Africai Year, Middle Easti Year). The former controls for any unobserved temporal trend between Muslim and non-Muslim countries, while the latter accounts for unobserved trends across regions.
https://doi.org/10.1017/9781009367509.010 Published online by Cambridge University Press
Aiding Political Violence
. A Specification While Figures . and . capture the underlying dynamics linking oil prices to aid inflows, and that aid to civil war, they do not control for recipient characteristics that might mediate the effect of aid on civil war (e.g., economic growth). To address this concern more systematically, I use a two-stage least squares (SLS) setup to identify the causal effect of oil prices on civil war via the aid channel and limit my analysis to developing countries that do not produce any oil. Specifically, I estimate variants of the following sets of regressions: First stage : Aidit ¼ a þ b∗ Muslimi pðoilÞt þ c∗ Xit þ di þ dt þ eit Second stage : Civil Warit ¼ α þ β∗ Aidit þ γ∗ Xit þ δi þ δt þ εit where Civil Warit measures the incidence of high-intensity civil war in country i in year t; Aidit is foreign aid (% GDP); Muslim is a dummy variable equal to if a country is Muslim majority and zero otherwise; p(oil)t is the world price of oil in year t; Xit is a vector of time-varying recipient characteristics (e.g., log GDP per capita, economic growth), δi and δt are a set of country and year fixed effects respectively, and εit is a stochastic error term. So as to not estimate the direct effect of oil prices on civil war, the sample is restricted to eighty-seven developing, non-oil-producing countries. By interacting whether an aid recipient is Muslim with the price of oil, the instrument identifies the differential effect of oil prices on aggregate aid inflows into Muslim countries (relative to non-Muslim countries) and thus captures movements in the aid differential in Figure .. Thus, the instrument measures how exogenous changes in world oil prices are propagated to Muslim non-oil producers via the aid channel. The inclusion of country fixed effects in both stages control for observed and unobserved time-invariant characteristics that affect a country’s receipts of foreign aid (e.g., colonial relationship, distance) and propensity for conflict (e.g., ethnic fragmentation, history of violence, geography) and subsumes the main effect of Muslimi in the interaction term. In the second stage regression, the inclusion of these fixed effects means the coefficient β gauges the withincountry effect of aid on civil war. Finally, the inclusion of year fixed effects ðδt Þ
I use foreign aid as a share of GDP as my preferred measure as this is the most prevalent measure in existing work on aid and conflict (and growth/development). The results are robust to alternate measures such as foreign aid per capita and the aggregate level (dollar amount) of foreign aid, see Table .. Following the same logic as the preceding chapter, the standard errors are corrected for serial correlation with the Newey-West procedure for -period autocorrelation. I also run the results clustering at the level of the instrument, or Muslim*Year, and the standard errors shrink and hence are more statistically significant (results not reported).
https://doi.org/10.1017/9781009367509.010 Published online by Cambridge University Press
How Rents Affect Dictatorship and Violence
captures all common temporal trends/shocks that impact all countries (e.g., oil shocks, the Cold War) and subsumes the annual variation in oil prices in the interaction term. I control for important recipient characteristics that can affect both the allocation of aid in the first stage and the incidence of civil war in the second stage. For example, existing studies find that poorer countries (e.g., Fearon and Laitin ) and those experiencing slower (negative) growth are more prone to conflict (e.g., Savun and Tirone ) and these countries are more likely to receive foreign aid (e.g., Alesina and Dollar ). Accordingly, I control for: economic development (log GDP per capita), growth (GDP per capita growth), and population size (in logarithmic units). I also account for the possibility that Muslim countries may exhibit structural demographic and/or economic conditions that might be differentially affected by oil prices. Here, I control for the share of rural population in (i.e., prior to the treatment period) and whether a country was autocratic in , each interacted with p (oil)t.
B Main results Table . presents the effect of oil-induced aid flows on civil war, conditional on potential confounding effects. I begin by examining the relationship between oil prices and aid inflows to highlight that oil prices explain variation in aid receipts in Muslim countries only. Columns and capture the dynamics underlying the correlation between the aid differential and oil prices in Figure .. In column , I restrict the analysis to Muslim recipients only. It shows that oil prices have a robust positive and statistically significant effect on aid inflows. In contrast, the price of oil does not affect aid inflows in nonMuslim recipients (column ). Pooling the samples, column reports the first stage regression describing the effect of oil prices on foreign aid. The coefficient estimate implies that oil prices have a positive and statistically significant effect in raising aid disbursements to Muslim recipients: A $ increase in oil prices raises aid receipts in Muslim countries by . percent of GDP. The instrument’s F-statistic ð¼ 25:28Þ exceeds the threshold for weak instruments of (Staiger and Stock ), thus strengthening my confidence in drawing causal inferences in the second stage regression. Armed with this strong instrument, the results in columns – shows that foreign aid causes a reduction in the incidence of civil war in Muslim recipients. In column , a one percentage point increase in aid lowers the propensity of civil war by . percentage points; alternatively, a standard deviation change in aid in Muslim recipients reduces the incidence of civil war by about one
The economic and demographic variables are from the World Bank (). Measures of political institutions are from the POLITY IV data set.
https://doi.org/10.1017/9781009367509.010 Published online by Cambridge University Press
https://doi.org/10.1017/9781009367509.010 Published online by Cambridge University Press
TABLE
. The causal effect of oil-price-induced foreign aid on civil war
Dependent variable:
Oil price
Foreign aid (% GDP) ()
()
Muslim . (.)***
Non-Muslim . (.)
Muslim Oil price
Incidence of civil war ()
()
()
()
. (.)*** −. (.)**
Foreign aid (% GDP)
−. (.)**
Foreign aid per capita
−. (.)**
Foreign aid (Millions, US$)
−. (.)**
Log foreign aid Recipient characteristics Country fixed effects Year fixed effects F-statistic on instrument No. of observations
()
Yes
Yes
Yes Yes Yes
Yes Yes Yes
Yes Yes Yes
Yes Yes Yes
Yes Yes Yes
.
.
.
.
.
Notes: Estimation via OLS in columns – and SLS in columns –. Newey-West corrected standard errors with up to lag autocorrelation in parentheses. **, *** = significant at and percent respectively. Recipient characteristics include low GDP per capita, GDP per capita growth (% annual), log population, POLITY in oil price, and percent rural in oil price. These coefficients, country and year fixed effects, and a constant are not reported. Oil prices are in US$. In column , foreign aid per capita is measured in US$. In columns and , foreign aid is measured in millions of US$ and logð1 þ foreign aidÞ, respectively.
How Rents Affect Dictatorship and Violence
standard deviation (equal to percentage points). While measuring aid (as a share of GDP) is prevalent in the conflict literature, columns – show that results do not hinge on this. Aid has a negative effect on civil war when measured in per capita terms (column ), in levels of US dollars (column ), and in logarithmic units (column ). The coefficient in column , for example, implies that a $ increase in aid per capita lowers the likelihood of conflict by percentage. To put this in historical context, the typical Muslim aid recipient was about percentage points less likely to experience civil war (relative to non-Muslim countries) during the period of high oil prices (–). The findings in Table . hold across a variety of alternate specifications, including those where the second stage is estimated via probit (since the dependent variable is binary); do not include fixed effects, but do appropriately control for the constituent terms of the interaction term, Muslimi and p(oil)t; and account for the potential budgetary effects of aid on conflict by including total government expenditures and fuel imports as additional controls. The main findings also do not hinge on whether aid is measured contemporaneously or with a lag. For example, as Burcu Savun and Daniel Tirone () show, some aid could be saved and used to stall rebellion in future years. Columns – in Table . shows that instrumented aid from one, two, and four years in the past reduces the incidence of contemporaneous civil war in Muslim recipients. And as one would expect, these effects tend to attenuate over time as the coefficient estimates decline in magnitude as the length increases. For instance, aid from the previous year reduces the contemporaneous incidence of civil war by . percentage points (column ). In contrast, this dampening effect is reduced to . percentage points with aid from four years in the past (column ). Relatedly, “smoothed” aid also dampens civil war in Muslim recipients. Columns and show that two- and four-year moving averages of foreign aid receipts also exhibits a robust negative effect on civil war. The results are also robust to specifications that account for various potential manifestations of unobserved heterogeneity. Here the concern is that factors unrelated to aid receipts could account for differences in conflict propensities across Muslim and non-Muslim countries. One possibility is the role associated with the role of geopolitics in shaping “technologies” of armed rebellion after the Cold War (Kalyvas and Balcells ). This argument, however, misses the point of the treatment and control groups in the quasiexperimental design: One would have to argue that the end of the Cold War differentially affected political violence in Muslim countries. Yet Cold War politics, such as “influence” from the United States and the Soviet Union, did not discriminate on religion. Indeed, as Figure . shows, the rise of civil war in Muslim aid recipients began in the mid-s, well before the Cold War ended. And more rigorously, the inclusion of year fixed effects in the baseline SLS
These results are reported in table from Ahmed and Werker ().
https://doi.org/10.1017/9781009367509.010 Published online by Cambridge University Press
https://doi.org/10.1017/9781009367509.010 Published online by Cambridge University Press
TABLE
. Alternate measures of foreign aid and accounting for unobserved heterogeneity
Dependent variable:
Incidence of civil war ()
()
()
()
()
−. (.) **
Foreign aid
Foreign aid in t-
Foreign aid in t-
Foreign aid in t- Foreign aid, year moving average Foreign aid, year moving average
()
() −. (.) **
() −. (.) **
() −. (.) **
−. (.) *** −. (.) ** −. (.) ** −. (.) *** −. (.) ** (continued)
https://doi.org/10.1017/9781009367509.010 Published online by Cambridge University Press
TABLE
. (continued)
Dependent variable:
Incidence of civil war ()
()
()
()
()
Additional controls Muslim x Cold War
()
()
()
−. (.) . (.)
Muslim x Year Region FE x Cold War Region FE x Year No. observations
()
Yes Yes
Notes: Estimation via SLS. Newey-West corrected robust standard errors with up to lag autocorrelation reported in parentheses. **, *** = significant at and percent respectively. Foreign aid is measured as a share of GDP. All specifications include recipient characteristics, country and year fixed effects, and a constant. These coefficients are not reported. Recipient characteristics include: Recipient characteristics include: low GDP per capita, GDP per capita growth (% annual), log population, POLITY in x oil price, and percent rural in x oil price. The instrument is “strong” in the first stage across all the specifications (F-statistic exceeds ).
Aiding Political Violence
specification subsumes the effects of temporal factors that affect all countries in the sample, such as the Cold War and post–Cold War period. Nevertheless, to account for this potential differential effect, I include the interaction of Muslimi with an indicator variable for the Cold War period (i.e., equal to zero before and one after ) as an additional control. As column in Table . shows, the robust negative effect of instrumented aid on civil war holds in such a specification, while Muslimi Cold Wart does not exhibit a robust effect. Another concern is whether Muslim countries are inherently more prone to political violence, particularly as many of these societies exhibit “sticky” authoritarian structures (as Part II of this book demonstrates). The decision to include country fixed effects in the baseline SLS specification mitigates this worry. By including country fixed effects, the SLS regressions estimate the within-country variation of aid on civil war and thus eliminates any potential “Muslim” effect. However, it is plausible that this proposed Muslim effect might be time-varying, for example, due to unobserved differential trends in demographics (e.g., fertility, “youth bulge”) and/or religious intensity between Muslim and non-Muslim countries. This worry is not borne out empirically: Controlling for this differential effect through the inclusion of a Muslimi Year trend as an additional control does not nullify the robust negative effect of aid on civil war (Table ., column ). Of course, rather than unobserved differential trends between Muslim and non-Muslim countries, skeptics may worry that unobserved differential trends across regions (e.g., regional “diffusion” of civil war and democracy) may be influencing the results. Again this does not seem to be the case empirically. Columns and in Table . address this concern by controlling for a vector of differential trends by region. Column interacts a vector of region dummies (e.g., Africa, Asia) with an indicator variable for the Cold War period, while column does so with a year trend. In both specifications, greater aid receipts dampen the incidence of civil war. C Evaluating the Exclusion Restriction As with all analyses that use instrumental variables, the validity of the causal estimates hinges on whether the exclusion restriction is met. In the context of the quasi-natural experiment in this chapter, the worry is whether other factors that are correlated with oil prices – but independent of foreign aid disbursements – may be driving the differential propensities of civil war within Muslim non-oil producers (i.e., potential violations of the exclusion restriction). Below, I investigate the main alternative explanations and provide evidence that despite the “contingent” nature of the instrumental variable, the chapter’s main results remain robust. Remittances The first alternative explanation pertains to another type of international capital received by Muslim non-oil producers during the period of high oil
https://doi.org/10.1017/9781009367509.010 Published online by Cambridge University Press
How Rents Affect Dictatorship and Violence
prices. At the same time Persian Gulf oil producers were disbursing foreign aid, they “imported” more foreign workers from non-oil-producing Muslim countries (e.g., Jordan, Pakistan, Bangladesh). These workers, in turn, remitted their income to their families back home. Like foreign aid, remittances to non-oilproducing Muslim countries are strongly and positively correlated with the price of oil (Ahmed ). During the period of high oil prices, Muslim non-oil producers experienced a remittance boom. However, unlike foreign aid, the main beneficiaries of remitted income are households and not directly the government. Consequently, the link between remittances and civil war is not as direct. On the one hand, remittances may “buy” stability via a substitution effect between welfare goods and patronage, which tends to be stronger in countries governed by nondemocratic regimes (Ahmed , ) and decrease the risk of conflict by inducing economic growth (Regan and Frank ). Yet, on the other hand, higher remittance income should enrich individuals and may mobilize insurgency (e.g., by increasing demand for political liberalization). Given these opposite effects of remittances on civil war, reconciling the net effect remains an empirical matter. One strategy to do so is to control for aggregate remittances in the baseline specification. Column in Table . shows that accounting for remittances (% GDP) magnifies the dampening effect of instrumented aid on civil war in Muslim recipients. The estimated effect (= –.) is larger in magnitude than those reported in Tables . and .. There are two important implications of this finding. First, excluding remittance inflows does not constitute omitted variable bias. Second, and more importantly, controlling for remittances does not weaken the effect of instrumented aid in reducing civil war propensity in Muslim recipients and thus, does not comprise a violation of the exclusion restriction (at least econometrically). Exporting Islamic Extremism A second, and perhaps more troublesome, explanation is that higher oil prices empowered Gulf oil producers to meddle in the domestic politics of a few nonoil-producing Muslim countries, independent of the foreign aid channel. After all, when the price of oil was high, OPEC countries were not only giving money away, but they also often strove to “export” their brand of politics abroad. There were three major players whose influence benefited from the high oil prices: Ayatollah Khomeini in Iran, Muammar al-Qaddafi in Libya, and the religious hardliners in Saudi Arabia. Khomeini tried to export the Iranian revolution, Qaddafi funded insurgencies of all sorts, and the Saudis pushed Wahhabi beliefs on Muslims around the world. These policies had the potential to lead to instability down the road in Muslim countries that also were the
The sample size in column is smaller than those reported in Tables . and . since remittance data is spotty in the s.
https://doi.org/10.1017/9781009367509.010 Published online by Cambridge University Press
https://doi.org/10.1017/9781009367509.010 Published online by Cambridge University Press
TABLE
. Addressing potential violations of the exclusion restriction
Dependent variable
Foreign aid
Additional controls Remittances
Incidence of civil war (% GDP) ()
()
()
()
()
()
−. (.)***
−. (.)***
−. (.)***
−. (.)***
−. (.)***
−. (.)***
. (.)** . (.)*
Assassinations
. (.)***
Terrorist fatalities Excluded countries No. of observations
Lebanon
Morocco
Lebanon Morocco
Notes: Estimation via SLS. Newey-West corrected robust standard errors with up to lag autocorrelation reported in parentheses. **, *** = significant at and percent respectively. Foreign aid and remittances are measured as a share of GDP. All specifications include recipient characteristics, country and year fixed effects, and a constant. These coefficients are not reported. Recipient characteristics include: low GDP per capita, GDP per capita growth (% annual), log population, POLITY in oil price, and percent rural in oil price. The instrument is “strong” in the first stage across all the specifications (F-statistic exceeds ).
How Rents Affect Dictatorship and Violence
beneficiaries of the foreign aid windfalls. As I discuss in greater detail below, the balance of evidence, however, suggests that none of these three stories biases the findings in this chapter. There are two specific dangers that any of these stories might pose to the analysis. First, the political meddling by Iran, Libya, or Saudi Arabia could lead to the data being coded in a way that would falsely indicate a rise in civil war (i.e., non-internationalized internal conflict) after the fall of oil prices. The specific concern is that these countries may have been intervening in domestic conflicts while Muslim non-oil producers were receiving aid windfalls. Second, the foreign meddling by these countries during the high oil price years could have planted the seeds for future internal conflict in a channel unrelated to foreign aid. Most saliently, the export of radical Islam gained steam during the s and early s. Perhaps the ideas promulgated during the high oil price years needed time to germinate before they would realize themselves in domestic instability after the oil price had fallen. The first concern is the easiest to address. As it turns out, there were only two non-oil-producing Muslim countries that, according to the Armed Conflict Database (ACD), experienced internal conflict over the period –, which entailed involvement by other Muslim countries for some duration of the conflict. They were Lebanon (–) and Morocco (–). In and , Lebanon’s conflict is not coded as internal because of intervention on the part of Israel and Syria. It seems reasonable to code this as an internationalized internal conflict: The typical analysis of this conflict is that Lebanon served as the fighting grounds for the war between Israel and the Palestinian Liberation Organization as well as Syria (Murphy , –). In the ACD, Morocco experienced civil war from to . For the first five years, this conflict was coded as internationalized (only in by ACD though) because of Mauritanian involvement in putting down the rebel group Polisario’s secession aims for Western Sahara. After a coup in Mauritania, the new government signed a cease-fire with Polisario and granted them control over disputed territory. ACD codes the conflict from to between the Moroccan government and the rebels as internal. To make sure the results are
If Libya, for example, were intervening in an internal war in Niger while the price of oil was high, the war in Niger could have been coded as internationalized while Niger received a foreign aid bonus from the Gulf states. Then, when the price of oil fell, Libya might have run out of money and stopped intervening – resulting in the conflict being coded as “non-internationalized” just as the foreign aid fell. This would generate the same results that this chapter documents: a rise in foreign aid coinciding with the fall in civil war, then the fall in foreign aid coinciding with a rise in civil war. Certainly, this understates Libyan involvement. Geoff Simons (, ) cites an Israeli intelligence report that in Qaddafi was supporting some “fifty terror organizations and subversion groups, in addition to more than forty radical governments in Africa, Asia, Europe and America.” If all these interventions had been coded as internationalized, however, the results could be biased toward seeing all domestic conflict as internationalized.
https://doi.org/10.1017/9781009367509.010 Published online by Cambridge University Press
Aiding Political Violence
not driven by these two conflicts, I re-estimate the SLS specifications dropping observations from Lebanon, Morocco, or both countries (Table ., columns –). The results remain unchanged from those in Table .. The second concern, that Islamic political influences during the period – planted the seeds for future internal conflict, is much harder to discount. The evidence, however, suggests that the rise in Islamism and the increasing power and political activism of traditionalist religious groups is an outcome rather than a cause of the political economy of the state that is fed by foreign aid. For example, during in Iran, Ayatollah Khomeini led a successful revolution that seized power from the American-backed shah, whose regime had ignored crucial segments of the populace. The Iranian revolution spawned what Gilles Kepel () calls a “ripple effect”: Regimes in Muslim countries viewed the shah’s fate as an object lesson, and many of them became ostentatiously religious, in the hope of avoiding what had befallen the Persian monarch, who had never bothered to hide his contempt for the “men in black.” Governments sought to head off social movements that, by annexing the vocabulary of Islam, threatened to unite everyone with an axe to grind and bring down the established power. The ulemas, having been steadily harassed during the nationalist period, now found themselves fawned upon by princes eager for the Islamic legitimacy their blessing could confer. In return, the clerics demanded greater control over culture and morals. (, )
In other words, the Islamic Revolution in Iran had an indirect, positive effect on Islamism in other Muslim countries (and potentially non-Muslim countries where the successes in Iran could be adapted). Governments, feeling distant from and threatened by their constituencies that they were otherwise repressing or buying off with the foreign aid rents, made concessions to the segment of the populace that was most threatening. In this case, that segment was religious, but there are other examples during the same time period where the segment was not religiously defined, as in Somalia. The foreign aid windfall could certainly have been used to help make these concessions. With respect to the spread of Wahhabism from Saudi Arabia, Gilles Kepel acknowledges the shift in power in the Muslim world to the Gulf states upon the beginning of the oil embargo. Yet he is skeptical that this wealth was able to influence any fundamental change in developing Muslim countries: “[Saudi Arabia’s] immense financial generosity had won it a following that was more venal than sincere, and the Wahhabization it wished to implement had tended to fluctuate with the price of a barrel of oil” (Kepel , ). An example of the failure of Saudi’s export of Islam can be seen in the far west of Africa. In Senegal, stable except for the separatist Casamance region in the south (which was not an Islamist movement), the influence of fundamentalist Islam was easily controlled. According to Kepel (, ): When Islamism began to appear in Senegal in the late s among students who were inspired by the Iranian revolution or educated in Arab universities in the Middle East, it
https://doi.org/10.1017/9781009367509.010 Published online by Cambridge University Press
How Rents Affect Dictatorship and Violence
immediately collided with the power of the marabouts. Radical Islamism was allowed a hearing only so long as it did no harm to the marabouts’ interests; as soon as the movement showed signs of doing so, the religious leaders squashed it.
Likewise, in Morocco – where King Hassan II had claimed to be a direct descendant of Prophet Mohammed – the reach of Wahhabism was limited. The Islamist movement encountered “difficulty in contriving a political rupture with the state without calling into question the monarch’s sacredness, a taboo to which the mass of the populace was resolutely opposed” (Kepel , ). More generally, even if the Islam exported by Saudi Arabia was not inherently successful in infiltrating the dominant practice of Islam in recipient countries, it may have encouraged radical groups at the fringe of society. This is equivalent to saying that the level of extremism increased as a result of Saudi influence. If such a story is correct, this implies that this chapter’s analysis thus far ignores an important factor: religious extremism. In order to measure the effect of aid on instability independent of extremism, one must control for the latter. To check whether this indeed might be the case, I account for assassinations and terrorist-related fatalities as proxies for extremism to evaluate whether extremism produced alongside the foreign aid windfalls can explain the patterns of civil war. I repeat the empirical analysis controlling for each country’s annual number of assassinations (available from Banks ) and terroristrelated fatalities (from START ). Controlling separately for assassinations and terrorist fatalities (Table ., columns and ) does not appreciably change the pattern between the foreign aid windfall and civil war. On balance, accounting for the potential effects associated with the “exporting” of Islamic extremism does not affect the chapter’s substantive findings. In summary, the evidence supports the notion that movements in Islamism among developing Muslim nations – though coinciding with oil price movements and Saudi foreign policy – were a fundamental part of those nations’ internal politics and the consequences of elites otherwise repressing the population. The robustness checks also suggest that the specificities of coding the data cannot alone explain the pattern in conflict, nor can the omission of religious extremism that might have been independently generated by the foreign policies of Iran and Saudi Arabia in particular. D Mamluk Recipients The results so far demonstrate the average treatment effect of oil-price induced foreign aid across Muslim non-oil producers comprised of countries that were both exposed to mamluk institutions (e.g., Jordan, Lebanon, Morocco, Pakistan) and those that were not (e.g., Bangladesh, Mali, Niger). Therefore, this average treatment effect is likely to mask whether cross-national differences in preexisting cohesive institutions matter in explaining the relationship between foreign aid and civil war. Indeed, a key prediction from the formal
https://doi.org/10.1017/9781009367509.010 Published online by Cambridge University Press
Aiding Political Violence
model in Chapter stipulates a heightened propensity for civil war following a decline in rents in countries whose preexisting institutions are less cohesive. In the context of Muslim recipients – and recalling the main finding in Part II of the book – this implies that greater amounts of foreign aid should engender greater political stability (i.e., less civil war) in countries exposed to mamluk institutions. This is observationally equivalent to showing that a decline in aid increases the likelihood of civil war in countries whose preexisting institutions are less cohesive (i.e., societies exposed to mamluk institutions). To evaluate this conjecture, I parse the treatment effects associated with Muslim recipients exposed to mamluk institutions (“mamluk aid”) and those that were not (“Non-mamluk aid”). Specifically, mamluk aid is equal to the product of a country’s annual foreign aid receipts and the percentage of territory governed by mamluk institutions by (i.e., Mi from Chapters –). Thus, mamluk aid essentially weights a Muslim non-oil producer’s annual receipts of aid by its historical exposure to mamluk institutions. This weight is higher among countries where Mi is greater; for example, Mi ¼ 0:85 for Pakistan and Mi ¼ 1 for Jordan and Lebanon. I then use the instrumental variable to derive clean causal estimates for mamluk and non-mamluk aid in Muslim non-oil producers. Table . reports the results. As a benchmark, I begin by examining the effect of oil-price induced aid in non-mamluk recipients. Column shows that foreign aid lowers the propensity of civil war in non-mamluk Muslim recipients. The causal estimate is statistically significant and similar in magnitude to those reported in Tables .–.. Importantly, in confirmation of corollary A from Chapter , column reports a more pronounced effect of foreign aid on civil war in mamluk recipients. The coefficient estimate is about three times larger than that associated with non-mamluk aid recipients and implies that a percentage point increase in foreign aid (as a share of GDP) lowers the propensity for civil war by percentage points in mamluk recipients (e.g., Afghanistan, Lebanon). This pacifying effect in mamluk recipients is present when measuring aid on a per capita basis (column ) and holds when controlling for potential confounders associated with oil prices that might violate the exclusion restriction (column ). As noted above, an equivalent interpretation of the results in columns – is that a decline in foreign aid heightens the incidence of civil war in mamluk recipients. Thus, since countries exposed to mamluk institutions tend to exhibit less cohesive institutions in the contemporary period (as demonstrated in Chapter ), this is a direct confirmation of corollary A in Chapter .
This observational equivalence is therefore a direct test of proposition A. Non-mamluk aid is equal to a country’s annual foreign aid receipts for Muslim countries not exposed to mamluk institutions, and set to zero for Muslim countries exposed to any mamluk institutions. The coefficient estimate is less precisely estimated (i.e., lower p-value) due in part to a smaller sample of mamluk aid recipients.
https://doi.org/10.1017/9781009367509.010 Published online by Cambridge University Press
How Rents Affect Dictatorship and Violence
. The heightened impact of aid on civil war in countries exposed to mamluk institutions
TABLE
Dependent variable:
Civil war ()
Non-mamluk aid
()
()
−. (.)*** −. (.)*
Mamluk aid
−. (.)** −. (.)**
Mamluk aid per capita
. (.) −. (.) . (.)**
Remittances Assassinations Terrorist fatalities Recipient characteristics Country fixed effects Year fixed effects No. observations
()
Yes Yes Yes
Yes Yes Yes
Yes Yes Yes
Yes Yes Yes
Notes: Estimation via SLS. Newey-West corrected robust standard errors with up to lag autocorrelation reported in parentheses. *, **, *** = significant at , , and percent respectively. Foreign aid and remittances are measures as a share of GDP. All specifications include recipient characteristics, country and year fixed effects, and a constant. These coefficients are not reported. Recipient characteristics include GDP per capita, GDP per capita growth (% annual), log population, POLITY in oil price, and percent rural in oil price.
. A Aiding Dictatorship Despite experiencing political turbulence when aid flows declined, Muslim recipients remained staunchly nondemocratic during periods of high and low aid receipts. Figure . illustrates this authoritarian resilience. It plots the average levels of democracy (using the normalized POLITY score) and aid receipts among Muslim recipients. The figure distills two important patterns. First, the average level of democracy has remained remarkably stable, with a value hovering around . and within a very tight band of . index points. Second, these Muslim non-oil producers tended to become slightly more authoritarian when aid receipts were buoyant. Estimations via SLS regression confirm these patterns. In a specification that controls for recipient characteristics, country and year fixed effects, instrument aid reduces a country’s normalized POLITY score. The estimated effects imply that during the period of high
https://doi.org/10.1017/9781009367509.010 Published online by Cambridge University Press
10 5 0
0.2
Foreign aid (% GDP)
POLITY (normalized) 0.21 0.22 0.23
15
0.24
Aiding Political Violence
1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 Year POLITY FIGURE
Foreign aid
. Aiding authoritarianism in Muslim recipients
oil prices, the typical Muslim recipient experienced a −. reduction point reduction in its normalized POLITY score. In relation to the variation illustrated in Figure ., this is equivalent to the entire movement in democracy in Muslim aid recipients during the period of high oil prices (–). The resilience of dictatorship is reassuring, even in those countries that experienced civil war. Table . describes the trajectory of democracy one, five, and ten years after the end of civil war in Muslim recipients. For both the modal (median) and average aid recipient, the normalized POLITY score remained well below a conservative cutoff for democracy (e.g., normalized POLITY score greater than .). It is worth noting that the average level of democracy (ranging from . to .) exceeds the modal value. This is due to two Muslim democracies that experienced civil war: Lebanon and Pakistan. Patterns of political violence in these two countries demonstrate that even Muslim recipients that qualify as democracies are vulnerable to internal violence and highlight the fragility of democracy in these countries. For example, since its independence from the United Kingdom in , Pakistan has cycled
The coefficient estimate on instrumented aid (% GDP) is −. with a standard error of . ðp-value ¼ 0:000Þ. The typical country received aid equal to . percent of GDP between –. Thus, the total effect is 10:7∗ −0:002 ¼ −0:021. To be clear, the sample of countries in Table . are restricted to Muslim non-oil-producing (aid recipients) that experienced civil war.
https://doi.org/10.1017/9781009367509.010 Published online by Cambridge University Press
TABLE
How Rents Affect Dictatorship and Violence . Trajectory of democracy in Muslim recipients after civil war POLITY (normalized) in … Terminal year
Mean Median
Years after the end of civil war
()
()
()
()
. (.) .
. (.) .
. (.) .
. (.) .
Note: Standard deviations for means reported in parentheses.
back and forth between periods of democracy and military dictatorship. While in Lebanon, despite constitutional provisions that formally share political power among the country’s dominant ethnic groups as part of its confessional democratic framework, the country has experienced intense periods of violent civil strife.
B Geopolitical Rents The greater vulnerability of civil war in Muslim aid recipients (non-petrostates) contrasts sharply with those in Muslim petrostates which have remained relatively stable and authoritarian. As the previous chapter demonstrated, this stems in part from the additional stable supply of US geopolitical rents to Muslim petrostates, particularly in the Persian Gulf. These rents have served to thwart insurgency while preserving dictatorship in Muslim petrostates. In contrast, the provision of geopolitical rents to Muslim non-petrostates has been less stable. In Muslim non-petrostates, US geopolitical rents have come mainly in the form of military and economic aid with very minimal (if any) formal troop deployments (unlike in Muslim petrostates). And unlike US geopolitical rents to Muslim petrostates, US assistance to Muslim non-oil producers has not sought to directly ensure their domestic political stability. Below, I identify a few cases. In the Middle East, the United States has provided economic and military assistance to several Muslim non-oil producers, provided they advance American foreign policy interests in the region. For instance, Egypt became a major recipient of American military assistance after the signing of the EgyptIsrael Peace Treaty in , which required the Egyptian government to formally recognize Israel’s existence and demilitarize the Sinai. Since then, the
While the United States has also supplied geopolitical rents to several Muslim non-oil producers (e.g., Jordan, Pakistan), this support has exhibited greater variability and frequently imposes political conditions for the recipients.
https://doi.org/10.1017/9781009367509.010 Published online by Cambridge University Press
Aiding Political Violence
Egyptian military has received about $. billion per annum in US military aid, assistance that has helped strengthen the military’s support of nondemocratic rule (Ahmed ). US foreign aid has also been an important source of authoritarian stability in Jordan; again, provided the ruling monarchy remains pro-American (Jamal ). Beyond the Middle East, American support has also been driven by geopolitical concerns, albeit for reasons unrelated to oil prices and largely indifferent to the quality of democracy in Muslim aid recipients. For instance, to repel the Soviet Union’s invasion of Afghanistan in , the United States ramped up its military assistance to Kabul (as well as to Islamabad, who provided logistic support to Afghan forces). After the demise of the USSR and the end of the Cold War in , US assistance to Afghanistan and Pakistan ceased. In the s, Afghanistan would undergo civil war with the nondemocratic Taliban emerging as the country’s de facto government (and training ground for Al Qaeda in the late s and early s). Across the border, with Afghanistan no longer a geopolitical priority, US interest in Pakistan waned in the s. Only after the events of / and the start of the War on Terror did Pakistan reemerge as an area of geopolitical interest for the United States and with that, a dramatic surge in US military and economic assistance. In neither instance did concerns about access to stable oil supplies (or other potential natural resources) or the quality of domestic politics in Afghanistan or Pakistan drive the allocation of US geopolitical rents. On balance, while the United States has provided some geopolitical rents to Muslim non-oil producers, these rents have not entailed formal troop deployments, nor have they comprised a consistent US foreign policy strategy aimed at thwarting domestic insurgencies and preserving the recipient’s nondemocratic longevity (in stark contrast to US foreign policy toward Muslim petrostates).
. As an important source of nontax government revenue (rent), foreign aid is frequently envisioned to promote economic and political development. And for many donors, an important aim for their financial assistance is to facilitate peaceful transitions to democracy. Tragically, this has not been the case in many non-oil-producing Muslim aid recipients. Cognizant of several empirical challenges, most notably endogeneity, I leverage a quasi-natural experiment of oil-price induced aid disbursements to identify causal relationships between foreign aid and political violence in Muslim aid recipients. This chapter presented several findings, the most important of which demonstrated that buoyant levels of foreign aid have strengthened authoritarianism
Of course, when the political winds changed in Egypt and democracy emerged briefly (in the short window after the Arab Spring), US military aid continued to flow to the Egyptian military provided the new government continue to comply with the Egypt-Israel Peace Treaty.
https://doi.org/10.1017/9781009367509.010 Published online by Cambridge University Press
How Rents Affect Dictatorship and Violence
in Muslim non-oil producers by reducing the propensity for civil war. Moreover, these effects have been more pronounced in Muslim aid recipients exposed to the institutional legacy of Muslim conquest via mamluk institutions. In combination with the inferences from Part II of this book, this finding supports corollary A from Chapter : A decline in rents can heighten the prospect of civil war in countries whose preexisting institutions are less cohesive. In contrast, in societies where preexisting institutions are relatively more cohesive, corollary B posits that a decline in rents can lead to peaceful transitions to democracy. The next chapter evaluates this proposition for foreign transfers in societies whose preexisting institutions tend to be more cohesive, at least in relation to many Muslim societies. In doing so, I provide an important analysis of “negative cases” and broaden the empirical evaluation of my analytical framework to non-Muslim societies.
https://doi.org/10.1017/9781009367509.010 Published online by Cambridge University Press
Appendix D Sample
TABLE D
Sample of non-oil-producing foreign aid recipients
Afghanistan Albania Bangladesh Comoros Djibouti Eritrea Guinea
Armenia Belarus Benin Bhutan Bolivia Bosnia Botswana Bulgaria Burkina Faso Burundi Cambodia Cent. Afr. Rep. Chad Chile Congo (DRC) Costa Rica Cote d’Ivoire Croatia Czech Rep. Dominican Rep. El Salvador Estonia Ethiopia
Muslim countries Jordan Kyrgyz Rep. Lebanon Mali Mauritania Morocco Niger Non-Muslim countries Fiji Gambia Georgia Ghana Guatemala Guinea-Bissau Guyana Haiti Honduras Hungary Jamaica Kenya Laos Latvia Lesotho Liberia Lithuania Macedonia Madagascar Malawi Mauritius Moldova Mongolia
Pakistan Senegal Somalia Sudan Tajikistan Turkey
Mozambique Namibia Nepal Nicaragua Panama Paraguay Philippines Poland Rwanda Sierra Leone Slovak Rep. South Africa Sri Lanka Swaziland Tanzania Togo Uganda Ukraine Uruguay Zambia Zimbabwe
https://doi.org/10.1017/9781009367509.010 Published online by Cambridge University Press
Democratic Transitions in Non-Muslim Societies
Declines in rents do not always lead to civil war but may facilitate a pathway toward democracy in societies whose stock of preexisting institutions are more cohesive. Drawing on collaborative research with Dan Schwab and Eric Werker (Ahmed, Schwab, and Werker ), this chapter presents the negative case to the findings in the previous chapter. It does so by probing the effect of increases and decreases in rents on patterns of democracy, dictatorship, and civil war in non-Muslim societies. An important aspect of this chapter’s inferences is the contention that non-Muslim societies tend to have political institutions in the contemporary period that are more cohesive relative to Muslim societies. This assertion stems from the arguments and empirical findings in Part II of the book: Since (most) non-Muslim societies did not experience Muslim conquest and its subsequently institutional trajectory, these societies are less likely to feature a sticky set of authoritarian political structures in the contemporary period relative to Muslim societies (that were exposed to conquest). Combining these features – temporal variation in rents and cross-national differences in preexisting cohesive institutions – provides a narrative to suggest that a decline in rents in many non-Muslim societies can be associated with transitions to democracy that do not incite civil war. Substantively, the findings
Two points are worth clarifying for this assertion. First, there are a few modern societies that were exposed to Muslim conquest that are not Muslim-majority today, such as Spain and Portugal. These countries are not in the “groups” of non-Muslim societies examined in this chapter (i.e., Eastern Europe and Latin America). Second, the absence of Muslim conquest and its subsequent institutional trajectory in non-Muslim societies does not necessarily explain the institutional trajectory among the non-Muslim societies examined in this chapter. For example, many Latin American countries exhibit authoritarian political structures in the contemporary period that may be linked to different historical processes, such as their rates of settler mortality and/or initial factor endowments at the time of European colonial settlement (Acemoglu et al. , Engerman and Sokoloff ).
https://doi.org/10.1017/9781009367509.011 Published online by Cambridge University Press
Democratic Transitions in Non-Muslim Societies
in this chapter corroborate corollary B from Chapter : A decline in rents can facilitate a peaceful transition from dictatorship to democracy in societies whose preexisting institutions are more cohesive. I test this conjecture by probing the effects of various types of foreign transfers – particularly those that can accrue to a government’s coffers – on patterns of democracy and dictatorship. As I describe in next section, while not identical to foreign aid, these foreign transfers exhibit characteristics that are like aid inflows. By probing several plausible drivers of democratic transitions, this chapter contributes to an expansive literature in comparative democratization. This literature consistently finds that richer countries with a high quality of civil society (e.g., high levels of human capital, political organization, civil liberties) and less ethnic, religious, and income inequality/heterogeneity tend to be democratic, are more likely to democratize, and are also less prone to civil conflict (e.g., Blattman and Miguel , Dahl , Fish and Brooks , Przeworski et al. , Putnam ). Good economic performance can help consolidate democratic governance in nascent democracies, while economic crises can be a source of democratization as well as democratic backsliding (Svolik ). In general, these factors capture domestic level characteristics. This chapter does not refute these findings but is careful to account for them in the empirical analysis. For example, the cross-national statistical tests control for country-level income, measures of societal fragmentation, and timeinvariant characteristics (e.g., colonial legacy) with country fixed effects. Rather, this chapter’s qualitative and quantitative evidence draws attention to the substantive role that foreign transfers and their interplay with preexisting cohesive institutions may have played in many democratic transitions in the late twentieth century (i.e., the so-called Third Wave of democratization). I substantiate this assertion with evidence from case studies and corroborative cross-national quantitative analysis. Beginning with the case studies, I describe how an increase in petrodollars in Latin America and Soviet trade subsidies in Eastern Europe helped sustain dictatorship in these regions in the s and early s. However, when these transfers declined after , many of these societies were able to peacefully transition to democracy. I then argue that countries in Latin America and Eastern Europe exhibit preexisting institutions that are more cohesive compared to those in Muslim conquest societies and consider other possible explanations for the difference in institutional cohesion, such as precolonial state capacity and ethnic fragmentation. Taken together, these two empirical patterns comprise substantively important historical episodes of democratization in Latin America and Eastern Europe that are consistent with proposition and corollary B from Chapter . Section . provides statistical corroboration for this narrative and leverages a research
For a recent review of the literature, see Haggard and Kaufman ().
https://doi.org/10.1017/9781009367509.011 Published online by Cambridge University Press
How Rents Affect Dictatorship and Violence
design akin to a difference-in-differences identification strategy to provide a causal interpretation. In Section ., I then broaden my scope of inquiry beyond Latin America and Eastern Europe and present statistical evidence from a global sample of transfer recipients. Cognizant of endogeneity concerns, I embed an instrumental variables research design in a difference-in-differences approach to better gauge causal effects. The results from this analysis show that during a period associated with an exogenous decline in foreign transfers, societies with preexisting institutions that are less cohesive experienced greater political violence (dictatorship, civil war). Together, the evidence in Sections . to . demonstrates that () greater inflows of foreign transfers can entrench dictatorship but () their subsequent decline can foster movements to democracy, particularly in societies whose preexisting institutions are more cohesive.
. In this section, I describe and trace the effect of two types of foreign transfers on patterns of democracy and dictatorship in societies with preexisting institutions that are more cohesive, at least relative to Muslim conquest states. I focus on the inflows of petrodollar investments in Latin America and the supply of trade subsidies from the Soviet Union to its “satellites” in Eastern Europe. Like foreign aid in the previous chapter, these foreign transfers share three important features: () a significant fraction accrued to the government, () they, therefore, comprise rents (i.e., they are nontax government income), and () were highly correlated with oil prices. The magnitudes of these capital flows were quite substantial: Gulf aid comprised some . percent of total (global) aid between and , and much of it went to Muslim recipient countries (Neumayer , ). In Latin America, petrodollar borrowing amounted to over $. billion per annum in the late s, which was equivalent to percent of combined exports over the same period (Devlin , and ). In a less appreciated case, the Soviet Union transferred somewhere between $ and $ billion in subsidies (measured in dollars) to Eastern Europe (Liberman , ). In analyzing these cases, my narrative mimics the research design from the previous chapter in two important ways. First, I leverage a similar common external shock – that is, variation in world oil prices – to track temporal
As I show in the subsequent quantitative sections, the exogeneity of oil prices allows me to draw causal inferences. My rough calculations for the total transfers to the non-oil Muslim countries and Latin American borrowers are that they are of a similar order of magnitude. In contrast, total global GDP for lowand middle-income countries in , according to World Bank data, was around $ trillion; thus each type of transfer, though spread out over a decade, was on the order of . percent of developing-country GDP.
https://doi.org/10.1017/9781009367509.011 Published online by Cambridge University Press
Democratic Transitions in Non-Muslim Societies
movements in foreign transfers. Second, to purge the direct effects of oil on politics, I limit my analysis to non-oil-producing transfer recipients in Latin America (e.g., Argentina but not Brazil) and Eastern Europe (e.g., Poland). Third, the exogeneity of oil prices allows me to draw a causal interpretation for the impact of foreign transfers on patterns of dictatorship and democracy across Latin American and Eastern European non-oil-producing transfer recipients. A Petrodollars in Latin America From the mid-s until the early s, governments in Latin America – many of whom were dictatorships – were awash in cheap finance, which they used to maintain political control. Once their source of borrowing dried up, or they over-borrowed into self-perpetuated default, many governments found they did not have enough legitimate support to remain in power. In some countries, power changed hands; in many others, the regime changed, which, according to Karen Remmer (), led to the longest and deepest wave of democratization in Latin America’s history. Oil producers in the Persian Gulf did not seek to cause regime change halfway around the world, but they could not domestically absorb all the rents of the high oil prices in the s, and some of those rents ended up being made available to Latin American dictators through the international banking system. The oil exporters were experiencing huge increases in their export revenues, and they had three choices to balance their international payments: increase imports, increase foreign aid (which they did through transfers to Muslim recipients – see Chapter ), or increase investments abroad. They did all three of each in large numbers (Gerner ). Regarding the third option, Gulf oil producers parked their investments abroad, often in Eurodollar accounts, which created a surge of lending to developing countries (Riner , ). Latin American countries were the main beneficiaries of these petrodollars (Devlin , ). Though the transfers came in the form of loans to economic actors with limited time horizons, it sometimes felt like a gift akin to an economic rent. For instance, one Chilean banker put it well: “We look upon foreign loans as a bonus from the rest of the world to Chile” (quoted in Riner , ). In practice, most of the funds went through the private banking system and allowed governments to borrow heavily. In the early s, percent of external finance to Latin America was through official channels. By the late s, that figure had fallen to percent; percent of external finance came through private banks (Devlin , ). The lending to developing countries was boosted by the recession in the advanced countries, which reduced the appetites of the private banks’ traditional borrowers for loans at just the time they were awash with liquidity (Riner , ). This led to low interest spreads in the OECD area, while Latin America was posting strong economic
https://doi.org/10.1017/9781009367509.011 Published online by Cambridge University Press
60 40 0
0
20
Oil price (2009 US$)
Commercial loans (% GDP) 2 4 6
80
8
100
How Rents Affect Dictatorship and Violence
1970
1975
1980
1985
1990
1995
2000
2005
2010
Year Loans FIGURE
Oil price (2009 US$)
. Commercial loans and oil prices in Latin American non-oil producers
growth, and its governments were eager to borrow (Devlin , ). It was the perfect combination of supply and demand. These financial flows were substantial. For Latin American non-oil exporters (excluding Brazil, Ecuador, Mexico, and Venezuela), the annual net external transfer went from less than $. billion in the s to average $. billion in the first half of the s and $. billion in the second half. The net inflows would remain at $. billion in before the Mexican debt crisis of would cause a reversal; by the net outflow from these countries was over $ billion (Devlin , ). For a sense of magnitude, the net transfer in the s was equivalent to percent of the countries’ combined exports over the same period (Devlin , ). If one adds up the net transfers in just the non-oil exporters – after all, the oil exporters were getting a firstorder surge in rents from the increase in oil prices – the total amounts to around $ billion. Annual flows of petrodollars tended to co-vary with recent movements in oil prices. Figure . plots the average amount of commercial loans (as a share of GDP) across Latin American non-oil producers and oil prices. The figure shows that commercial borrowing tracks the price of oil fairly well through the end of the twentieth century. While the timing of the start of the financing was clearly correlated with the oil price increases, for each country the timing of the end of the easy financing was more driven by internal dynamics of borrowing and
https://doi.org/10.1017/9781009367509.011 Published online by Cambridge University Press
Democratic Transitions in Non-Muslim Societies
over-extension, or the Mexican debt crisis which put an end to affordable borrowing for the region as a whole. There was certainly new borrowing after , but most of it was “involuntary,” or loans that resulted from rolling over or rescheduling previous debt (Devlin , ). For most countries, the credit cycle tended to follow a similar pattern. First, loans generated net financial inflows. With the new inflows, the borrower’s appetite for new loans increased, to finance both citizens’ expectations of continued economic growth, as well as the cost of servicing existing debt. With higher debt levels, the borrower’s creditworthiness started to decrease and the borrower found it had to pay more for new finance (in diminishing quantities) and to just service and roll over existing debt. Eventually, the transfer became negative as nervous creditors demanded still higher interest rates and only lent to enable the borrower to meet its existing obligations. Some countries, like Peru, started the cycle as early as (Devlin , ) and others, like politically unstable Argentina, did not start until (Frieden , ). Unlike loans from official agencies, the lending from private banks came with very few strings attached, enabling the governments to continue their policies without outside intervention. Notably, loans were made unconditionally, without technical or political requirements (Devlin , ). For example, according to Robert Devlin (, ) in Peru, only percent of loans were even tied to actual projects or capital goods imports; in Bolivia, the figure was percent. Deborah Riner (, ) argues that access to Eurocurrency market increased the autonomy of developing-country governments. One Chilean official she interviewed expressed the sentiment: For a government “. . . whose attitude is that no one is going to tell us what to do,” borrowing from foreign private banks is more attractive than borrowing from the World Bank or IDB “. . . who can refuse us loans because the US and Venezuela don’t like our politics” (Riner , ). Since many regimes in the continent were still under military control, this autonomy served to consolidate the generals’ power through patronage channels and control (e.g., Pinochet’s repression in Chile, Velasco’s military rule in Peru). Of course, when the financial tides turned with the crisis, so did the ability of many governments to maintain their power, as Jeffry Frieden () meticulously describes. The loans allowed them to expand benefits to constituents when the net financial flows were positive, but once the banks were calling in their loans, “debtor governments had to take resources from society rather than distribute them. Governments inevitably failed to meet all of the social demands they faced during the crisis” (Frieden , ). In some countries, governments changed hands while in others there were demands to change the system itself, which in many cases led to democratization (Frieden , ). The variation in commercial borrowing from a relatively high level in the s to a lower level starting in the mid-s coincides with the region’s entrenchment of authoritarianism (in the s) and its transition to
https://doi.org/10.1017/9781009367509.011 Published online by Cambridge University Press
0
0.2
0.4 0.6 0.8 POLITY (normalized)
Commercial loans (% GDP) 2 4 6
1
8
How Rents Affect Dictatorship and Violence
1970
1975
1980
1985
Loans FIGURE
1990 Year
1995
2000
2005
2010
POLITY (normalized)
. Commercial loans and democracy in Latin American non-oil producers
democracy (from early s onward). Figure . illustrates this by plotting the average level of democracy (using the normalized POLITY score) along with the average level of commercial loans among Latin American non-oil producers. In the s, the typical Latin American country was nondemocratic with an average POLITY score of .. In the early to mid-s, many countries became increasingly less authoritarian and by the average score had increased to .. In combination, Figures . and . suggest that higher levels of commercial borrowing (driven in large part by petrodollar investments) sustained authoritarianism in Latin America in the s. As commercial borrowing fell in the s (due to declining petrodollar transfers), governments found it increasingly difficult to maintain their authoritarian rule and these countries democratized. Several cases illustrate this dynamic. For example, Peru managed to become heavily indebted, lose access to new borrowing, and even democratize well before the debt crisis in Mexico. Running budget deficits financed by foreign finance, the military government of Juan Velasco Alvarado sought to maintain legitimacy “with its two main constituencies: the generals and the popular masses” (Devlin , ). According to a former central banker, half of the $ billion borrowed in the late s was used for arms purchases and a fifth for food subsidies (Riner , ). One former official in the Velasco government described the mechanism:
https://doi.org/10.1017/9781009367509.011 Published online by Cambridge University Press
Democratic Transitions in Non-Muslim Societies
Foreign borrowing was the way Velasco maintained his position. Ministries were the fiefdoms of the generals who headed them . . . There was a lot of borrowing for corruption’s sake: The generals wanted their kickbacks . . . By giving the generals free rein, Velasco retained his position. (Quoted in Riner , )
The Americans were not happy with the policies pursued under Velasco because they involved nationalizations of foreign investment. As a result, they stalled the availability of development finance. Peru, however, could borrow on the Eurodollar market and maintain its stance (Riner , –). After rising imports and shocks to copper prices and anchovy stocks, Peru began to lose control of its economy in the late s; with it came a gradual erosion of political control by the military (Riner , –, and –). During the s, Peru successfully (and peacefully) consolidated democratic rule. Argentina and Chile also went through democratizations in the s but the relation with the credit cycle was not quite as clean. In Argentina, a hardline government took the political opening brought about by the economic decline, so the country in fact moved further to the right in the wake of economic crisis. It took military defeat in the Falklands to finally create the space for democratization in (Frieden , ). In Chile, where the Pinochet government had not borrowed that much during the credit spree, the credit cycle was not as devastating to government even as the economy contracted. Meanwhile the business community was sufficiently afraid of the left that there was little defection to the opposition. As a result, when Chile finally democratized in it was during a period of economic strength (Frieden , –).
B Trade Subsidies in Eastern Europe Even though they were often operating without market prices, Soviet “satellites” in Eastern Europe experienced the oil price rise just like the rest of the world. The Soviet Union exported oil and gas to many Western countries as well as its allies in Eastern Europe. While oil was traded with market economies according to market pricing, trade terms among Eastern Bloc countries were determined by an agreement brokered by the Council for Mutual Economic Assistance (CMEA, also known as COMECON, an economic organization founded by the Soviet Union) (Marer ). Under the Bucharest Formula of , prices in each half-decade were determined by the average world market price of the prior half-decade. However, in early , the formula was modified to be a five-year moving average of world oil prices. At first, the fixing of oil prices had little effect on a subsidy to Eastern Europe since oil prices only fluctuated mildly during this period; for example, the world market price for a barrel of oil (in dollars) was just less than $ in , pretty much the same value as it was in . But when real oil prices rose dramatically to $ in , the CMEA would have been pricing oil at under $, generating a subsidy of over $ per barrel of oil for that year. Following the price formula
https://doi.org/10.1017/9781009367509.011 Published online by Cambridge University Press
How Rents Affect Dictatorship and Violence
would lead to nearly a decade of heavy subsidies before a falling oil price reversed them. Why did the Soviet Union decide to subsidize its Eastern European allies, principally the “CMEA-” of East Germany, Czechoslovakia, Poland, Bulgaria, Hungary, and Romania? There is agreement among scholars that there was likely intentionality on the part of the Soviets to maintain the subsidies to secure state interests. For one, the fixed price contracts in the CMEA were difficult to amend, according to one scholar, for “institutional and political as well as economic reasons” (Marer , ). But difficult did not mean impossible, and the subsidies played an important role during the twilight of the Communist era. Two, they may have been used to reward military loyalty to the Soviet Union, for instance, since they were correlated with defense spending rates across countries (Liberman ). More broadly, a study authored by the RAND Corporation argued that the subsidies were continued by the Soviet Union “to buttress its important strategic, ideological, bureaucratic, and political stakes in the region” and “forestall domestic unrest in Eastern Europe” (Crane , ). The trade subsidies appear to have played an important role in propping up the regimes in the recipient countries. Their totalitarian regimes were not cheap to run, requiring a police state in which there were four times as many police and internal security officers, for example, in East versus West Germany (Liberman , ). Moreover, changes in economic variables such as price cuts or wages were particularly salient “triggering” events for political collective action in Eastern Europe given the inability to otherwise organize (Liberman , ). This meant that both economic decline as well as increased prices for key consumer goods would have been seen by governments as things to avoid in order to maintain political stability. Subsidies from the USSR appeared to have been crucial in keeping prices stable as well as preventing wage cuts from economic recession or budgetary cuts, at least for a while. The supply of Soviet trade subsidies – proxied using export data – was highly correlated with oil prices. Figure . plots the average level of goods exports (as a share of GDP) across Eastern European Soviet satellites and the COMECON oil price (i.e., a five-year moving average of the world price of oil). While trade data prior to is unavailable, the figure illustrates a strong positive correlation: A higher COMECON oil price is associated with greater exports. These subsidies arrived at a time of increasing fiscal strain for the Eastern European economies. Growth rates were slowing and hard currency debt was increasing, as governments increasingly looked to new ways to fund the regime. Net hard currency debt to Western banks and institutions of the CMEA- in was only $ billion in , but by it had reached $ billion and by $ billion (Wharton Econometrics ). By the late s, Western creditors expected debt service payments back from the Eastern European governments rather than just refinancing and providing new loans (Marer ). Creditors had legitimate worries about potential default, and the
https://doi.org/10.1017/9781009367509.011 Published online by Cambridge University Press
80 20
40
60
Good exports (% GDP) 30 40
0
20 1970
1975
1980
1985
Goods exports FIGURE
COMECON oil price (US$ 2009)
50
100
Democratic Transitions in Non-Muslim Societies
1990 Year
1995
2000
2005
2010
COMECON oil price (US$ 2009)
. Trade subsidies and oil prices in Eastern European Soviet satellites
Eastern European recession of – vindicated those fears (Crane ). Without the Soviet subsidies, Eastern European governments would have gotten into trouble sooner than they did. Of course, by the end of the decade, the former CMEA- would democratize; there are many explanations for why these Communist regimes of Eastern Europe fell in . These explanations include the roles of ideas; of Gorbachev, Reagan, or Pope John Paul; of domestic opposition movements; and of the relative economic inefficiency of communism (see Brown , Okey ). Writing toward the end of the period of subsidies, Samuel Huntington (, ) rated the prospect of democratization as “virtually nil” because of the threat of the Soviet veto. Adam Przeworski (, ) wrote (unlike Huntington, with the benefit of hindsight) that the threat of Soviet invasion was indeed a constraint on internal political change in Eastern Europe: “But it was only that: a constraint, a dam placed against pressing water.” As oil prices halved from to , the subsidies began to reverse. While Eastern European economies continued to grow in the late s, their incomes
Eastern European governments may have received an additional petrodollar tailwind from the availability of private lending facilitated by the glut of savings caused by Middle Eastern oil rents being recycled through the global financial system. This extended the centrally planned “investment cycle” well beyond its usual several-year tenure as ambitious plans normally came up against balance-of-payments realities (Crane , ).
https://doi.org/10.1017/9781009367509.011 Published online by Cambridge University Press
0.6 0.4 0
20
0.2
POLITY (normalized)
Good exports (% GDP) 30 40
0.8
1
50
How Rents Affect Dictatorship and Violence
1970
1975
1980
1985
Goods exports FIGURE
1990 Year
1995
2000
2005
2010
POLITY (normalized)
. Trade subsidies and democracy in Eastern European Soviet satellites
were diverging from their Western counterparts. The end of a monumental transfer of unearned income undoubtedly reduced those governments’ legitimacy and funds to govern in the same heavy-handed way. Figure . shows that as oil prices fell in the mid-s – driving a sharp reduction in COMECON subsidies and trade – these satellites began their transitions toward democracy. By the late s, as the Soviet Union disintegrated, these Eastern European countries were firmly on the path toward democracy. Russian reformer Yegor Gaidar () dates the beginning of the end of the USSR to September , , the day the Saudis decided to flood oil markets. According to Gaidar, the Soviet Union was able to borrow its way out of its funding dilemma caused by plummeting export earnings – cut military expenditure, or food imports? – for several years; by , needing $ billion in loans from the West, it could no longer intervene militarily in Eastern European affairs so the Soviet veto over Eastern European democratization was effectively removed. A few more recent analyses take the oil price explanation seriously for the fall of the USSR, perhaps inspired by the obvious role of oil in Russia’s recent rise. Douglas Reynolds and Marek Kolodziej () argue that a decline in oil production in the USSR led to a fall in Soviet GDP rather than vice versa. Stephen Kotkin () notes that “[o]il seemed to save the Soviet Union in the s, but it merely delayed the inevitable” by encouraging the “delay of painful economic restructuring” (p. , ).
https://doi.org/10.1017/9781009367509.011 Published online by Cambridge University Press
Democratic Transitions in Non-Muslim Societies
Mikhail Gorbachev’s response to the weakening state budgets was to initiate perestroika, or restructuring, and his way to push it through the Communist Party was glasnost, or opening. Thus a fall in unearned income in the Soviet Union led to a fall in unearned income transferred to Eastern Europe; at the same time, it took away the Soviet veto from Eastern European democratization from within, and it planted the seeds for challengers to the regimes in both Eastern Europe and the USSR itself to have a fighting chance. Within a few years, they had taken the challenge and peacefully removed the incumbents. C Comparison of Preexisting Cohesive Institutions The evidence in the previous two sub-sections describes how buoyant amounts of external transfers funded dictatorship in Latin America and Eastern Europe, but their subsequent decline was correlated with relatively peaceful transitions to democracy in these regions. These dynamics contrast sharply with those in the previous chapter, where a decline in transfers – in the form of foreign aid – heightened civil war and did not usher in democracy. But why were Muslim countries more prone to conflict when their transfers declined? The formal model in Chapter provides conditions for such dynamics: Civil war is more likely in countries with a lower exogenous (or “fixed” level) of θ, which corresponds to preexisting institutions that are less cohesive. To date, scholars have identified two important sources of fixed θ across countries. The first recognizes that “deep” historical processes often associated with critical junctures explain contemporary political institutions (see Nunn for an overview). Jacob Hariri (, ), for example, argues that countries with “longer” state histories tend to exhibit less cohesive political institutions today. The second, and often related, source of fixed θ links it to the stickiness of societal heterogeneity associated with ethnic, linguistic, and/or religious fractionalization. Here, societies with greater social fragmentation tend to exhibit less cohesive political institutions (e.g., Alesina et al. , Weingast ). In Part II of this book, I examined the first source and demonstrated how the institutional legacy of Muslim conquest is an important source of less cohesive institutions in many contemporary Muslim societies (i.e., persistence of dictatorship). Following the initial period of Muslim conquest, conquered territories were governed with two institutions – the use of elite slave soldiers (mamluks) and their compensation through a non-hereditary form tax-collection (iqta) – that effectively centralized political authority in dictatorship. This trajectory of less cohesive institutions persisted through to the modern period. Thus, when foreign aid disbursements (external transfers) in non-oil-producing Muslim countries fell in the mid-s, these societies were more predisposed to experience civil war and not transition to democracy.
Recall, θ measures the share of state resources (e.g., rents) the incumbent distributes to out of power groups. A lower value of θ implies the incumbent distributes less to these groups.
https://doi.org/10.1017/9781009367509.011 Published online by Cambridge University Press
How Rents Affect Dictatorship and Violence
In contrast, countries in Latin America and Eastern Europe did not experience Muslim conquest and thus were not exposed to mamluk institutions and the subsequent trajectory of centralized autocracy. Empirically, as the results in Chapter demonstrate, institutions in these two regions were more cohesive (relative to Muslim societies) at the time of political change; thus it allowed for the possibility of transition to democracy with a lower level of external transfers. It is worth emphasizing that my analysis of institutional formation in Part II of this book (contrasting conquest to non-conquest societies) does not necessarily explain the institutional trajectory in non-Muslim societies. Other historical processes might account for the unique institutional trajectories in these regions. For example, in the contemporary period, many societies in Latin America exhibit less cohesive institutions (e.g., authoritarian political structures) that may be linked to their rates of settler mortality and/or initial factor endowments at the time of European colonial settlement (Acemoglu et al. , Engermann and Sokoloff ). That said, for the purposes of my comparative analysis in this chapter, a robust empirical regularity is present: In the contemporary period, Muslim societies have tended to exhibit less institutional cohesion relative to societies in Latin America and Eastern Europe. For example, unlike Muslim countries, many Latin American ones had prior experiences with democracy, making a transition back to democracy more likely (under opportune circumstances, e.g., lower external transfers). An examination of Gerring et al.’s () measure of accumulated democratic experience (“democratic stock”) – the main dependent variable in Section . – shows that Latin American countries had longer experiences with democracy prior to the s than Muslim countries. Furthermore, several Latin American dictatorships in the s were the result of coups that overthrew democratically elected governments, such as in Brazil (), Peru (), and Chile (), and were often run under the premise of being a caretaker government. While the former Soviet satellite countries had not previously experienced democracy for sustained periods, they shared several societal characteristics amenable to democracy (Huntington ): the existence of strong and effective civic organizations, limited ethnic pluralism and – especially relevant to Communist Eastern Europe – relatively low levels of inequality. These conditions correspond neatly to a higher value of θ in Chapter ’s formal model. As the exception that proves the role, in the former Yugoslavia, with much more intense ethnic pluralism, the transition from Communist rule was incredibly violent and only achieved peaceful democracies after several partitions and mass migration reduced the level of ethnic pluralism. Analysis of various quantitative indicators corroborates my characterization of θ above. Table . reports the average values of societal fragmentation and “state history” across Muslim and non-Muslim non-oil-producing countries. To measure state history, I use a country’s precolonial state history and “timing” since the Neolithic Revolution (both from Hariri ). For each, a higher value corresponds to a longer state history. I use Philip Roeder’s ()
https://doi.org/10.1017/9781009367509.011 Published online by Cambridge University Press
Democratic Transitions in Non-Muslim Societies
. Correlates of cohesive institutions in Muslim and non-Muslim non-oilproducing societies
TABLE
Muslim Non-Muslim
Neolithic transition
Precolonial state history
ELF in
()
()
()
. (.) . (.)
. (.) . (.)
. (.) . (.)
Note: Group means with corresponding standard deviations reported in parentheses. In column , ELF refers to ethnolinguistic fractionalization.
ethnolinguistic fractionalization (ELF) index to measure each country’s societal fragmentation. This index ranges from to , where a higher value implies greater social heterogeneity. Columns and show that Muslim countries, on average, have longer state histories than non-Muslim states. According to Jacob Hariri’s (, ) finding this implies that Muslim non-oil producers will tend to exhibit less cohesive institutions in the modern era. With respect to societal heterogeneity, column shows that Muslim countries tend to exhibit greater ethnolinguistic fragmentation. According to Alesina et al. (), incumbents in Muslim societies will tend to share less of the state’s resources (rents) with out of power groups. In short, on two important dimensions of cohesive institutions, Muslim countries tend to have a low value of θ. (In Section ., I provide additional evidence using “exogenous” institutions.)
. In this section, I present statistical evidence that seeks to corroborate the qualitative evidence in Section . and contrast it with patterns of political violence in Muslim transfer (aid) recipients. Specifically, I investigate the effect of external transfers on democracy and conflict at the country-year level using regression analysis. This allows me to control for confounding effects, such as country income, geopolitics, and time-invariant country characteristics (e.g., geography, historical legacies) with country fixed effects. Using a difference-indifferences setup (see Equation .) allows me to plausibly infer a causal relationship: External transfers in countries cause differential patterns of democracy and conflict. Throughout the analysis, I limit my sample to non-oilproducing recipients to purge concerns that direct oil-price induced government income (e.g., nationalization of oil industries, tax income from foreign oil companies) may unduly affect governance (e.g., Mahdavi ).
For instance, this means I exclude Brazil, Mexico, and Venezuela from the main sample of Latin American countries as these nations produce significant amounts of oil. I also exclude oil-
https://doi.org/10.1017/9781009367509.011 Published online by Cambridge University Press
How Rents Affect Dictatorship and Violence
A Statistical Associations To corroborate the patterns in Figures .–., I first examine the statistical association between each foreign transfer and democracy with the regression below. For each group (i.e., Latin America, Eastern Europe, Muslim) of nonoil-producing countries (denoted with i) in year t, I regress a country’s normalized POLITY score on its relevant financial transfer ðTRANSFERit Þ; log per capita income, GDPPCit (to account for ‘modernization’ effects); a post–Cold War dummy, POSTCW t (to account for the upward trend in democracy after the Cold War); and country fixed effects ðδi Þ. Including fixed effects imply the coefficients estimate within-country effects. If higher levels of external transfers are associated with less democracy, I expect β < 0. POLITY it ¼ α þ β TRANSFERit þ π GDPPCit þ κ POSTCW t þ δi þ εit ð8:1Þ The results, reported in Table ., show that higher levels of each financial transfer are negatively and robustly correlated with democracy. In column , higher levels of foreign aid lower the normalized POLITY score in non-oilproducing Muslim recipients, while higher amounts of commercial loans (column ) and exports (column ) hurt democracy in Latin American and Soviet satellite countries, respectively. Across the specifications, the controls have their expected effects: Richer countries tend to be more democratic (although, not on average, in Muslim-majority societies), while the post–Cold War period exhibits a robust positive effect on democracy.
B Difference-in-Differences Estimates The results in Table . show that greater levels of unearned transfers are negatively correlated with democracy in Latin American, Eastern European (Soviet satellites), and Muslim non-oil-producing recipients. This finding is consistent with proposition from Chapter . This, however, does not necessarily imply that a reduction in transfers contributes to political change. To evaluate this, I examine how changes in the levels of transfers from the period of high oil prices (–) to the period of low oil prices (–) affected democracy and civil war in Latin America, Soviet satellites, and Muslim transfer recipients. In order to better infer a causal relationship, I leverage the plausibly exogenous change (drop) in world oil prices in the mid-s in a differences-indifference (DD) research design. Specifically, I estimate the following baseline specification with exogenous covariates: producing Muslim-majority countries such as Saudi Arabia and Iran, as well as aid recipients that produce some oil including Egypt.
https://doi.org/10.1017/9781009367509.011 Published online by Cambridge University Press
Democratic Transitions in Non-Muslim Societies TABLE
. Foreign transfers are associated with less democracy
Dependent variable
Democracy (normalized)
Sample: Foreign aid
()
()
()
Muslim −. (.)**
Latin America
Soviet Satellites
−. (.)***
Commercial loans
. (.)*** . (.)*** . (.)***
. (.) . (.)*** −. (.)
−. (.)** . (.)** . (.)*** −. (.)**
Yes .
Yes .
Yes .
Goods exports Log GDP per capita ( US$) Post–Cold War Constant Country fixed effects No. observations R-squared
Notes: Sample of non-oil-producing transfer recipients. Estimation via OLS. Robust standard errors, clustered by country in parentheses. **, *** = significant at and percent respectively. All the specifications include country fixed effects. These coefficients are not reported.
PV ijt ¼ α þ
X
X β R POST θ R þ δPOST t þ Ci þ μijt þ j t j j¼k j¼κ j j ð8:2Þ
where PV ijt measures civil war or democracy (depending on the specification) in country i in region j in year t. There are three regions in the set κ: Latin America, Soviet satellites, and Muslim countries. Rj is a dummy variable equal to if country i is in region j and otherwise. POST t is a dummy variable equal to in the period of low oil prices (–) and otherwise. Ci is a vector of country fixed effects and μijt is an error term. Since my aim is to identify causal effects, I do include potentially endogenous covariates, such as country income and growth. In Equation ., POST t is the exogenous “treatment” since it is assigned based on when world oil prices fell, which was plausibly exogenous to economic and political conditions in non-oil-producing transfer recipients. And by interacting POST t with Rj , I appropriately ‘assign’ the exogenous treatment to countries in the relevant regions.
For example, for Pakistan, Rmuslim ¼ 1, Rlatin ¼ 0, and Rsoviet ¼ 0. For instance, civil conflict can affect growth and country income.
https://doi.org/10.1017/9781009367509.011 Published online by Cambridge University Press
TABLE
How Rents Affect Dictatorship and Violence . An exogenous decline in transfers and political violence
Dependent variable:
Post high oil period (POST) Muslim POST Latin America POST Soviet Sat. POST Muslim Latin America Soviet Satellite Country fixed effects R-squared No. observations
Civil war
Democracy
()
()
()
()
−. (.) . (.)** −. (.) −. (.) −. (.) . (.) −. (.)***
−. (.) . (.)* −. (.) −. (.)
. (.)**** −. (.) . (.)*** . (.)*** −. (.) −. (.) −. (.)***
. (.)*** . (.) . (.)*** . (.)***
.
Yes .
.
Yes .
Notes: Sample of non-oil-producing transfer recipients. Estimation via OLS. Robust standard errors, clustered by country reported in brackets. *, **, *** = significant at , , and percent respectively. Muslim, Latin America, and Soviet Satellite are dummy variables for whether a country belongs to any of these “regions.” The excluded category is any country that does not belong to any of these regions. The controls and a constant are not reported.
In regression (.) the coefficients of interest are the β‘s, which evaluate the effect of transitioning from a period of high to low oil prices for countries in each region on the country’s level of democracy or civil war (depending on the specification). For example, if the dependent variable is the incidence of civil war, a positive coefficient on β implies that transfer recipients in that region were more likely to experience civil war during the period of low oil prices, relative to the earlier period of high oil prices. I expect β to be positive and statistically significant for Muslim recipients and to have no effect for recipients in Latin America and Eastern Europe. In contrast, with the normalized POLITY score as the dependent variable, I expect β to be positive and statistically significant for recipients in Latin America and Eastern Europe and to have no (or even a negative) effect in Muslim transfer recipients. Table . presents the DD estimates corresponding to Equation (.). Columns and show that Muslim recipients became percent more likely to experience civil conflict during the period of low oil prices (relative to the period of high oil prices). There was no statistically significant change in conflict in Latin American and Soviet satellite transfer recipients between the
https://doi.org/10.1017/9781009367509.011 Published online by Cambridge University Press
Democratic Transitions in Non-Muslim Societies
periods of high and low prices. Rather, these countries experienced statistically significant improvements in democracy after oil prices tanked (columns and ). The typical country experienced around a .-point increase in its normalized POLITY score (relative to the period of high oil price). Muslim recipients did not experience any meaningful change in their normalized POLITY scores. They remained nondemocratic but marred with heightened conflict.
. - Skeptics may worry that the case studies are neither causally attributable nor generalizable. To allay this concern, in this section I expand the analysis to a global sample of non-oil-producing transfer recipients and embed an instrumental variables (IV) strategy in a DD setup to better identify causal effects. The identification strategy seeks to gauge the causal impact associated with a decline in transfers on political violence across countries with different preexisting cohesive institutions. To do so, requires a plausibly exogenous source of variation in institutions, particularly in the mid-s (when transfers declined). My identification strategy is deployed in two steps, which I describe in greater detail below. A Identification Strategy Exogenous Cohesive Institutions In the first step, I leverage a country’s transition to the Neolithic Revolution associated with “early statehood” as an instrumental variable for cohesive institutions in the modern era (specifically, in ). In a cross-sectional study of democracy, Jacob Hariri (, ) argues that outside Western Europe, “the historical experience of an early development of statehood has been autocracy, not democracy.” According to Hariri (, ): This occurred because indigenous state institutions constrained the European colonial endeavor and limited the diffusion of institutions and ideas from Europe. Some countries were strong enough to resist colonization; others had enough state infrastructure that the colonial powers would rule to a considerable degree through existing institutions. Neither group experienced institutional transplantation nor European settlement. As a result, in societies with more precolonial state development it was more likely that the traditional mode of rule – autocracy – was preserved. In countries with less precolonial state development, the indigenous state was not an obstacle to European settlement and institutional transplantation. These countries were more likely therefore to embark on a democratic regime trajectory.
Hariri documents that “older states” with more well-developed indigenous state institutions constrained the European colonial endeavor and limited the diffusion of European institutions and ideas and their associated democratic trajectory. Thus, countries with a longer “state history” exhibit less cohesive institutions in the modern era.
https://doi.org/10.1017/9781009367509.011 Published online by Cambridge University Press
How Rents Affect Dictatorship and Violence
Yet, examining effect of early statehood on a country’s subsequent institutional trajectory is complicated by the endogeneity of statehood. For instance, the way a state is governed affects the probability that it will survive. Moreover, unobserved cultural and historical factors (e.g., the influence of important individuals, such as Napoleon or Stalin) might plausibly affect both a state’s history and its regime trajectory. To mitigate this concern, I follow Hariri and leverage the timing of the Neolithic Revolution (the historical transition from foraging to farming) as an exogenous source of variation in precolonial state development. The underlying rationale is based on the idea that an early transition from hunting-gathering to settled agriculture predicts an early development of state institutions. Thus, Hariri’s historical narrative and corroborative empirical analysis implies that NEOLITHIC ! STATE HISTORY ! COHESIVE INSTITUTIONS. Since NEOLITHIC is causally prior to modern institutions, econometrically, one can express these relationships in a two-stage least squares (SLS) setup of the form: First stage : STATEHISTORY i ¼ a þ b NEOLITHICi þ c Xi þ ui ð8:3aÞ Second stage : INSTITUTIONSi ¼ α þ β STATEHISTORY i þ δ Xi þ ei ð8:3bÞ In the first-stage regression (.a), I regress the length of country i’s state history on its transition to the Neolithic Revolution and a vector of confounders (e.g., per capita income, population). I expect countries that transitioned to the Neolithic Revolution at an earlier point in history (i.e., a higher value of NEOLITHIC) to have a longer state history. Thus, I expect b > 0. In the second stage regression (.b), I use predicted values from the first stage for STATE HISTORY to explain the cross-sectional variation in cohesive institutions in the modern era, specifically a country’s normalized POLITY score in . From Jacob Hariri’s () historical narrative, a longer state history should be associated with less cohesive (democratic) institutions in ; thus, I expect β < 0. Exogenous Decline in Transfers In the second step of the identification strategy, I embed this exogenous source of variation for state history (NEOLITHIC) in a DD framework like the one described in the previous section. Recall, here I leverage the plausibly exogenous drop in world oil prices in to “explain” the reduction in foreign transfers across countries. I pick as this corresponded to the large annual price decline in world oil prices. Thus, with these two exogenous covariates
The results are robust with alternate cutoff years.
https://doi.org/10.1017/9781009367509.011 Published online by Cambridge University Press
Democratic Transitions in Non-Muslim Societies
(NEOLITHICi, POSTt), I estimate the following DD specification with exogenous covariates to draw causal inferences. Reduced form : PV it ¼ α þ βðNEOLITHICi POST t Þ þ δPOST t þ Ci þ εit
ð8:4Þ d Instrumented form : PV it ¼ ϕ þ τSTATE HISTORY ι POST t þ πPOST þ Ci þ νit
ð8:5Þ where PV it is an increasing three-point index of political violence ð2 ¼ civil war, 1 ¼ repression, 0 ¼ peaceÞ based on Besley and Persson’s (b) categorization. A country is assigned a value of if it experiences civil conflict with at least , battle deaths per annum, if the government engages in any purge (which could be in the presence of conflict not exceeding , battle deaths), and otherwise. As a consequence, PV it directly measures the step-wise concept of political violence in Chapter ’s formal model. POST t is a dummy variable equal to in the period of low oil prices (–) and zero otherwise. In Equations . and ., POST t is the treatment since it is assigned based on when world oil prices fell, which was plausibly exogenous to economic and political conditions in non-oil-producing transfer recipients. Ci is a vector of country fixed effects (which subsume the main effects associated with NEOLITHICi and STATE HISTORY i since they are each time invariant) and εit and νit are error terms. Thus, the DD specification exploits two plausibly exogenous sources of variation: across countries with NEOLITHICi and over time due to a sharp drop in oil prices (with POST t ). In interpreting the coefficients, a higher value of NEOLITHICi and correspondingly, STATE d HISTORY ι (in Equation .) implies less cohesive institutions. The coefficients of interest are the interaction HISTORY ι where a of POST t with NEOLITHICi and POST t with STATE d positive coefficient on the interaction terms ðβ, τÞ implies that countries with less cohesive institutions experienced higher levels of political violence during the period of low oil prices (i.e., POST t ¼ 1). It corresponds to testing proposition A from Chapter . Since the dependent variable is ordinal, I use ordered probit to estimate the regressions. I use a two-step procedure based on Newey () to derive the instrumented estimates in Equation (.). Specifically, I estimate the
This approach is inspired by the “reduced form” approach in Besley and Persson (b, –). Moreover, since my aim is to identify causal effects, I do not include potentially endogenous covariates, such as country income and growth. Reassuringly, this ordinal variable is negatively correlated with a country’s level of democracy. Note, I use democracy (normalized POLITY) and civil war separately as the dependent variables in Section . since my aim is to statistically model the trends in Figures .–., which plot democracy and conflict separately. Since the dependent variable is ordinal, SLS is not appropriate. Rather a non-linear estimator is required, such as ordered probit.
https://doi.org/10.1017/9781009367509.011 Published online by Cambridge University Press
1
How Rents Affect Dictatorship and Violence
MUS JAM
CRI
Normalized POLITY 0.4 0.6 0.8
URY BOL FJI
BWADOM
GMB SLV HND
LKA
ZAF
ZWE UGA SENGTM NIC
SDN
LBN NPL
0.2
PAK
0
MDG PAN TZA LBR CHL CUB COM LSO RWA TCD BFA CAF MLI BEN TGO GIN SLE BDIGUY GHA KEN MRT SOM NER MOZ GNBMARPRY DJI ETH HTI MWI ZMB CIV SWZ
0
2
PHL MNGBGDPOL LAO MMR
HUN BGR ALB
JOR
BTN
4 6 Time since Neolithic Revolution Normalized POLITY
FIGURE
TUR
8
10
Fitted values
. Timing since the Neolithic Revolution and democracy in
first-stage regression using OLS and substitute the predicted values into the second stage regression, which I estimate via ordered probit. B Cohesive Institutions in The validity of the two-step identification strategy necessitates establishing the first step. As an initial confirmation, the “fitted line” in Figure . shows that countries that transitioned to settled agriculture (i.e., exhibit a higher value of NEOLITHIC) tend to exhibit less cohesive (democratic) institutions in . Table . provides more stringent validation. It shows that NEOLITHICi is robust and plausibly exogenous to precolonial state history and subsequent institutions in . In column , I first show that a country’s “stock” of democratic experience (up to ) is positively associated with a country’s normalized POLITY score in . Column shows that countries with a longer precolonial state history (drawn from Hariri ) tend to have accumulated less democratic experience. Together, the associations in columns and suggest that precolonial state history affects a country’s accumulated democratic experience, and in turn, its quality of democratic institutions in .
Observe, this is my preferred measure of contemporary democracy used in Chapter .
https://doi.org/10.1017/9781009367509.011 Published online by Cambridge University Press
https://doi.org/10.1017/9781009367509.011 Published online by Cambridge University Press
TABLE
. The Neolithic determinants of (modern) cohesive institutions
Dependent variable:
Method of estimation Democratic stock (up to ) Precolonial state history
POLITY in
Democ. stock
Precol. state history
Democ. stock
POLITY in
()
()
()
()
()
OLS . (.)***
OLS −. (.)*
. (.)* −. (.) −. (.) −. (.)
. (.)*** . (.) . (.) −. (.)
. (.)*** −. (.) . (.)*** . (.) −. (.)*
.
.
.
NEOLITHIC Log GDP per capita ( US$) GDP per capita growth (% annual) Log population Constant R-squared F-statistic
OLS
SLS
SLS
−. (.)***
−. (.)*
. (.)*** . (.)** . (.)** −. (.)**
. (.)*** . (.) . (.) −. (.)*
. .
. .
Notes: Sample of non-oil-producing transfer recipients. Estimation via OLS. Robust standard errors reported in parentheses. *, **, *** = significant at , , and percent respectively.
How Rents Affect Dictatorship and Violence
The remaining specifications in Table . firmly establish a causal relationship using SLS. Column reports the first-stage regression. It shows that countries that transitioned to sedentary agriculture at an earlier date (i.e., with a larger value of NEOLITHIC) exhibit longer precolonial state histories. This relationship is highly statistically significant, with a corresponding F-statistic ð¼ 49:1Þ that exceeds the threshold for a strong instrument of . Armed with this instrument, columns and report the second stage estimates. Column shows that precolonial state history adversely affects a country’s democratic trajectory. Instrumented precolonial state history causes a reduction in a country’s accumulated democratic experience. Focusing on the year in which oil prices tanked, column shows that instrumented precolonial state history (with a p-value ¼ 0:06) lowers a country’s POLITY score in . In combination, the results in columns – show that NEOLITHICi is a robust determinant of institutions in non-oil-producing recipients. Thus, I use this exogenous component of cohesive institutions in the DD framework. C Decline in Transfers and Heightened Political Violence I now turn to the second step. Columns – in Table . present the reduced form DD estimates corresponding to Equation .. While countries, on average, experienced less political violence during the period of low oil prices (i.e., the coefficient on POST t is negative), this pacifying effect was increasingly offset in countries with less cohesive institutions, as indicated by the positive and statistically significant interaction term. The interactive effect is substantively important. For example, the typical Muslim transfer recipient (with NEOLITHICi equal to .) experienced a . net index point increase in political violence when oil prices were low ð¼ 5:7∗ 0:118−0:553Þ. In contrast, Latin American countries experienced a net . index point decline in political violence. In column , this interactive effect remains unchanged in a specification that controls for a country’s geographic location (corresponding to the regional categorizations from the case studies), level of ethnolinguistic fragmentation, and an indicator variable for the period after the Cold War. The latter two controls account for several competing explanations for civil war and democracy in the late s and s, including the role of the Cold War’s termination (e.g., Kalyvas and Balcells ) and societal fragmentation (e.g., Montalvo and Reynal-Querol ). In general, these confounders are not robust determinants of political violence. In column , I subsume the effect of geography and societal divisions with country fixed effects. While this specification explains more of the variation in political violence (as indicated with the higher R-squared), it does not attenuate the interactive effect. In fact, the point estimate on NEOLITHICi POST t increases in magnitude (to .) and remains statistically significant. To strengthen the relationship between a country’s transition to settled agriculture, political institutions (in ), and political violence, I now turn to a set of
https://doi.org/10.1017/9781009367509.011 Published online by Cambridge University Press
https://doi.org/10.1017/9781009367509.011 Published online by Cambridge University Press
TABLE
. Political violence with changes in oil prices and exogenous institutions
Dependent variable:
Political violence Reduced form
POST NEOLITHIC NEOLTHIC
Instrumental variables
()
()
()
. (.)** −. (.)
. (.)** −. (.)
. (.)**
POST State History State History POST
−. (.)**
ELF in Muslim Latin America Soviet Satellite Post–Cold War Country fixed effects R-squared No. observations
.
−. (.)** . (.) . (.) . (.) −. (.)* . (.) .
−. (.)**
()
()
()
. (.)** −. (.) −. (.)**
. (.)** −. (.) −. (.)* . (.) . (.) . (.) −. (.)* . (.)
. (.)**
. (.) Yes .
.
.
−. (.)**
. (.) Yes .
Notes: Robust standard errors, clustered by parentheses. *, **, *** = significant at , , and percent respectively. The dependent variable is a -point index: (peace/democracy), (repression), (civil war). The constant is not reported.
How Rents Affect Dictatorship and Violence
instrumental variable (IV) results corresponding to Equation .. Column shows that instrumented STATE HISTORY i POST t exhibits a robust, positive effect on political violence. This positive interactive effect implies that countries with less cohesive institutions (as measured with longer state histories) experienced higher levels of political violence during the period of oil prices (when foreign transfers declined). The direct effect of instrumented STATE HISTORY i has no effect on political violence. The instruments are strong in the first-stage regression, notably for the interaction term. The causal effect of instrumented STATE HISTORY i NEOLITHICi increases in magnitude in specifications that control for several competing explanations, such as ethnic fragmentation (column ) and with country fixed effects (column ).
. This chapter presented the negative cases to the findings in the previous one by probing the effects of largely unearned (nontax) foreign transfers on political violence in non-Muslim countries in Latin America and Eastern Europe and for a global sample of transfer recipients. In combination, this chapter’s case study evidence and cross-national statistical analysis demonstrate a decline in foreign transfers – particularly those that accrue to the central government – can explain transitions from dictatorship to democracy in the late twentieth century in societies whose preexisting institutions are more cohesive (at least relative to conquest societies). This central finding is consistent with corollary B from Chapter that a decline in rents can facilitate peaceful transition from dictatorship to democracy in societies whose preexisting institutions are more cohesive. This finding, however, is opposite to that presented in the previous chapter in which a decline in foreign aid – another type of foreign transfer that accrues to a government – led to civil war in Muslim recipients; countries exhibiting ex ante institutions that tend to be less cohesive. Together, the findings in this and the previous two chapters ( and ) illustrate three central points to understanding how rents can affect patterns of political violence. First, greater level of rents can sustain dictatorship. Second, their decline can contribute to civil war or transitions to democracy. Third, whether a decline in rents is peaceful or violent hinges on the quality of preexisting cohesive institutions: Where these institutions are more cohesive, democracy is possible (e.g., Latin America, Eastern Europe). In contrast, where preexisting institutions are less cohesive – and where other forms of rents cannot compensate (e.g., geopolitical influence is limited) – a decline in rents can foster high-intensity civil war (as was the case in Muslim aid recipients). Thus, understanding the interplay of rents with the quality of preexisting cohesive institutions offers a lens to study cross-national patterns of democracy, dictatorship, and civil war in the late twentieth century.
https://doi.org/10.1017/9781009367509.011 Published online by Cambridge University Press
P ART IV CONCLUSION
https://doi.org/10.1017/9781009367509.012 Published online by Cambridge University Press
https://doi.org/10.1017/9781009367509.012 Published online by Cambridge University Press
Conclusion
This book has endeavored to explain why many contemporary Muslim societies are predisposed to political violence: dictatorship and civil war. My answer focuses on the confluence of two factors: the institutional legacy of Muslim conquest and variation – both increases and decreases – in the level of rents. This final chapter briefly summarizes the book’s central argument and evidence. I then discuss some implications of the book’s two key independent variables: rents and conquest. I situate the former in a broader discussion of economic and political development. For the latter, I discuss the book’s implications for political and economic liberalization in Muslim societies.
. This book has embarked on understanding why many contemporary Muslim countries exhibit an elevated propensity for political violence, characterized by dictatorship and civil war. In diagnosing why, I developed a theory with supportive empirics that the institutional legacy of Muslim conquest and variation in rents help explain patterns of dictatorship and civil war across many contemporary Muslim societies. The theoretical argument stems from an analytical framework developed in Part I of the book. Here, the main contribution is the development of a new theory linking variation in rents and preexisting cohesive institutions to patterns of dictatorship/democracy and civil war/peace in a unified framework. To summarize, the theory generates three core predictions: . Buoyant levels of rents sustain dictatorship, particularly in societies that already feature less cohesive institutions. . A decline in rents can foster civil conflict in countries where preexisting institutions are less cohesive.
https://doi.org/10.1017/9781009367509.013 Published online by Cambridge University Press
Conclusion
. A decline in rents can foster transitions to democracy in countries where preexisting institutions are more cohesive. The theory’s predicted outcomes – dictatorship, civil war, and democracy – are conditional. They covary with two parameters within a particular society: its quality of preexisting cohesive institutions and prevailing level of rents. Accordingly, I then show in the empirical parts of this book how these parameters affect political violence in many contemporary Muslim societies. Part II of the book examines why many contemporary Muslim societies feature a “stock” of institutions that are less cohesive in the contemporary period. As described in Chapter , societies with less cohesive institutions are those where the incumbent does not equitably allocate the state’s resources to opposition groups. As I argue and empirically document, less cohesive societies tend to be nondemocratic (e.g., see Chapter ). With this interpretation, Part II proceeds to show how many contemporary Muslim societies exhibit a sticky set of authoritarian political institutions and it traces their origins to events in the distant past. Organized in three chapters (, , and ), Part II argues and presents evidence that the expansion of Islam via military conquest (from to ) set these “conquest” societies on a trajectory of authoritarian governance. Following the death of Prophet Muhammad in CE, Islam spread rapidly through military conquest. This process introduced two institutions that consolidated political authority in a dictator (caliph): the use of elite slave soldiers (mamluks) as the backbone for the regime’s repressive capacity and a non-hereditary form of compensation for their service (iqta). Legitimized by clerics (ulema) this institutional equilibrium of centralized autocracy in conquered territories persisted through to the modern era. As a consequence, in the contemporary era, societies exposed to Muslim conquest inherited a stock of less cohesive political institutions. In the contemporary era, the presence of these preexisting less cohesive institutions has shaped whether in conquest societies, buoyant levels of rents – e.g., government revenue from oil and foreign aid – has sustained dictatorship or fostered civil war when rents are lower (decline). Structured in three chapters
Chapter provides a succinct historical narrative of Muslim conquest and its governing institutions. Chapter corroborates this historical account with quantitative analysis showing how Muslim conquest helped “autocratize” conquered societies and generated an institutional equilibrium of centralized autocracy. Chapter extends the analysis to the present by tracing the political legacy of Muslim conquest: Societies whose modern-day territory experienced greater control by Muslim armies (and its subsequent governing institutions) tend to be significantly less democratic today. Together, the analysis in Chapters – show how the institutional legacy of Muslim military conquest (and a supporting governing coalition involving a dictator, the military, and clergy) endowed many contemporary Muslim societies with relatively “sticky” less cohesive institutions.
https://doi.org/10.1017/9781009367509.013 Published online by Cambridge University Press
Conclusion TABLE
. Rents and politics in conquest and non-conquest states since
Country sample:
Democracy Civil war
Conquest
Non-conquest
Petrostates
Recipients
Petrostates
Recipients
()
()
()
()
. (.) . (.)
. (.) . (.)
. (.) . (.)
. (.) . (.)
Notes: Group means and corresponding standard deviations below in parentheses based on data from Chapters and . “Petrostates” refer to countries that produce oil. “Recipients” refer to nonoil-producing countries that are the recipients of various financial transfers (e.g., foreign aid, trade subsidies, petrodollars). Democracy is the normalized POLITY score, where a value closer to corresponds to greater democracy. Civil war is the incidence of high intensity civil war with at least , battle deaths per year.
(, , and ), this is empirically documented in Part III of the book with robust cross-national evidence that is careful with causal identification. Table . summarizes some of the key findings in Parts II and III. It reports the average and standard deviation for the level of democracy and incidence of civil war across groups of countries categorized by their conquest status and type of rent. The table shows conquest states tend to be less democratic (columns and ), on average, relative to countries not exposed to Muslim conquest (columns and ). Among conquest states, petrostates tend to be more stable dictatorships than recipient states. In any given year, conquest recipient states (column ) – with more variable sources of rents (see Chapter ) – are more conflict prone and exhibit higher and more variability in their democracy scores (for the latter, as indicated with its higher standard deviation). The dynamics are reversed in non-conquest states: Recipients tend to be more democratic and less conflict prone (column ) compared to nonconquest petrostates (column ). Having summarized the book’s argument and evidence, I now discuss the book’s two key variables – rents and the institutional legacy of Muslim conquest – and their implications for economic development and democracy.
. A Political Transfer Problem The importance of rents, especially those that emanate from abroad, such as foreign aid and geopolitical rents (e.g., US troop deployments in oil producers), in explaining patterns of political violence contributes to a rich literature describing how unearned income can affect economic and political
https://doi.org/10.1017/9781009367509.013 Published online by Cambridge University Press
Conclusion
development in recipient countries. Motivated by these works, the formal model developed in Chapter bridges competing predictions among existing theories linking rents to political stability. Some theories show that rents foster political stability, while others demonstrate the opposite. I provide conditions under which lower levels of rent can facilitate democracy or civil war. Upon receiving these rents, nondemocratic regimes can use them to entrench their autocratic rule. As these rents wane, however, these regimes become increasingly vulnerable to challenges from opposition groups. In response, I provide conditions under which these regimes can either democratize or engage in armed conflict. Together, these dynamics can be characterized as a “political transfer problem.” As Eric Werker and I first described in an unpublished working paper (Ahmed and Werker ), this political transfer problem is a reference to the original “transfer problem,” which described the relative price distortions inherent in Germany’s reparation payments after World War I. Writing in The Economic Journal, John Maynard Keynes (, ) described the transfer problem as “converting the German money so received into foreign currency,” not as the German authorities’ ability to collect sufficient taxes to be used as reparations. If making substantial reparations meant that German demand had to fall, but foreign demand was for products other than those that would have been produced for German consumers, then a shift in production would need to occur in order for the value of those taxes not to be diminished away in relative price movements. The transfer problem enjoyed a vigorous debate in the interwar years (most famously between Keynes and Bertil Ohlin) and fell out of fashion, with only the occasional attempt at resurrection (e.g., Brakman and van Marrewijk ). Like the economic transfer problem, the political transfer problem also involves an effect on two macroeconomic equilibria resulting from a transfer of resources from one country to another. In the economic transfer problem, the flow of resources may lead to the production of goods and services being unbalanced. In the political transfer problem, the flow of resources may lead to the incentives for investment in political violence to become unbalanced. This
As cited throughout the book, some prominent work includes but is not limited to Ahmed (), Besley and Persson (a), Dunning (), Karl (), Levi (), Madhavy (), Mahdavi (), Menaldo (), Morrison (), Ross (), Adam Smith (). That the literature on the transfer problem has been largely silent over the past eighty years provides some possible hint as to why the political transfer problem has remained unarticulated. Since the World Wars, there has been no situation of war reparations involving such a share of global GDP. After World War I, for example, Germany was required to repay around $ billion over thirty years, although by the time Keynes was writing in the Young Plan had reduced the debt to $. billion, approximately percent of global GDP, paid out over fifty-nine years. Although Germany never completed its reparations payments (thanks in no small part to the hyperinflation of the interwar years), the sheer size of the payments, and the potential distortions to the economics, did not go unnoticed.
https://doi.org/10.1017/9781009367509.013 Published online by Cambridge University Press
Conclusion TABLE
.: Developmental challenges associated with rents Economic
Political
Domestic rents
Dutch disease
Resource curse
Foreign rents
The transfer problem
The political transfer problem
Note: Based on table in Ahmed and Werker ().
book identifies three sources of foreign transfers in Muslim countries that have affected their macro-level politics: in oil producers, the sale of oil abroad and the provision of geopolitical rents (by Western states) in strengthening their repressive capacity, while in non-oil producers, the financing of political stability through foreign aid. More generally, characterizing the political transfer problem also squares off a two-by-two matrix that has been heretofore incomplete. Unearned income (rents) can theoretically generate both economic and political problems domestically. When a state experiences a surge in income from natural resources (Corden and Neary ) or foreign aid (Rajan and Subramanian ), the increased demand on the currency can bring about real exchange rate appreciation, or Dutch disease, making other potential exports less attractive. From a political perspective, the increase in unearned income can lead to the so-called resource curse, which describes the political conflict and corruption that may result over the rents. The traditional transfer problem results when unearned income must be transferred to another country and the price mechanism gets tested (possibly necessitating a reallocation of production), just as with Dutch disease. Similarly, the political transfer problem represents how the politics are distorted when unearned income is transferred from one country to another; it is the internationalization of the resource curse. Table . depicts this in a twoby-two matrix. Table . clarifies how rents can pose challenges to economic and political development. The top row describes how domestic rents can affect economic and political development, through channels consistent with existing theories such as Dutch disease and the resource curse. This book’s original theoretical and empirical contribution is depicted in the bottom right cell in which foreign transfers that accrue to government budgets (e.g., foreign aid, petrodollar investments, trade subsidies, and geopolitical rents) can pose challenges to economic and political development in recipient countries. This articulation of a political transfer problem is pertinent for scholars and practitioners in various fields, especially economic development and international politics.
For example, the political transfer problem is relevant for scholars and practitioners of aid effectiveness (e.g., Alesina and Dollar , Bermeo , Bueno de Mesquita and Smith , Dietrich , , Kuziemko and Werker , Winters and Wright ), the political economy of
https://doi.org/10.1017/9781009367509.013 Published online by Cambridge University Press
Conclusion
B Private Rents In some instances, unearned income may not necessarily have to accrue directly to governments to generate pernicious political effects. In dictatorships and some kleptocratic democracies (e.g., Mexico under the single party rule of the PRI), governments may have a political incentive to encourage rents that accrue to private actors, such as firms and individuals, as a strategy to stay in power (Ahmed , Bueno de Mesquita et al. ). Governments in many Muslim societies may have acted in this fashion. For example, in prior work, I have documented how migrant income remitted back to their families has helped many less democratic governments in Muslim countries finance corruption and extend their duration in office, through, for example, an expenditure-switching strategy (Ahmed , ). Governments may also be able to generate rents through policy choices, particularly in international trade (Grossman and Helpman ). One such example is partial trade liberalization, which may strategically shield regime supporters from trade competition, sustain their stream of economic rents, and concomitantly buttress nondemocratic rule in many Muslim societies (Ahmed and Malik ). This finding corroborates recent scholarship with industry level data from Egypt and Jordan, where the reigning dictatorship has used a combination of tariff and non-tariff measures to better ensure its political longevity. These policies seek to protect specific sectors with ties to important constituencies, such as co-ethnics, the military, and business elites (Eibl and Malik , Monroe ).
international investment and trade (e.g., Ahmed , Cordova and Meissner , Freeman and Quinn , Jensen et al. , Milner and Kubota , Pinto , Tomz ) and those at the intersection of the international security and international economics (e.g., Brooks , Gartzke , Gartzke et al. , Gowa , Gowa and Mansfield ). Rather than taxing remittance income, governments in Muslim societies have resorted to an expenditure-switching strategy to incorporate remittance income into their revenue base. As households spent some of their remittance income on various substitutable welfare goods (e.g., basic healthcare), governments concurrently reduced their provision of these goods and diverted the unused funds toward patronage that helps win support from a narrow group of regime supporters (e.g., increase spending on the military, bloated public sector payrolls that reward regime loyalists). In ongoing work, Adeel Malik and I document how Muslim societies have missed two important global “waves” since the s: democracy (as studied in this book) and globalization. Despite being in regions conducive for international trade, most Muslim societies have above-average tariff rates that protect certain industries, particularly those with ties to the regime. Our analysis reveals a strong relationship between economic closure in international trade (e.g., high tariffs rates, prevalence of non-tariff barriers) and the longevity of nondemocratic governments in Muslim countries. In Egypt, Ferdinand Eibl and Adeel Malik () study the impact of political connections on trade protection. They compile novel sector-level data on non-tariff measures (NTMs) and politically connected businessmen in Mubarak-era Egypt, exploiting within-sector variation in NTMs to demonstrate that sectors populated by politically connected businessmen experienced
https://doi.org/10.1017/9781009367509.013 Published online by Cambridge University Press
Conclusion
. A Governing Coalitions The importance of rents should not overshadow the other key variable in this book’s account of political violence in Muslim societies: the institutional legacy of Muslim conquest. While mamluk institutions no longer exist in contemporary Muslim societies, the propagating agents supporting this original institutional configuration remain important actors in political and economic activity. These agents are the dictator – monarch (e.g., Saudi Arabia, Jordan), personalist ruler (e.g., Libya under Qaddafi, Syria under Assad), or a military officer (e.g., Egypt, Pakistan –) – clerics, and the military. In conjunction with sufficient rents, these agents help legitimize nondemocratic rule and serve to suppress domestic opposition. In recognizing the composition of this tripartite governing coalition, two points are worth highlighting in understanding the prospect for political change in many contemporary Muslim societies. First, political preferences are not necessarily static among these agents. Political change is possible – perhaps in the direction of political liberalization – if the preferences of these actors change. For example, the prospect of declining oil revenues in the future has incentivized some conquest petrostates ruled by monarchies (e.g., the United Arab Emirates, Kuwait) to diversify and “open” their economies, a process that may engender greater political freedoms. In several conquest non-petrostates, the ruling dictatorship – whether a monarchy (e.g., Algeria, Jordan) or military regime (e.g., Egypt) – has permitted parliamentary elections, although the devolution of power from the executive to the legislature has been largely muted. In these instances, the advent of parliaments can offer opportunities to identify and co-opt anti-regime political leaders (Gandhi , Gandhi and Lust-Okar ). In Muslim societies outside of the MENA region, the military has sometimes served as a “caretaker government” as the country transitions to
systematically higher incidence and density of non-tariff protection. Eibl and Malik interpret their findings as evidence of cronyism that helped buttress Mubarak’s dictatorial rule in Egypt. In Jordan, Steve Monroe () probes how ethnic ties and selective trade liberalization underlie King Abdullah’s governing strategy. Monroe exploits plausibly exogenous variation in ministerial turnover and system of ethnic quotas in King Abdullah’s cabinet to study the formulation of trade policy. His analysis reveals that Jordanian policymakers are systematically less likely to liberalize non-ethnics’ industries in two domains of trade policy: free trade agreements and temporary tariff cases. Other actors and groups within a society may also support dictatorship. This includes business elites and co-ethnics of the ruling regime. And often, there is significant overlap between the composition of commercial elites and ethnic identity (Monroe ). It is worth noting that economic diversification may engender varying political consequences. For example, as Michael Herb (, ) observes in “Kuwait the parliament challenges the predominance of the ruling family . . . and the United Arab Emirates remains resolutely authoritarian.”
https://doi.org/10.1017/9781009367509.013 Published online by Cambridge University Press
Conclusion
greater democracy (e.g., Bangladesh in and ). In short, while the political preferences of dictators in contemporary Muslim societies may change, there is no guarantee that this will necessarily lead to political liberalization. A related second point is the important role of the ulema (clerics) in day to day politics in many Muslim societies. As argued and demonstrated in this book, prior studies that find a negative association between Islam and democracy are likely capturing the anti-democratic effects of religious legitimization (from a society’s ulema) as part of the institutional legacy of Muslim conquest (see Section . in Chapter ). This insight calls for understanding why the ulema in many contemporary Muslim societies may prefer dictatorship over democracy. One account argues the ulema exhibits a status quo bias and is resistant to political change that may undermine its prominent status in many Muslim societies, such as the interpreters of Islamic law (Kuran ). Relatedly, many clerics fear political liberalization may induce secularism and an overall decline in their society’s level of piety. A fear of change, of course, is not the sole impetus for why the ulema may prefer authoritarianism. Another factor is the policy concessions favored by religious elites that dictators may extend. The legitimation that clerics provide for many dictatorial regimes, particularly Arab monarchies, elevates their significance as political allies. In turn, this political pact can affect a state’s domestic and foreign policies. For example, on the domestic front, dictators may be more inclined to implement and enforce Sharia law, while on the international front, dictators may be more inclined to “export” a certain brand of Islam to other Muslim societies, such as the financing of Wahhabi madrassas overseas by the Saudi government and the Iranian regime’s strategy of expanding Shi’ite influence abroad (Kepel ). In each instance, these foreign policies frequently align with the policy preferences of that society’s religious elite.
As discussed shortly, international pressure/support and a changing economic environment may incentivize the tripartite governing coalition to prefer liberal economic and political reform. The infusion of religion and politics, of course, applies in many non-Muslim societies as well. For example, evangelical Christians in the United States and the Catholic church in Latin America (e.g., Andersen and Jensen , Lindsay ). When separating the effect of Islam from the institutional legacy of Muslim conquest, Islam tends to have no association on democracy (e.g., see Table . column and Figure .) These policy preferences will vary across Muslim societies and the versions/brands of Islam the clerics follow, such as Wahhabism, Shi’ism, or Sufism. It is worth nothing, the ulema’s political and policy ambitions are not confined to dictatorships nor conquest societies. For instance, in non-conquest Muslim societies with elections (e.g., Bangladesh, Malaysia), Islamic parties tend to reflect the policy preferences of the ulema (e.g., greater implementation of Sharia law, rescind democratic/pro-Western laws). In recent years, Islamic parties have greatly shaped political discourse in these societies and some of these religious parties have performed quite well in legislative elections. In conquest societies (e.g., Pakistan, Egypt), some religious parties have greatly influenced the formation of the government. Most notably was the election of the Muslim Brotherhood in Egypt following the Arab Spring.
https://doi.org/10.1017/9781009367509.013 Published online by Cambridge University Press
Conclusion
B Democratic Transitions? The discussion in the preceding section suggests that political and economic liberalization is possible in contemporary Muslim societies if each country’s governing coalition chooses to do so. The fact that many governing coalitions have not done so, as I argue in this book, can be traced back in large part to the creation of mamluk institutions (and associated political alliances) during the period of Muslim conquest (see Part II). In the modern era, the institutional legacy of Muslim conquest has created a sticky set of authoritarian institutions (and supporting governing coalitions) that do not incentivize political actors to change their behavior. Furthermore, this authoritarian equilibrium has been reinforced with buoyant levels of domestic and foreign rents (see Part III). The importance of rents in sustaining dictatorship in conquest states suggests that their decline may offer an opportunity for political change. The analytical framework developed in Chapter predicts that in conquest societies, a reduction in rents can elevate the prospect for violence between the incumbent regime and opposition groups. The model, however, is agnostic as to whether after the cessation of violence, politics will revert to autocracy or whether some political liberalization is possible. For the latter, several recent studies offer insights as to how civil conflict may provide pathways toward peace with less repressive politics and possibly democracy after the cessation of violence. For example, Alia Matanouk () identifies electoral participation provisions as an important component of post-conflict settlements that make peace more durable and less repressive. In contrast, Ryuku Huang () argues the prospects for greater democracy after civil war depend on strategies of “rebel governance” during the civil war: “Democratization is more likely to follow wars in which the rebels rely heavily on civilians for warfighting support” (Huang , ). Huang notes that the availability of rents – whether geopolitical (e.g., external support from foreign governments) or economic (i.e., oil, diamonds, foreign assistance) – tends to lower the incentives for rebels to seek support from civilians.
Reyko Huang () examines the data from all civil wars since and renders two observations. “First, civil war and democratization have often come hand in hand; many states have in fact emerged from civil war more democratic than they were before the war. And second, there is variation in how political regimes change through the experience of civil war; some democratize, some autocratize and some exhibit little change in their regime type” (Huang , ). Matanock notes that “peace agreements are increasingly based on provisions that establish former rebel groups as political parties set to participate alongside government parties in postconflict elections, and compared to other settlements, these provisions are associated with an percent increase in the chance that a settlement with produce enduring peace” (Matanock , ). Huang articulates some of the choices rebel groups have in fighting: “Some choose to directly involve ordinary people to incite and encourage their active support; others choose to fight with little concern for popular involvement, whether because they have ready access to external
https://doi.org/10.1017/9781009367509.013 Published online by Cambridge University Press
Conclusion
These insights shed light on the prospect for peace and perhaps, democracy in conquest states after periods of civil conflict. Here, the evidence presented in this book is mixed. In conquest aid recipients, the analysis in Chapter showed that either scenario – reversion to autocracy or movement to democracy – is possible. In Somalia, for example, the propitious decline in foreign aid starting in the mids incited civil war and an eventual collapse of the central government. More than thirty years after the outbreak of violence, Somalia today is politically fragmented and is deemed a “failed state.” Somalia’s tragic post-conflict political trajectory, however, is not universal across Muslim aid recipients. During the s and early s, buoyant amounts of foreign aid from the Persian Gulf helped finance military rule in Pakistan and Bangladesh. In the mids, as oil prices tanked and with it a decline in aid flows, military rule in each country eventually ended, offering an opportunity for greater democracy. While both countries have reverted to military rule at times since , politics today in each society features electoral politics. On average, however, after the cessation of violence, Muslim aid recipients have tended to remain nondemocratic (see Figure . and Table .). The autocratic legacy of Muslim conquest in aid recipients may not necessarily be undone after periods of civil unrest. In conquest petrostates, the prospects for democracy seem limited. First, the institutional footprint of Muslim conquest is often much stronger in these countries: Their modern-day territory was completely conquered by Muslim armies by (e.g., Saudi Arabia, Iran, Libya). Second, oil rents comprise a significant share of their nations’ economic production and government revenue. The analysis in Chapter shows that oil revenues have sustained dictatorship without heightening the prospect of civil war, the latter due in tandem with an implied security guarantee provided by the United States. A decline in oil wealth in the future – perhaps due to changing fossil fuel consumption to combat climate change and/or greater price competitiveness of renewable energy – may bring about political change, possibly toward greater political liberalization. The formal model in Chapter predicts that in conquest petrostates, a decline in rents is likely to elevate the prospect for civil conflict. Due in part to its rare occurrence in these societies (given their buoyant levels of oil wealth and availability of geopolitical rents), the analysis in Chapter does not directly examine post-conflict dynamics in conquest petrostates. Nevertheless, one can speculate on the prospects for democracy after civil war in these countries. To “clear the
patrons or to profits from natural resources that precludes the need to draw on civilian support, or because their political ideologies do not demand a popular focus, or because they lack the skill or capacity for mass mobilization” (Huang , ). Based on The Fund for Peace’s “Fragile States Index.” Data available at: https:// fragilestatesindex.org/country-data/ (Accessed: May , ). Section . does illustrate the trajectory of democracy in conquest and non-conquest petrostates. Whereas non-conquest petrostates (e.g., Nigeria, Venezuela) have tended to become more democratic over time, this has not been the case in conquest petrostates.
https://doi.org/10.1017/9781009367509.013 Published online by Cambridge University Press
Conclusion
brush” on democratization after civil war, Virginia Fortna and Ryuku Huang () statistically evaluate a battery of factors, such as economic development, resource wealth, and peacekeeping missions. They find “the determinants of democratization to be much the same for post–civil war societies as for other societies” where “economic development aids democratization while oil wealth hinders it” (Fortna and Huang , Abstract). This conclusion in conjunction with the prevalence of less cohesive institutions (stemming from the institutional legacy of Muslim conquest) suggests that if conquest petrostates experience civil war, they are likely to remain nondemocratic afterward. Of course, with all research in the social sciences – which deals with probabilistic effects – this grim outcome for conquest petrostates is not set in stone. While geopolitical rents, in the form of an implied external security guarantee (from the United States), have helped many conquest petrostates thwart internal rebellion (see Chapter ), these policies by foreign powers can change. After the Cold War, foreign interventions have been increasingly deployed by the international community to facilitate transitions to durable peace and possibly democracy after the cessation of civil conflict. Climate change and/ or the emergence of renewables as a viable alternative to fossil fuels may weaken the United States’ incentives to provide a security guarantee to many Muslim petrostates in the future. On balance, however, the theory and empirical evidence marshaled in this book point to muted prospects for political and economic liberalization in many Muslim societies. This conclusion contributes to burgeoning scholarship among economists, historians, and political scientists that seeks to explain the persistently low levels of economic and political development in many contemporary Muslim societies. Like several of these recent works, my explanation relies on historical factors; I focus on the institutional legacy of Muslim conquest. History, though, provides an incomplete explanation for important temporal patterns of political dynamics observed in many contemporary Muslim societies. A crucial addendum to the legacy of Muslim conquest is the role of contemporary rent streams that accrue to governments. In conjunction with a society’s quality of institutional cohesion, variation in rents help explain patterns of political violence across countries and crucially, over time. Importantly, the theory and empirics in this book shows that Islam and related factors, such as Arab culture are not the source of dictatorship and violence in many contemporary Muslim societies; rather, political institutions and rents are more consequential causal factors.
While not a test of the arguments advanced in this book, the US-led invasion of Iraq in is illustrative of the difficulty in consolidating democracy in a conquest petrostate. After the ousting of Saddam Hussein, international efforts at establishing a relatively peaceful democracy have been complicated with competition for the control and allocation of oil wealth among the various regions and ethnic/religious groups in Iraq (i.e., the Kurds, Shia, Sunni). See for example, Doyle and Sambanis (), Heldt (), Steinert and Grimm ().
https://doi.org/10.1017/9781009367509.013 Published online by Cambridge University Press
https://doi.org/10.1017/9781009367509.013 Published online by Cambridge University Press
References
Abadie, Alberto, Alexis Diamond, and Jens Hainmueller. . “Comparative Politics and the Synthetic Control Method,” American Journal of Political Science, (): –. Abramson, Scott and Carles Boix. . “Endogenous Parliaments: The Domestic and International Roots of Long-Term Economic Growth and Executive Constraints in Europe,” Working paper. Acemoglu, Daron and Simon Johnson. . “Unbundling Institutions,” Journal of Political Economy, (): –. Acemoglu, Daron and James A. Robinson. . Economic Origins of Dictatorship and Democracy, New York: Cambridge University Press. . Why Nations Fail: The Origins of Power, Prosperity, and Poverty, New York: Crown Publishing. . “Rents and Economic Development: The Perspective of Why Nations Fail,” Public Choice, : –. Acemoglu, Daron, Simon Johnson, and James A. Robinson. . “The Rise of Europe: Atlantic Trade, Institutional Change, and Economic Growth,” American Economic Review, (): –. Acemoglu, Daron, Simon Johnson, James A. Robinson, and Pierre Yared. . “Income and Democracy,” American Economic Review, (): –. Acemoglu, Daron, Suresh Naidu, Pascual Restrepo, and James A. Robinson. . “Democracy Does Cause Growth,” Journal of Political Economy, (): –. Adam, Hussein M. . “Somali Civil Wars,” in T. M. Ali and R. O. Matthews (eds.), Civil Wars in Africa: Roots and Resolution, Montreal: McGill-Queen’s University Press, . Ahmadov, Anar K. . “Oil, Democracy, and Context: A Meta-analysis,” Comparative Political Studies, (): –. Ahmed, Faisal Z. . “The Perils of Unearned Foreign Income: Aid, Remittances, and Government Survival,” American Political Science Review, (): –. . “Remittances Deteriorate Governance,” Review of Economics and Statistics, (): –.
https://doi.org/10.1017/9781009367509.014 Published online by Cambridge University Press
References
. The Perils of International Capital, New York: Cambridge University Press. . “Muslim Conquest and Institutional Formation,” Explorations in Economic History, , . Ahmed, Faisal Z. and Adeel Malik. . “Crony Globalization,” Working Paper. Ahmed, Faisal Z. and Eric Werker. . “Unobserved State Fragility and the Political Transfer Problem,” Working Paper. Ahmed, Faisal Z., Daniel Schwab, and Eric Werker. . “The Political Transfer Problem: How Cross-Border Financial Windfalls Affect Democracy and Civil War,” Journal of Comparative Economics, (): –. Albertus, Michael and Victor Menaldo. . “Coercive Capacity and the Prospects for Democratization,” Comparative Politics, (): –. Alesina, Alberto and David Dollar. . “Who Gives Aid to Whom and Why?” Journal of Economic Growth, (): –. Alesina, Alberto, Reza Baqir, and William Easterly. . “Public Goods and Ethnic Divisions,” Quarterly Journal of Economics, (): –. Alesina, Alberto, Arnaud Devleeschauwer, William Easterly, Sergio Kurlat, and Romain Wacziarg. . “Fractionalization,” Journal of Economic Growth, : –. Alesina, Alberto, William Easterly, and Janina Matuszeski. . “Artificial States,” Journal of the European Economic Association, (): –. Alvarez, Michael, José Antonio Cheibub, Fernando Limongi, and Adam Przeworski. . “Classifying Political Regimes,” Studies in Comparative International Development, (): –. Andersen, Jorgen J. and Silje Aslasken. . “Oil and Political Survival,” Journal of Development Economics, (): –. Andersen, Jorgen J. and Michael L. Ross. . “The Big Oil Change: A Close Look at the Haber-Menaldo Analysis,” Comparative Political Studies, (): –. Anderson, Perry. . Lineages of the Absolutist State. New York: Verso. Andersen, Thomas B. and Peter S. Jensen. . “Preaching Democracy: The Second Vatican Council and the Third Wave,” Journal of Comparative Economics, : –. Arias, Luz Marina and Desha Girod. . “Indigenous Origins of Colonial Institutions,” Quarterly Journal of Political Science, (): –. Arjomand, Said Amir. . “Legitimacy and Political Organization: Caliphs, Kings, and Regimes,” in Robert Irwin (ed.), New Cambridge History of Islam, Vol ., Islamic Cultures and Societies to the End of the Eighteenth Century, Cambridge: Cambridge University Press, . Asher, Sam and Paul Novosad. . “Dirty Politics: Natural Resource Wealth and Politics in India,” unpublished manuscript. Aslasken, Silje. . “Oil as Sand in the Democratic Machine?” Journal of Peace Research, (): –. Ayalon, David. . “Mamluk: Military Slavery in Egypt and Syria,” in David Ayalon (ed.), Islam and the Abode of War: Military Slaves and Islamic Adversaries, New York: Routledge. Baccini, Leonardo and Johannes Urpelainen. . Cutting the Gordian Knot of Economic Reform: When and How International Institutions Help, New York: Oxford University Press.
https://doi.org/10.1017/9781009367509.014 Published online by Cambridge University Press
References
Baldwin, Kate and John D. Huber. . “Economic versus Cultural Differences: Forms of Ethnic Diversity and Public Goods Provision,” American Political Science Review, (): –. Banks, Arthur. . “Banks Cross-National Time Series Data Archive” [computer file]. Jerusalem: Databanks International. Retrieved from: www.databanks.sitehosting .net/. Bapat, Navin. . Monsters to Destroy: Understanding the War on Terror, London: Oxford University Press. Barkey, Karen. . Bandits and Bureaucrats: The Ottoman Route to State Centralization, Ithaca, NY: Cornell University Press. Barkey, Karen and George Gavrilis. . “The Ottoman Millet System: NonTerritorial Autonomy and Its Contemporary Legacy,” Ethnopolitics, (): –. Baron, Rueben M. and David A. Kenny . “The Moderator-Mediator Variable Distinction in Social Psychological Research: Conceptual, Strategic, and Statistical Considerations,” Journal of Personality and Social Psychology, (): –. Barro, Robert J. . “Determinants of Democracy,” Journal of Political Economy, (S): –. Basedau, Matthias and Jann Lay. . “Resource Curse or Rentier Peace? The Ambiguous Effects of Oil Wealth and Oil Dependency on Violent Conflict,” Journal of Peace Research, (): –. Basedau, Matthias and Thomas Richter. . “Why Do Some Oil Exporters Experience Civil War but Others Do Not? A Qualitative Comparative Analysis of Net Oil-Exporting Countries,” European Political Science Review, (): –. Bates, Robert H. and Da-Hsiang Donald Lien. . “A Note on Taxation, Development, and Representative Government,” Politics and Society, (): –. Beblawi, Hazem. . “The Rentier State in the Arab World,” in Hazem Beblawi and Giacomo Luciani (eds.), The Rentier State, New York: Croom Helm, –. Bellin, Eva. . “The Robustness of Authoritarianism in the Middle East: Exceptionalism in Comparative Perspective,” Comparative Politics, (): –. Bermeo, Sarah B. . “Foreign Aid and Regime Change: A Role for Donor Intent,” World Development, (): –. . “Aid is Not Oil: Donor Preferences, Heterogeneous Aid, and the AidDemocratization Relationship,” International Organization, : (Winter): –. . “Aid Allocation and Targeted Development in an Increasingly Connected World,” International Organization, (): –. Bernstein, Thomas P. and Lu Xiaobo. . “Taxation without Representation: Peasants, the Central and the Local States in Reform China,” The China Quarterly, (September): –. Besley, Timothy, and Torsten Persson. . “Repression or Civil War?” American Economic Review, (): –. . “State Capacity, Conflict, and Development,” Econometrica, (): –. a. Pillars of Prosperity: The Political Economics of Development Clusters, Princeton, NJ: Princeton University Press. b. “The Logic of Political Violence,” The Quarterly Journal of Economics, (): –. Besley, Timothy and Marta Reynal-Querol. . “The Legacy of Historical Conflict: Evidence from Africa,” American Political Science Review, (): –.
https://doi.org/10.1017/9781009367509.014 Published online by Cambridge University Press
References
Besteman, Catherine. . “Violent Politics and the Politics of Violence: The Dissolution of the Somali Nation-State,” American Ethnologist, (): –. Betz, Timm, Scott J. Cook, and Florian Hollenbach. . “On the Use and Abuse of Spatial Instruments,” Political Analysis, (): –. Bjorvatn, Kjetil and Alireza Naghavi. . “Rent Seeking and Regime Stability in Rentier States,” European Journal of Political Economy, (): –. Blattman, Christopher and Edward Miguel . “Civil War,” Journal of Economic Literature, (): –. Blaydes, Lisa. . “State Building in the Middle East,” Annual Review of Political Science, : –. . State of Repression: Iraq Under Saddam Hussein, Princeton, NJ: Princeton University Press. Blaydes, Lisa and Eric Chaney. . “The Feudal Revolution and Europe’s Rise: Political Divergence of the Christian West and the Muslim World before CE,” American Political Science Review, (): –. Boix, Carlos. . Democracy and Redistribution. Cambridge: Cambridge University Press. . “Democracy, Development, and the International System,” American Political Science Review, (): –. Boix, Carlos, Michael K. Miller, and Sebastian Rosato. . “A Complete Data Set of Political Regimes, –,” Comparative Political Studies, (): –. Boschini Anne D., J. Pettersson, and J. Roine. . “Resource Curse or Not: A Question of Appropriability,” Scandinavian Journal of Economics, ():–. Bosworth, Edmund. . “The Steppe Peoples of the Islamic World,” in David O. Morgan and Anthony Reid (eds.), New Cambridge History of Islam, Vol . The Eastern Islamic World Eleventh to Eighteenth Centuries, New York: Cambridge University Press, –. Brakman, Steven and Charles van Marrewijk. . The Economics of International Transfers, Cambridge: Cambridge University Press. Bray, Julia. . “Arabic Literature,” in Maribel Fierro (ed.), New Cambridge History of Islam, Vol. . The Western Islamic World Eleventh to Eighteenth Centuries, Cambridge: Cambridge University Press, –. Brett, Michael. . “State Formation and Organization,” in Maribel Fierro (ed.), New Cambridge History of Islam, Vol. . The Western Islamic World Eleventh to Eighteenth Centuries, Cambridge: Cambridge University Press, –. Brocksette, Valerie, Areendam Chanda, and Louis Putterman. . “States and Markets: The Advantage of an Early Start,” Journal of Economic Growth, (): –. Brollo, Fernanda, Tommaso Nannicini, Roberto Perotti, and Guido Tabellini. . “The Political Resource Curse,” American Economic Review, (); –. Brooks, Stephen. . Producing Security: Multinational Corporations, Globalization, and the Changing Calculus of Conflict, Princeton, NJ: Princeton University Press. Brown, Archie. . The Rise and Fall of Communism, London: Random House. Brownlee, Jason M. . “Low Tide after the Third Wave: Exploring Politics under Authoritarianism,” Comparative Politics, (): –. Brownlee, Jason M., Tarek Masoud, and Andrew Reynolds. . The Arab Spring: Pathways of Repression and Reform, Oxford: Oxford University Press.
https://doi.org/10.1017/9781009367509.014 Published online by Cambridge University Press
References
Bueno de Mesquita, Bruce and Alastair Smith . “A Political Economy of Aid,” International Organization, (): –. . “Leader Survival, Revolutions, and the Nature of Government Finance,” American Journal of Political Science, (): –. Bueno de Mesquita, Bruce, Alastair Smith, James D. Morrow, and Randolph M. Siverson. . The Logic of Political Survival, Cambridge: MIT Press. Bulliet, Richard W. . “Neo-Mamluk Legitimacy and the Arab Spring,” Middle East Law and Governance, : –. Cahen, Claude. . “L’evolution de l’iqta,” Annales: ESC, :–. Cammett, Melani, Ishac Diwan, Alan Richards, and John Waterbury. . A Political Economy of the Middle East (th Edition), Boulder, CO: Westview. Casella, Alessandra and Barry Eichengreen. . “Can Foreign Aid Accelerate Stabilization?” Economic Journal, (): –. Caselli, Francesco and Andrea Tesei. . “Resource Windfalls, Political Regimes, and Political Stability,” The Review of Economics and Statistics, (): –. Caselli, Francesco, Dominic Rohner, and Massimo Morelli. . “The Geography of Interstate Resource Wars,” Quarterly Journal of Economics, (): –. Cederman, Lars-Erik, Nils B. Weidmann, and Kristian S. Gleditsch. . “Horizontal Inequalities and Ethnonationalist War: A Global Comparison”, American Political Science Review, (): –. Chaney, Eric. . “Democratic Change in the Arab World, Past and Present”, Brookings Papers on Economic Activity, Spring: –. Chaudhry, Kiren A. . The Price of Wealth: Economies and Institutions in the Middle East, Ithaca, NY: Cornell University Press. Cheibub, José A., Jennifer Gandhi, and James R. Vreeland. . “Democracy and Dictatorship Revisited,” Public Choice, (): –. Colgan, Jeff. . Petro-Aggression: When Oil Causes War, New York: Cambridge University Press. . “Oil, Domestic Conflict, and Opportunities for Democratization,” Journal of Peace Research, (): –. Collier, Paul. . Wars, Guns, and Votes: Democracy in Dangerous Places, New York: Harper Collins. Collier, Paul and Anke Hoeffler. . “On Economic Causes of Civil War,” Oxford Economic Papers, : –. . “Greed and Grievance in Civil War,” Oxford Economic Papers, (): –. Corden, W. Max and Peter J. Neary. . “Booming Sector and De-industrialization in a Small Open Economy,” Economic Journal, (December): –. Cordova, J. Ernesto Lopez and Christopher M. Meissner. . “The Globalization of Trade and Democracy, –,” World Politics, (): –. Crane, Keith. . The Soviet Economic Dilemma of Eastern Europe, Rand Corporation. Crone, Patricia. . Slaves on Horses: The Evolution of the Islamic Polity. Cambridge: Cambridge University Press. Crystal, Jill. . Oil and Politics in the Gulf: Rulers and Merchants in Kuwait and Qatar, Cambridge: Cambridge University Press. Cunningham, David E. . “Preventing Civil War: How the Potential for International Intervention Can Deter Conflict Onset,” World Politics (): –.
https://doi.org/10.1017/9781009367509.014 Published online by Cambridge University Press
References
Cust, James and Torfinn Harding. . “Institutions and the Location of Oil Exploration”, Journal of the European Economic Association, (): –. Dahl, Robert A. . Polyarchy: Participation and Opposition, New Haven, CT: Yale University Press. Dell, Melissa. . “The Persistent Effect of Peru’s Mining Mita,” Econometrica, (): –. Devlin, Robert. . Debt and Crisis in Latin America: The Supply Side of the Story, Princeton, NJ: Princeton University Press. Diamond, Jared. . Guns, Germs, and Steel: The Fates of Human Society, New York: W.W. Norton and Company. Dietrich, Simone. . “Donor Political Economies and the Pursuit of Aid Effectiveness,” International Organization, (): –. Dincecco, Mark. . State Capacity and Economic Development: Past and Present. New York: Cambridge University Press. Dincecco, Mark and Yuhua Wang. . “Violent Conflict and Political Development over the Long Run: China versus Europe,” Annual Review of Political Science, : –. Donner, Fred. . The Early Islamic Conquests, Princeton, NJ: Princeton University Press. Doyle, Michael and Nicolas Sambanis. . Making War and Peacebuilding: United Nations Peacekeeping Missions, Princeton, NJ: Princeton University Press. Downing, Brian. . The Military Revolution and Political Change: Origins of Democracy and Autocracy in Early Modern Europe. Princeton, NJ: Princeton University Press. Dreher, Axel and James Vreeland. . The Political Economy of the United Nations Security Council, New York: Cambridge University Press. Dunning, Thad. . “Conditiong the Effects of Aid: Cold War Politics, Donor Credibility, and Democracy in Africa,” International Organization, (): –. . Crude Democracy: Natural Resource Wealth and Political Regimes, Cambridge: Cambridge University Press. Eibl, Ferninand and Adeel Malik. . “The Politics of Partial Liberalization: Cronyism and Non-Tariff Protection in Mubarak’s Egypt,” CSAE Working Paper WPS/-. Eichengreen, Barry and David Leblang. . “Democracy and Globalization,” Economics and Politics, (): –. Elbadawi, Ibrahim and Samir Makdisi. . Democratic Transitions in the Arab World, New York: Cambridge University Press. Engerman, Stanley and Kenneth Sokoloff. . “Institutions, Factor Endowments, and the Paths of Development in the New World,” Journal of Economic Perspectives, (): –. Entelis, John P. . “Oil Wealth and the Prospects for Democratization in the Arabian Peninsula: The Case of Saudi Arabia,” in Naiem A. Sherbiny and Mark A. Tessler (eds.), Arab Oil: Impact on the Arab Countries and Global Implications, New York: Praeger, –. Fearon, James. . “Rationalist Explanations for War,” International Organization, (): –.
https://doi.org/10.1017/9781009367509.014 Published online by Cambridge University Press
References
Fearon, James and David Laitin. . “Ethnicity, Insurgency, and Civil War,” American Political Science Review, (): –. Fieldhouse, David K. . Western Imperialism in the Middle East, –, Oxford: Oxford University Press. Finer, Samuel E. . The History of Government: Volume II, The Intermediate Ages, Oxford: Oxford University Press. Fish, M. Steven. . “Islam and Authoritarianism,” World Politics, (): –. . Are Muslims Distinctive? A Look at the Evidence, Oxford: Oxford University Press. Fish, M. Steven and Robin S. Brooks. . “Does Diversity Hurt Democracy?” Journal of Democracy, (): –. Freeman, John R. and Dennis P. Quinn. . “The Economic Origins of Democracy Reconsidered,” American Political Science Review, (): –. Frieden, Jeffry R. . Debt, Development, and Democracy: Modern Political Economy and Latin America, –, Princeton, NJ: Princeton University Press. Gaidar, Yegor. . “The Soviet Collapse,” American Enterprise Institute, URL: www .aei.org/feature/the-soviet-collapse/, accessed March , . Gallup, John L., Jeffrey D. Sachs, and Andrew D. Mellinger. . “Geography and Economic Development,” International Regional Science Review, (): –. Gandhi, Jennifer. . Political Institutions under Dictatorship, New York: Cambridge University Press. Gandhi, Jennifer and Ellen Lust-Okar. . “Elections under Authoritarianism,” Annual Review of Political Science, : –. Gartzke, Eric. . “The Capitalist Peace,” American Journal of Political Science, (): –. Gartzke, Erik, Quan Li, and Charles Boehmer. . “Investing in the Peace: Economic Interdependence and International Conflict,” International Organization, (): –. Gassebner, Martin, Michael J. Lamla, and James R. Vreeland. . “Extreme bounds of democracy,” Journal of Conflict Resolution, (): –. Geddes, Barbara. . “What Do We Know about Democratization after Twenty Years?” Annual Review of Political Science, : –. Geddes, Barbara, Joseph Wright, and Eric Frantz. . “Autocratic Breakdown and Regime Transitions: A New Data Set,” Perspectives on Politics, (): –. Gehlbach, Scott. . Representation through Taxation: Revenue, Politics, and Development in Postcommunist States, Cambridge: Cambridge University Press. Gerner, Deborah J. . “Petro-Dollar Recycling: Imports, Arms, Investment and Aid,” Arab Studies Quarterly, –. Gerring, John, Philip Bond, William T. Barndt, and Carola Moreno. . “Democracy and Economic Growth: A Historical Perspective,” World Politics, (): –. Gerring, John, Daniel Ziblatt, Johan Van Gorp, and Julian Arevalo. . “An Institutional Theory of Direct and Indirect Rule,” World Politics, (): –. Gibb, Hamilton A. . “An Interpretation of Islamic History – II,” Muslim World, (): –. Girod, Desha and Meir R. Walters. . “Imperial Origins of the Oil Curse,” Journal of Arabian Studies, : –.
https://doi.org/10.1017/9781009367509.014 Published online by Cambridge University Press
References
Gledistch, Kristian S. and Michael D. Ward. . “Double Take: A Reexamination of Democracy and Autocracy in Modern Politics,” Journal of Conflict Resolution, (): –. Gleditsch, Nils Petter. . “Whither the Weather? Climate Change and Conflict,” Journal of Conflict Resolution, (): –. Glick, Thomas F. . Islamic and Christian Spain in the Early Middle Ages, Princeton University Press. Goldberg, Ellis, Erik Wibbels and Eric Mvukiyehe. . “Lessons from Strange Cases: Democracy, Development, and the Resource Curse in US States,” Comparative Political Studies, (–): –. Gowa, Joanne. . Allies, Adversaries, and International Trade, Princeton, NJ: Princeton University Press. Gowa, Joanne and Edward D. Mansfield. . “Power Politics and International Trade,” American Political Science Review, (): –. Grossman, Gene M. and Elhanan Helpman. . “Protection for Sale,” American Economic Review, (): –. Grossman, Herschel I. . “Foreign Aid and Insurrection,”Defense Economics (): –. Guardado, Jenny. . “Office Selling Corruption and Long-Term Development in Peru,” American Political Science Review, (): –. Gurr, Ted. . Why Men Rebel. Princeton, NJ: Princeton University Press. . People versus States: Minorities at Risk in the New Century. Washington, DC: US Institute of Peace Press. Gunitsky, Seva. . Aftershocks: Great Powers and Domestic Reforms in the Twentieth Century, Princeton, NJ: Princeton University Press. Guyer, Jane I. . “Representation without Taxation: An Essay on Democracy in Rural Nigeria, –,” African Studies Review, (): –. Haber, Stephen and Victor Menaldo. a. “Do Natural Resources Fuel Authoritarianism? A Reappraisal of the Resource Curse,” American Political Science Review, (): –. b. “Rainfall, Human Capital, and Democracy,” Stanford University and University of Washington. Habyarimana, James, Macartan Humphreys, Daniel N. Posner, and Jeremy M. Weinstein. . “Why Does Ethnic Diversity Undermine Public Goods Provision,” American Political Science Review, (); –. Haggard, Stephen and Robert R. Kaufman. . “Democratization During the Third Wave,” Annual Review of Political Science, : –. Hallwood, Paul and Stuart Sinclair. . Oil, Debt and Development: OPEC in the Third World, London: George Allen & Unwin Ltd. Harbom, Lotta, Eric Melander and Peter Wallensteen. . “Dyadic Dimensions of Armed Conflict, –,” Journal of Peace Research, (): –. Hariri, Jacob G. . “The Autocratic Legacy of Early Statehood,” American Political Science Review, (): –. . “A Contribution to the Understanding of Middle Eastern and Muslim Exceptionalism,” Journal of Politics, (): –. Heldt, Birger. . “Peacekeeping and Transitions to Democracy,” in Hanne Fjelder and Kristine Hoglund (eds.), Building Peace, Creating Conflict? Conflictual
https://doi.org/10.1017/9781009367509.014 Published online by Cambridge University Press
References
Dimensions of Local and International Peacekeeping, Lund: Nordic Academic Press, –. Herb, Michael. . “A Nation of Bureaucrats: Political Participation and Economic Diversification in Kuwait and the United Arab Emirates,” International Journal of Middle East Studies, (): –. Hodgson, Marshall. . The venture of Islam: Conscience and History in a World Civilization. Vol , Chicago: University of Chicago Press. Hodler, Roland. . “The Curse of Natural Resources in Fractionalized Countries,” European Economic Review, (): –. Hollyer, James R., B. Peter Rosendorff, and James Raymond Vreeland. . Information, Democracy, and Autocracy: Transparency and Political (In) Stability, New York: Cambridge University Press. Houbens, Vincent J. . “Southeast Asia and Islam,” Annals of the American Academy, (July): –. Huff, Toby E. . The Rise of Early Modern Science: Islam, China and the West, Cambridge: Cambridge University Press. Hunter, Shireen. . OPEC and the Third World: The Politics of Aid, London: Croom Helm. Huntington, Samuel P. . “Will More Countries Become Democratic?” Political Science Quarterly, –. . The Third Wave of Democratization: Democratization in the Late th Century, Norman, OK: Oklahoma University Press. Hunziker, Philipp and Lars-Erik Cederman. . “No Extraction without Representation: The Ethno-Regional Oil Curse and Secessionist Conflict,” Journal of Peace Research, (): –. Ivanhoe, L.F. . World Oil Supply: Production, Reserves, and EOR. Golden CO: M. King Hubert Center for Petroleum Supply Studies, Colorado School of Mines. Jablonski, Ryan. . “How Aid Targets Votes: The Effect of Electoral Strategies on the Distribution of Foreign Aid,” World Politics, (): –. Jackson, Peter. . The Delhi Sultanate: A Political and Military History, Cambridge: Cambridge University Press. . “The Mamluk institution in early Muslim India,” in Peter Jackson (ed.), Studies on the Mongol Empire and Early Muslim India, Farnham: Ashgate Variorum. –. Jamal, Amaney A. . Of Empires and Citizens: Pro-American Democracy or No Democracy at All? Princeton: Princeton University Press. Jensen, Nathan M., Glen Biglaiser, Quan Li, Edmund J. Malesky, Pablo Pinton and Santiago Pinto. . Politics and Foreign Direct Investment, Ann Arbor: University of Michigan Press. Jensen, Nathan and Leonard Wantchekon. . “Resource Wealth and Political Regimes in Africa,” Comparative Political Studies, (): –. Johns, Anthony H. . “The Role of Sufism in the Spread of Islam to Malaya and Indonesia,” Journal of Pakistan Historical Society, –. Kalyvas, Stathis. . “Civil Wars,” in Carlos Boix and Susan Stokes (eds.), The Oxford Handbook of Comparative Politics, Oxford: Oxford University Press, –.
https://doi.org/10.1017/9781009367509.014 Published online by Cambridge University Press
References
Kalyvas, Stathis and Laia Balcells. . “International System and Technologies of Rebellion: How the End of the Cold War Shaped Internal Conflict,” American Political Science Review, (): –. Karl, Terri L. . The Paradox of Plenty: Oil Booms and Petro-States, Berkeley: University of California Press. Karpat, Kemal H. . “Millets and Nationality: The Roots of the Incongruity of Nation and State in the Post-Ottoman Era” in Benjamin Braude and Bernard Lewis (eds.), Christians and Jews in the Ottoman Empire: The functioning of a plural society, New York: Holmes and Meier, –. Kedourie, Ellie. . Democracy and Arab Political Culture, Portland, OR: Frank Cass & Co. Kennedy, Edward S. . “The Arabic Heritage in the Exact Sciences,” Al-Abtath, : –. Kennedy, Hugh. . An Historical Atlas of Islam, Boston: Brill. . “The Decline and Fall of the First Muslim Empire”, Der Islam, : –. . The Great Arab Conquests: How the Spread of Islam Change the World We Live In, Philadelphia, PA: Da Capo Press. . “The City and the Nomad,” in Robert Irwin (ed.), New Cambridge History of Islam, Vol . Islamic Cultures and Societies to the End of the Eighteenth Century, Cambridge: Cambridge University Press, . Kepel, Gilles. . Jihad: The Trail of Political Islam (trans. A. F. Roberts), Cambridge, MA: Harvard University Press. Keynes, John M. . “The German Transfer Problem,” Economic Journal, (): –. Kohli, Atul. . Imperialism and the Developing World: How Britain and the United States Shaped the Global Periphery, New York: Oxford University Press. Kotkin, Stephen. . Armageddon Averted: Soviet Collapse, –, Oxford: Oxford University Press. Koubi, Vally, Gabriele Spilker, Tobias Boehmelt, and Thomas Bernauer. . “Do Natural Resources Matter for Interstate and Intrastate Armed Conflict?” Journal of Peace Research, (): –. Kurtz, Marcus J. and Sarah M. Brooks. . “Condition the Resource Curse: Globalization, Human Capital and Growth in Oil-Rich Nations,” Comparative Political Studies, (): –. Kuran, Timur. . The Long Divergence: How Islamic Law Held Back the Middle East, Princeton, NJ: Princeton University Press. . “Islam and Economic Performance: Historical and Contemporary Links,” Journal of Economic Literature, (): –. Kuru, Ahmet T. . Islam, Authoritarianism, and Development: A Global and Historical Comparison, Cambridge: Cambridge University Press. Kuziemko, Ilyana and Eric Werker. . “How Much Is a Seat on the Security Council Worth?” Journal of Political Economy, (): –. Lapidus, Ira. . A History of Islamic Societies, New York: Cambridge University Press. Lake, David A. and Matthew A. Baum. . “The Invisible Hand of Democracy: Political Control and the Provision of Public Services,” Comparative Political Studies, (): –.
https://doi.org/10.1017/9781009367509.014 Published online by Cambridge University Press
References
Lawrence, Adria. . Imperial Rule and the Politics of Nationalism: Anti-Colonial Protest in the French Empire, New York: Cambridge University Press. Le Billon, Philippe. . “The Political Ecology of War: Natural Resources and Armed Conflicts,” Political Geography, (): –. Lee, Melissa. . Crippling Leviathan: How Foreign Subversion Weakens the State, Ithaca, NY: Cornell University Press. Lei, Yu-Hsiang and Guy Michaels. . “Do Giant Oilfield Discoveries Fuel Internal Armed Conflict?” Journal of Development Economics, : –. Lenin, Vladimir I. . Imperialism, the Highest Form of Capitalism: A Popular Outline, Available: www.marxists.org/archive/lenin/works//imp-hsc/ Accessed: May . Levi, Margaret. . Of Rule and Revenue, Berkeley, CA: University of California Press. Levitsky, Steven and Lucan A. Way. . “The Rise of Competitive Authoritarianism,” Journal of Democracy, (): –. . Competitive Authoritarianism: Hybrid Regimes after the Cold War, New York: Cambridge University Press. Levtzion, Nehemia. . Ancient Ghana and Mali, London: Methuen and Company. . “The Sahara and the Sudan from the Arab conquest of the Maghrib to the rise of the Almoravids,” in J. D. Fage and Roland Oliver (eds.), The Cambridge History of Africa, Vol . BC to AD , Cambridge: Cambridge University Press, –. Lewis, Bernard. . “Islam and Liberal Democracy,” Atlantic Monthly, (): –. . “Islam and Liberal Democracy: A Historical Overview,” Journal of Democracy, (): –. Liberman, Peter. . Does Conquest Pay?: The Exploitation of Occupied Industrial Societies, Princeton, NJ: Princeton University Press. Lindsay, Michael. . Faith in the Halls of Power: How Evangelicals Joined the American Elite, New York: Oxford University Press. Lizzeri, Alessandro, and Nicola Persico. . “The Provision of Public Goods under Alternative Electoral Incentives,” American Economic Review, (): –. Lujala, Päivi. . “The Spoils of Nature: Armed Conflict and Rebel Access to Natural Resources,” Journal of Peace Research, (): –. Luong, Pauline J. and Erika Weinthal. . Oil Is Not a Curse: Ownership Structure and Institutions in Soviet Successor States, New York: Cambridge University Press. Lust, Ellen. . “Missing the Third Wave: Islam, Institutions, and Democracy in the Middle East,” Studies in Comparative International Development, (): –. Mahdavi, Paasha. . “Explaining the Oil Advantage: Effects of Natural Resource Wealth on Incumbent Reelection in Iran,” World Politics, (): –. . Power Grab: Political Survival Through Extractive Resource Nationalization, New York: Cambridge University Press. Mahdavy, Hussein. . “The Patterns and Problems of Economic Development in Rentier States: The Case of Iran,” in M. A. Cook (ed.), Studies in Economic History of the Middle East, London: Oxford University Press, –. Mann, Michael. . “The Autonomous Power of the State: Its Origins, Mechanisms, and Results,” European Journal of Sociology, (): –.
https://doi.org/10.1017/9781009367509.014 Published online by Cambridge University Press
References
Marer, Paul. . “The Political Economy of Soviet Relations with Eastern Europe,” Soviet Policy in Eastern Europe, –. Marshall, Monty G. and Keith Jaggers. . “Polity IV: Political Regime Characteristics and Transitions.” Masoud, Tarek. . Counting Islam: Religion, Class, and Elections in Egypt, New York: Cambridge University Press. Mazaheri, Nimah. . Hydrocarbon Citizens: How Oil Transformed People and Politics in the Middle East, Unpublished manuscript. Meltzer, Allan H. and Scott F. Richard. . “A Rational Theory of the Size of Government,” Journal of Political Economy, (): –. Menaldo, Victor. . The Institutions Curse: Natural Resources, Politics, and Development, New York: Cambridge University Press. Michalopoulos, Stelios, Alireza Naghavi, and Giovanni Prarolo. . “Trade and Geography in the Spread of Islam,” The Economic Journal, –. Miguel, Edward, Shanker Satyanath, and Ernest Sergenti. . “Economic Shocks and Civil Conflict: An Instrumental Variables Approach,” Journal of Political Economy, (): –. Milner, Helen V. and Keiko Kubota. . “Why the Move to Free Trade? Democracy and Trade Policy in the Developing Countries,” International Organization, (): –. Milner, Helen V. and Bumba Mukherjee. . “Democratization and Economic Globalization,” Annual Review of Political Science, :–. Monroe, Stephen L. . “Varieties of Protectionism: Ethnic Politics and Resistance to Neoliberalism in the Arab World,” PhD Dissertation, Princeton University. Montalvo, Jose G. and Marta Reynal-Querol. . “Ethnic Polarization, Potential Conflict, and Civil Wars,” American Economic Review, (): –. Moore, Barrington. . Social Origins of Dictatorship and Democracy: Lord and Peasant in the Making of the Modern World, Boston, MA: Beacon Press. Morby, John. . Dynasties of the World: A Chronological and Genealogical Handbook, New York: Oxford University Press. Morony, Michael J. . Iraq after the Muslim conquest, Princeton, NJ: Princeton University Press. Morrison, Kevin M. . Nontaxation and Representation: The Fiscal Foundations of Political Stability, Cambridge: Cambridge University Press. Murphy, John F. Jr. . Sword of Islam. Amherst, NY: Prometheus Books. Neumayer, Eric. . “Arab-Related Bilateral and Multilateral Sources of Development Financing: Issues, Trends, and the Way Forward,” United Nations University: World Institute for Development Economics Research Discussion Paper No. /. . “What Factors Determine the Allocation of Aid by Arab Countries and Multilateral Agencies?” Journal of Development Studies, (): –. Newey, Whitney K. . “Efficient Estimation of Limited Dependent Variable Models with Endogenous Explanatory Variables,” Journal of Econometrics, (): –. Nielsen, Richard, Michael Findley, Zach Davis, Tara Candland, and Daniel Nielsen. . “Foreign Aid Shocks as a Cause of Violent Armed Conflict,” American Journal of Political Science, (): –.
https://doi.org/10.1017/9781009367509.014 Published online by Cambridge University Press
References
Nordvik, Frode Martin. . “Does Oil Promote or Prevent Coups? The Answer is Yes,” The Economic Journal, (): –. North, Douglass. . “Institutions,” Journal of Economic Perspectives, (): –. North, Douglass, John Wallis and Barry Weingast. . Violence and Social Orders: A Conceptual Framework for Interpreting Recorded Human History, New York: Cambridge University Press. Nunn, Nathan. . “The Importance of History for Economic Development,” Annual Review of Economics, : –. Nunn, Nathan and Nancy Qian. . “US Food Aid and Civil Conflict,” American Economic Review, (): –. Nunn, Nathan and Leonard Wantchekon. . “The Slave Trade and the Origins of Mistrust in Africa,” American Economic Review, (): –. Nüssli, Christos. . “Euratlas.” www.euratlas.com/about.html. Ober, Josiah. . The Rise and Fall of Classical Greece, Princeton: Princeton University Press. O’Callaghan, Joseph F. . A History of Medieval Spain, Ithaca: Cornell University Press. Okey, Robin. . The Demise of Communist East Europe: in Context, London: Arnold. Olson, Mancur. . “Dictatorship, Democracy, and Development,” American Political Science Review, (); –. Olsson, Ola. . “On the Democratic Legacy of Colonialism,” Journal of Comparative Economics, : –. Østby, Gudrun, Ragnhild Nordå and Jan Ketil Rød. . “Regional Inequalities and Civil Conflict in Sub-Saharan Africa,” International Studies Quarterly, (): –. Oto-Peralías, Daniel and Diego Romero-Ávila. . “The Economic Consequences of the Spanish Reconquet: The Long-Term Effects of Medieval Conquest and Colonization,” Journal of Economic Growth, : –. Ottaway, Marina. . Democracy Challenged: The Rise of Semi-Authoritarianism, Washington, DC: Carnegie Endowment for International Peace. Pepinsky, Thomas. . Economic Crises and the Breakdown of Authoritarian Regimes: Indonesia and Malaysia in Comparative Perspective, New York: Cambridge University Press. Pevehouse, Jon C. . Democracy from Above? Regional Organizations and Democratization, New York: Cambridge University Press. Pinto, Pablo. . Partisan Investment in the Global Economy: Why the Left Loves Foreign Direct Investment and FDI Loves the Left, New York: Cambridge University Press. Pipes, Daniel. . Slave Soldiers and Islam: The Genesis of a Military System, New Haven: Yale University Press. Polanyi, Karl. . The Great Transformation: The Political and Economic Origins of Our Time, Boston, MA: Beacon Press. Powell, Robert. . “The Inefficient Use of Power: Costly Conflict with Complete Information,” American Political Science Review (): –. . “War as a Commitment Problem,” International Organization, (): –. Przeworski, Adam. . “The ‘East’ Becomes the ‘South?’ The ‘Autumn of the People’ and the Future of Eastern Europe,” PS: Political Science & Politics, (): –.
https://doi.org/10.1017/9781009367509.014 Published online by Cambridge University Press
References
Przeworski, Adam, Michael E. Alvarez, Jose Antonio Cheibub, and Fernando Limongi. . Democracy and Development: Political Institutions and Well-Being in the World, –, New York: Cambridge University Press. Putnam, Robert. . Making Democracy Work: Civic Traditions in Modern Italy, Princeton: Princeton University Press. Putterman, Louis. . “State Antiquity Index (Version ),” Brown University. Rajan, Raghuram G. and Arvind Subramanian. . “Aid and Growth: What Does the Cross-Country Evidence Really Show?” Review of Economics and Statistics, (): –. Ramsay, Kris. . “Revisiting the Resource Curse: Natural Disasters, the Price of Oil, and Democracy,” International Organization, (): –. Regan, Patrick M. and Richard W. Frank. . “Migrant Remittances and the Onset of Civil War,” Conflict Management and Peace Science, –. Remmer, Karen L. . “The Process of Democratization in Latin America,” Studies in Comparative International Development, (): –. Reynolds, D. B., and M. Kolodziej. . “Former Soviet Union Oil Production and GDP Decline: Granger Causality and the Multi-cycle Hubbert Curve,” Energy Economics, (): –. Ricci, Matteo. . Della entrata della Compagnia di Giesu e Christianita nella Cina, vols. i. . Riner, Deborah. . Borrowers and Bankers: The Euro-Market and Political Economy in Peru and Chile, PhD dissertation, Princeton Department of Politics. Rodrik, Dani, Arvind Subramanian, and Francesco Trebbi. . “Institutions Rule: The Primacy of Institutions over Geography and Integration in Economic Development,” Journal of Economic Growth, (): –. Roeder, Philip G. . “Ethnolinguistic Fractionalization (ELF) Indices, and ,” Available from http//:weber.ucsd.edu/$\sim$proeder/elf.htm. Accessed: December . Ross, Michael L. . “Does Oil Hinder Democracy?” World Politics, (April): –. . “But Seriously: Does Oil Really Hinder Democracy,” Unpublished paper, UCLA. Available: https://leitner.yale.edu/sites/default/files/files/resources/papers/ ButSeriously.pdf. . The Oil Curse: How Petroleum Wealth Shapes the Development of Nations, Princeton, NJ: Princeton University Press. Rowley, Charles K. and Nathanael Smith. . “Islam’s Democracy Paradox: Muslims Claim to Like Democracy, So Why Do They Have So Little?” Public Choice, : –. Rubin, Jared. . Rulers, Religion, and Riches: Why the West Got Rich and the Middle East Did Not, New York: Cambridge University Press. Savun, Burcu and Daniel C. Tirone. . “Exogenous Shocks, Foreign Aid, and Civil War,” International Organization, (Summer): –. Schedler, Andreas. . Electoral Authoritarianism: The Dynamics of Unfree Competition, Boulder, CO: Lynne Rienner Publishers. Scheve, Kenneth F. and David Stasavage. . Taxing the Rich: A History of Fiscal Fairness in the United States and Europe, Princeton: Princeton University Press. Schumpeter, Joseph. . Capitalism, Socialism, and Democracy, New York: Harper.
https://doi.org/10.1017/9781009367509.014 Published online by Cambridge University Press
References
Schwarz, Rolf. . “The Political Economy of State-Formation in the Arab Middle East: Rentier States, Economic Reform, and Democratization,” Review of International Political Economy, –. Shaw, Stanford J. . “Some Aspects of the Aims and Achievements of the Nineteenth Century Ottoman Reformers,” in W. Polk and R. Chambers (eds.), Beginnings of Modernization in the Middle East, Chicago: University of Chicago Press, . . “The Nineteenth-Century Ottoman Tax Reforms and Revenue System,” International Journal of Middle East Studies, (): –. Simons, Geoff. . Libya: The Struggle for Survival, New York: St. Martin’s Press. Smith, Adam. . The Wealth of Nations, London: W. Strahan and T. Cordello. Smith, Alastair. . “The Perils of Unearned Income,” Journal of Politics, (): –. Smith, Benjamin. . Hard Times in the Lands of Plenty: Oil Politics in Iran and Indonesia, Ithaca, NY: Cornell University Press. Smith, Benjamin and David Waldner. . Rethinking the Resource Curse, New York: Cambridge University Press. Snyder, Richard. . “Does Lootable Wealth Breed Disorder? A Political Economy of Extraction Framework,” Comparative Political Studies, (): –. Soifer, Hillel. . “The Causal Logic of Critical Junctures,” Comparative Political Studies, (): –. Staiger, Douglas and James H. Stock. . “Instrumental Variable Regression with Weak Instruments,” Econometrica, (): –. Stasavage, David. . “When Distance Mattered: Geographic Scale and the Development of European Representative Assemblies,” American Political Science Review, (): –. . The Rise of Western Democracy: Why it Happened in Europe and Not China or the Middle East, Princeton, NJ: Princeton University Press. Steinert, Janna I. and Sonja Grimm. . “Too Good to Be True? United Nations Peacebuilding and the Democratization of War-Torn States,” Conflict Management and Peace Studies, (): –. Strayer, Joseph. . On the Medieval Origins of the Modern State, Princeton: Princeton University Press. Svensson, Jacob. . “Foreign aid and rent-seeking,” Journal of International Economics, (): –. Svolik, Milan. . “Authoritarian Reversals and Democratic Consolidation,” American Political Science Review, (): –. . The Politics of Authoritarian Rule, New York: Cambridge University Press. Tessler, Mark. . “Islam and Democracy in the Middle East: The Impact of Religious Orientations on Attitudes Towards Democracy in Four Arab Countries,” Comparative Politics, (April): –. Thompson, Elizabeth F. . How the West Stole Democracy from the Arabs: The Syrian Arab Congress and the Destruction of Its Historic Liberal-Islamic Alliance, New York: Atlantic Monthly Press. Tilly, Charles. . Coercion, Capital, and European States, AD –, London: Basil Blackwell. Tir, Jaroslav and Johannes Karreth. . Incentivizing Peace: How International Organizations Can Help Prevent Civil War in Member Countries, New York: Oxford University Press.
https://doi.org/10.1017/9781009367509.014 Published online by Cambridge University Press
References
Tomz, Mike. . Reputation and International Cooperation: Sovereign Debt across Three Centuries, Princeton, NJ: Princeton University Press. Tordo, Silvana, Brandon S. Tracy, and Noora Arfaa. . “National Oil Companies and Value Creation,” World Bank Working Paper . Tornell, Aaron and Philip R. Lane. . “The Voracity Effect,” American Economic Review, (): –. Tsugitaka, Sato. . State and Rural Society in Medieval Islam: Sultans, Muqta’s and Fallahun, Leiden, Netherlands: Brill. Tsui, Kevin K. . “More Oil, Less Democracy? Evidence from Worldwide Crude Oil Discoveries,” Economic Journal, (): –. Tullock, Gordon. . “The Welfare Costs of Tariffs, Monopolies, and Theft,” Western Economic Journal, (): –. Ulfelder, Jay. . “Natural Resource Wealth and the Survival of Autocracies,” Comparative Political Studies, (): –. von Grunebaum, Gustave E. (). Classical Islam: A History, AD to AD, New Brunswick, NJ: Transaction. Wantchekon, Leonard. . “Why Do Resource Dependent Countries Have Authoritarian Governments?” African Finance and Economic Development, (): –. Weber, Max. []. The Sociology of Religion. Translated from German by Ephraim Fischoff, Boston, MA: Beacon Press. Weingast, Barry. . “The Political Foundations of Democracy and the Rule of Law,” American Political Science Review, (): –. Werker, Eric. . “The Political Economy of Bilateral Foreign Aid,” in Gerard Caprio (ed.), Handbook of Safeguarding Global Financial Stability: Political, Social, Cultural, and Economic Theories and Models, London: Academic Press, (–). Werker, Eric, Faisal Z. Ahmed and Charles Cohen. . “How Is Aid Spent? Evidence from a Natural Experiment,” American Economic Journal – Macroeconomics, (): –. Wharton Econometrics. . “Centrally Planned Economics Outlook,” Washington, DC: Wharton Econometric Forecasting Associates, March . Wickham, Chris. . “The Uniqueness of the East,” Journal of Peasant Studies, (/): –. Wimmer, Andreas. . “Is Diversity Detrimental? Ethnic Fractionalization, Public Goods Provision, and the Historical Legacies of Stateness,” Comparative Political Studies, ():–. Winters, Matthew and Joseph Wright. . “The Politics of Effective Foreign Aid,” Annual Review of Political Science (June): –. Wright, Joseph . “How Foreign Aid Can Foster Democratization in Authoritarian Regimes,” American Journal of Political Science, (): –. Woodberry, Robert D. . “The Missionary Roots of Liberal Democracy,” American Political Science Review, (): –. Yergin, Daniel. . The Prize: The Epic Quest for Oil, Money and Power, New York: Free Press. Yetiv, Steve A. and Keterina Oskarsson. . Challenged Hegemony: The United States, China, and Russia in the Persian Gulf, Stanford, CA: Stanford University Press.
https://doi.org/10.1017/9781009367509.014 Published online by Cambridge University Press
Index
Abbasid Caliphate, –, – Abd al-Rahman III (Caliph), Abramson, Scott, absolutist rule in China, – in Europe, – Abu Bakr (Caliph), accumulated democracy, – democratic deficit and, military conquest reducing, religiosity and, – ACD. See Armed Conflict Database Acemoglu, Daron, , –, –, , on democratic transitions, on income and democracy, Afghanistan, Soviet invasion of, agricultural land. See iqta agriculture. See also iqta autocratization and, – democracy development and, –, Haber and Menaldo on, institutional development and, land quality and conquest, Ahmadov, Anar, Ahmed, Faisal Z., –, –, Ajnadayn, – Alesina, Alberto, , –, – Alfonso IX (King), – Arab Spring, , , Arab-Israeli conflict of , Argentina, democratization in, Arjomand, Said Amir, –
Armed Conflict Database (ACD), – asabiyyah (solidarity), assassinations, authoritarian resilience, – foreign aid linked to, – oil wealth linked to, authoritarianism, –. See also dictatorship in conquest societies, – in Latin America, – Svolik on, – autocracy. See also dictatorship after military conquest, – persistence of centralized, – autocratization, , agriculture and, – contemporary explanations for, – counterfactual countries in study of, – evaluation of competing historical explanations for, mamluk institutional presence and, – matched counterfactuals to, military conquest and, , –, – after Muslim expansion, –, – results of study on, – robustness of study on, – trade routes and, Bacon, Francis, Barkey, Karen, – Baron, Rueben, Barre, Said, – Bates, Robert H.,
https://doi.org/10.1017/9781009367509.015 Published online by Cambridge University Press
Battle of Poitiers, Bauer, Vincent, Beblawi, Hazem, Benavente, – Bermeo, Sarah, – Besley, Timothy, –, –, –, on cohesive institutions, on political violence, , Betz, Tim, Bin Laden, Osama, – Blaydes, Lisa, , , Boix, Carles, , , – on democratization, Borneo, Boschini, Anne, Bosworth, Edmund, –, Bray, Julia, Bucharest Formula, – Bueno de Mesquita, Bruce, Buwayhid, – Byzantine Empire, – Cahen, Claude, caliph, Cammett, Melani, capitalism, crony, case vignettes, – centralization. See state centralization cereal suitability, Chaney, Eric, , , , , , – Chaudhry, Kiran A., Chile, – democratization in, China, – absolutist rule in, – civil liberties, defining, civil war, –, causal estimates of foreign aid connection to, – in conquest petrostates, democracy following, , dictatorship and, , , , economic conditions linked to, – external support for, – foreign aid linked to, , greed-based explanations for, – groups in, mamluk institutional history and, in Muslim societies, in non-Muslim societies, , in non-oil producers, , ,
Index in oil producers, – oil wealth linked to, remittances and, – rents influencing, – transitions to, in theoretical model, – underdevelopment linked to, CMEA. See Council for Mutual Economic Assistance Cohen, Charles, –, – cohesive institutions, , –, Besley on, defining, , degree of cohesion in, democracy distinguished from, democracy linked to, , –, – exogenous, – in Latin America, – military conquest and, – in Muslim societies, – Neolithic determinants of, in , – in non-Muslim societies, – in non-oil producers, – Persson on, public goods in, rent levels in, rents and, – state resources in, theoretical predictions, Cold War, –, foreign aid recipients during, – Colgan, Jeff, Collier, Paul, – comparative democratization, conditional resource curse, Confucianism, – conquest. See military conquest, Muslim conquest petrostates, , civil war in, democracy in, conquest societies authoritarianism in, – percentages of conquered territory, rents in, , continental axis hypothesis, Cook, Scott J., Council for Mutual Economic Assistance (CMEA), – credit cycle, critical junctures, – Crone, Patricia, crony capitalism,
https://doi.org/10.1017/9781009367509.015 Published online by Cambridge University Press
Index cross-sectional variation, – Cunningham, David, –, Damascus, DD design. See differences-in-differences design de Gaulle, Charles, de-escalation, Delhi, – democracy. See also cohesive institutions accumulated, –, – Acemoglu on income and, agriculture and development of, –, alternative measures of, causal estimates of oil wealth and, – civil war preceding, , cohesive institutions distinguished from, cohesive institutions linked to, , –, – in conquest petrostates, contemporary explanations for development of, – criteria for, crude, deficits of, –, –, dictatorship as residual category for, in Eastern Europe, economic development linked to, European colonialism and, – European colonialism and development of, – foreign aid and, Islam and, –, –, – mamluk institutional legacy and, – in MENA, military conquest and contemporary, , , military conquest harming, – in Muslim societies, , , , – in non-Muslim societies, , in oil producers, –, , – oil wealth and, , , –, –, precolonial state development and, – pre-conquest trade and development of, – public goods and, religiosity and, – rents and, – social transfers in, – taxation and, , transitions to, – in theoretical model, – welfare spending and,
democratic transition, – in Nigeria, democratic transitions Acemoglu and Robinson theory of, foreign transfers and, – in non-Muslim societies, – in oil producers, – in Venezuela, – democratization in Argentina, Boix on, in Chile, comparative, Huntington on, Third Wave of, , –, Devlin, Robert, Diamond, Jared, continental axis hypothesis of, dictatorship, backsliding to, challenges of, – civil war and, , , , control in, defining, – entrenchment of, – foreign aid to, – military conquest and, – in Muslim societies, in non-oil producers, in oil producers, political conflict correlated with, political violence linked to, – power-sharing in, rents linked to, –, , –, as residual category of democracy, Ross on, – Smith, B., on, – Svolik on, underdevelopment linked to, Waldner on, – differences-in-differences (DD) design, –, , , –, – causal interpretation in, Dincecco, Mark, , Donner, Fred, donor intent, – Dunning, Thad, Easterly, William, Eastern Europe, – cohesive institutions in, – democracy in,
https://doi.org/10.1017/9781009367509.015 Published online by Cambridge University Press
Eastern Europe (cont.) Soviet Union relations with, – trade subsidies in, – economic conditions civil war linked to, – in Muslim societies, in non-Muslim societies, economic development democracy linked to, rents and, – Egypt, , , – mamluks in, Eibl, Ferdinand, – Elbaawi, Ibrahim, ELF. See ethno-linguistic fractionalization endogeneity bias, Engerman, Stanley, – Entelis, John P., – ethno-linguistic fractionalization (ELF), – eunuchs, Europe, – absolutist rule in, – political survival in, European colonialism, –, –, , –, – in conquest equilibrium, – democracy and, – impact of, on development of democracy, – indirect rule under, Muslim military conquest and, exclusion restriction evaluation of, – violations of, – executive constraints index, expenditure-switching strategy, extremism, Islamic, – fallahun (peasants), Fearon, James, Ferdinand III (King), – Finer, Samuel, , –, , – fiscal capacity, – military conquest and, Fish, Steven, foreign aid, , authoritarian resilience linked to, – causal estimates of civil war connection to, – civil war linked to, , Cold War recipients of, – democracy and,
Index to dictatorships, – donor intent, – exclusion restriction evaluated, – geopolitical rents and, – mamluk institutional recipients of, – measurement of, , – to non-oil producers, –, , oil prices and, in Persian Gulf, political stability bought by, – political violence patterns and, –, –, – to Somalia, foreign interventions, state centralization and, foreign transfers, , –, – calculations of, democratic transitions and, – exogenous decline in, – to Latin America, – political violence linked to decline in, – France, free resources, Freedom House, , Frieden, Jeffry, Gabon, Gaidar, Yegor, Gassebner, Martin, Genghis Khan, – geopolitical rents, –, – external support for civil wars, – foreign aid and, – Gerring, John, , –, Ghaznavid, – glasnost, – Glick, Thomas, Gorbachev, Mikhail, – government transfers, military conquest and, Great Britain, greed-based explanations, for civil war, – group formation effect, – Haber, Stephen, – on agriculture, Hallwood, Stuart, Hariri, Jacob, , , –, – on Middle Eastern exceptionalism, – Hodges, Marshall, Hodler, Roland, Hoeffler, Anke, – Hollenbach, Florian,
https://doi.org/10.1017/9781009367509.015 Published online by Cambridge University Press
Index Hollyer, James, Huang, Ryuku, – Hunter, Shireen, Huntington, Samuel, –, – on democratization, Hussein, Saddam, – Ibn Abi Zayd, – identification strategy, – incumbent advantage, – India mamluks in, – military conquest in, indigenous American societies, – indigenous state centralization, – Indonesia, industrializers, institutional development agriculture and, since , – institutions curse, – instrumental variables analysis, – international politics, international trade, expansion of Islam and, iqta (agricultural land), , –, efficacy of, historical introduction of, – mamluks and, Iranian revolution, Iraq, leader duration in, – Islam, , – democracy and, –, –, – international trade and expansion of, in MENA, military conquest spread of, – in non-conquest territories, – role of, in military conquest, –, – in Southeast Asia, – spread of, in SSA, – territories exposed to, by , – Islamic charitable trust. See wafq Islamic extremism, – Islamic Golden Age, Islamic law, , jagirs, Jamal, Amaney, James I (King), – janissary, Jerusalem, Johnson, Simon, –, Jordan, –
Kabul, Kalyvas, Stathis, Karpat, Kemal H., Kennedy, Hugh, –, –, –, , – Kenny, David, Kepel, Gilles, – Keynes, John Maynard, Khomeini (Ayatollah), – Kolodziej, Marek, Kotkin, Stephen, Kuran, Timur, , , , Kuru, Ahmet, , Laitin, David, Lamla, Michael, land grant system. See timar system land quality, military conquest and, Lane, Philip, Lapidus, Ira, –, Latin America, – authoritarianism in, – cohesive institutions in, – foreign transfers to, – oil prices in, Lawrence, Adria, , leader duration in Iraq, – in military conquest, Lebanon, , – Lei, Yu-Hsiang, – Levi, Margaret, , Lewis, Bernard, , – Libya, Lien, Da-Hsiang Donald, Lipset, Seymour, Lujala, Päivi, Machiavelli, Mahdavy, Hussein, Makdisi, Samir, Malaysia, Mali Kingdom, Malik, Adeel, – Malmud II (Sultan), mamluks (slave soldiers), , , , , , , autocratization and presence of, – civil war in countries with mamluk institutional history, cultures importing, – democracy and mamluk institutions, –
https://doi.org/10.1017/9781009367509.015 Published online by Cambridge University Press
mamluks (slave soldiers) (cont.) efficacy of institutions, – in Egypt, foreign aid received by countries with mamluk institutional history, – in India, – iqta and, loyalty of, in Muslim expansion, political equilibrium of, – political institutions, – recruitment of, – as repressive agents, – social position of, in Spain, , in SSA, – state centralization affected by, – Matanouk, Alia, Mayan civilization, – M’ba, Leon, median voter theorem, – medieval period parliaments in, – political survival in, , – MENA. See Middle East and North Africa Menaldo, Victor, , –, , , – on agriculture, Mexican debt crisis, – Michaels, Guy, – Michalopoulos, Stelio, , , – Middle East and North Africa (MENA), , –, , , –. See also specific countries democracy in, Islam in, Middle Eastern exceptionalism, , – Hariri on, – military conquest, alternate measures of, – autocratization and, – fiscal capacity and, legacy of, – military conquest, Muslim, –, –. See also conquest societies autocracy after, – autocratization and, , – cohesive institutions and, – conceptual clarification of, – contemporary explanations for democracy development, – decomposition of, – democracy and, , ,
Index democracy harmed by, – dictatorship resultant from, – effects of, traced, – equilibrium tracing, – European colonialism and, evaluation of competing theories on, – expansion via, – foundations of, – government transfers and, historical legacy of, – history of, identification strategy in tracing, – in India, institutional legacy of, –, institutional trajectory launched by, Islam role in, –, – Islam spread through, – land quality and, leader duration in, measurement of, – military conquest reducing, parallel trends in, – parliament development and, – percentages of conquered territory, political and religious elites imposed after, – political institutions in, – political survival and, – political violence and history of, – pre-conquest trade, – premodern explanations of, – of Spain, – start date of, – state centralization in, , – in Sub-Saharan Africa, territories exposed to, by , – undoing, – Miller, Michael, , – millet system, Ming dynasty, modernization effect, Mongol rule, – Moorish institutions, Morby, John, –, Morocco, Mubarak, Hosni, – Muhammad (Prophet), –, –, –, –, Muhammad XII (Caliph), – muqtas, Muslim Brotherhood,
https://doi.org/10.1017/9781009367509.015 Published online by Cambridge University Press
Index Muslim expansion, . See also military conquest, Muslim autocratization after, –, – mamluks institutions in, Muslim recipient states, – Muslim societies civil war in, cohesive institutions in, – democracy in, , , , – dictatorship in, economic well-being in, non-Muslim societies contrasted with, political stability in, – political violence in, –, –, , – POLITY scores of, socio-economic conditions in, standards of living in, underdevelopment in, – Mutszeski, Janina, Naghavi, Alireza, National Movement, natural resources, – Neolithic Revolution, –, –, cohesive institutions determined by, neo-Mamluks, Neumayer, Eric, –, – Newey-West procedure, Nigeria, , democratic transition in, non-conquest territories, Islam in, – nondemocracy. See dictatorship non-Muslim societies civil war in, , cohesive institutions in, – democracy in, , democratic transitions in, – economic well-being in, Muslim societies contrasted with, POLITY scores of, non-negative rule, – non-oil producers, civil war in, , , cohesive institutions in, – dictatorship in, foreign aid to, –, , political stability in, – rents in, – Nunn, Nathan, – Nüssli, Christos,
Obasanjo, Olusegun, official development aid (ODA), – OIC. See Organization of Islamic Cooperation oil policy, US, – oil prices, foreign aid and, in Latin America, political violence with changes in, oil producers civil war in, – democracy in, –, , – democratic transitions in, – dictatorship in, political stability of, –, – state-ownership of oil production, oil wealth, authoritarian resilience linked to, causal estimates of democracy and, – Chaudhry on, civil war linked to, democracy and, , , –, –, domestic peace and, – measurement of, political regimes and, – political stability and, –, –, – timing of, Olsson, Ola, – on Western colonies, OPEC. See Organization of Petroleum Exporting Countries Organization of Islamic Cooperation (OIC), Organization of Petroleum Exporting Countries (OPEC), – Oto-Peralías, Daniel, Ottoman Empire, –, –, out-of-region damage, – Pakistan, – Palestinian Liberation Organization, parliaments in medieval period, – military conquest and development of, – peace oil wealth and, – preservation of, in theoretical model, troops engendering, – peasants. See fallahun perestroika, –
https://doi.org/10.1017/9781009367509.015 Published online by Cambridge University Press
Persian Gulf, foreign aid in, Persson, Torsten, –, –, –, on cohesive institutions, on political violence, , Peru, – Pettersson, Jan, Pillars of Prosperity (Besley and Persson), – Pinochet, Augusto, Pipes, Daniel, – Platas, Melinas, political centralization. See state centralization political institutions, , . See also cohesive institutions cohesion of, –, – historical origins of, – mamluk, – in military conquest, – variation in, – political regimes oil wealth and, – rents and, – political resource curse, – political rights, defining, political stability foreign aid buying, – in Muslim societies, – in non-oil producers, – of oil producers, –, – oil wealth and, –, –, – rents and, –, political survival in Europe, in medieval period, , – military conquest and, – state centralization and, – political transfer problem, – political transitions, rents and, – political violence, –, –, Besley on, , corroborative statistical evidence on, – cross-national causal evidence on, – dictatorship linked to, – foreign aid and patterns of, –, –, – foreign transfer decline linked to, – history of military conquest and, – in Muslim societies, –, –, , –
Index with oil price changes, optimal investment in, – patterns of, – Persson on, , rents and, , – in state capacity theory, , – in theoretical model, – underdevelopment and, Polity index, – POLITY IV project, – POLITY scores, , –, –, , – of Muslim societies, of non-Muslim societies, post-colonial era, power-sharing, in dictatorship, Prarolo, Giovanni, precolonial state development, development of democracy and, – pre-conquest trade, development of democracy and, – PRIO/Armed Conflict Database, – private rents, – Przeworski, Adam, – public goods in cohesive institutions, democracy and, Puigcerdá, – Punto Fijo Pact, – Putnam, Robert, al-Qaddafi, Muammar, , – Al Qaeda, Qarakhanid, – quasi-natural experiment setting, – Quran, Ramsay, Kris, , –, – Reconquista, reduced form approach, remittances, civil war and, – Remmer, Karen, rentier effect, mechanisms of, – rentier states, Beblawi on, Mahdavy on, rents, –, –. See also oil wealth civil war influenced by, – cohesive institutions and, – conditional theories of, –
https://doi.org/10.1017/9781009367509.015 Published online by Cambridge University Press
Index in conquest societies, , defining, , –, – democracy and, – destabilizing effects of, dictatorship linked to, –, , –, economic development and, – geopolitical, –, – governance and, – heterodox views on, – levels of, in cohesive institutions, in non-oil producers, – outcomes of declines in, political economy of, – political regimes and, – political stability and, –, political transitions and, – political violence and, , – private, – Ross on, – Smith, A., on, sources of, – stabilizing effects of, theoretical puzzle of, – variation in, –, rent-seeking, – repression effect, repressive capacity, mamluks as, – Reynolds, Douglas, Ricci, Matteo, ridda wars, Riner, Deborah, Robinson, James, , –, –, on democratic transitions, Roeder, Philip, – Roine, Jesper, Romero-Ávila, Diego, Rosato, Sebastian, , – Rosendorff, Peter, Ross, Michael, –, on dictatorship and rents, – Rowley, Charles K., Rubin, Jared, , , Sahara, Sasanian Empire, – Saudi Arabia, – Scheve, Kenneth, Schwab, Daniel, , –, Seljuk, –, – Sharia law, –, Shaw, Stanford, Siad Barre, Mohamed, –
Sinclair, Paul, slave soldiers. See mamluks Smith, Alastair, – on rents, Smith, Benjamin, , , on dictatorship, – social transfers, in democracies, – Soifer, Hillel, Sokoloff, Kenneth, – solidarity. See asabiyyah Somalia, –, – Southeast Asia, Islam in, – sovereign borrowing, – Soviet Union, –, , – Afghanistan invaded by, Eastern Europe relations with, – Spain, –, – assemblies in, first parliament in, – mamluks in, , military conquest of, – reconquest of, –, spending effect, – SSA. See Sub-Saharan Africa Staiger, Douglas, standards of living, in Muslim societies, Stasavage, David, state capacity theory, political violence in, , – state centralization, challenges to causal interference, – components of, evaluation of competing historical explanations for, foreign interventions and, indigenous, – mamluks affecting, – matched counterfactuals to, in military conquest, , – political survival and, – pre-conquest, – pre-modern state, –, robustness of, – synthetic controls in study on, within-country level of, state prize, , –, state resources in cohesive institutions, defining, – sharing of, – Stock, James H., subnational transfers, –
https://doi.org/10.1017/9781009367509.015 Published online by Cambridge University Press
Index
Sub-Saharan Africa (SSA). See also specific countries Islam in, – mamluks in, – military conquest in, Sufis, – Sunni Muslims, Svensson, Jacob, Svolik, Milan, on authoritarianism, – on dictatorship, Sykes-Picot Agreement, Syria, –, ,
ulema, –, –, –, as legitimizing agents, – Umar (Caliph), –, – Umayyad dynasty, , umma, underdevelopment civil war linked to, dictatorship linked to, in Muslim societies, – political violence and, United States (US), – foreign oil policy, – troop deployments, –
T’ang Empire, – Tanzimat, – tarikahs, – taxation, – bargaining process in, democracy and, , taxation effect, – temporal variation, – Tessler, Mark, – theoretical model basic setup of, – empirical evaluation of, – political violence in, – preservation of peace in, proofs for, – transition to democracy or civil war in, – Thompson, Elizabeth, , Tilly, Charles, –, –, – timar system (land grants), – Tornell, Aaron, trade routes, autocratization and, trade subsidies, in Eastern Europe, – transatlantic trade, – troop deployments, US, – Tsugitaka, Sato, Tullock, Gordon, – Turkic rule, two-stage least (SLS) regressions, –, –, ,
Velasco Alvarado, Juan, – Venezuela, –, voracity effect, Vreeland, James, , wafq (Islamic charitable trust), Wahhabism, – Waldner, David, on dictatorship, – Wang, Yuhua, Wantchekon, Leonard, – War on Terror, Weinstein, Jeremy, welfare spending, democracy and, Werker, Eric, , –, –, –, , West Africa, Western colonies, Why Nations Fail (Acemoglu and Robinson), – Wickham, Chris, – World Bank, –, – World Bank Development Indicators, – Yergin, Daniel, – Yom Kippur War, . See Arab-Israeli conflict of
https://doi.org/10.1017/9781009367509.015 Published online by Cambridge University Press