Cambridge Core Constraining Dictatorship 9nov2022

C O N S T R A IN I N G D I C T A T O R S H I P
How do some dictatorships become institutionalized ruled-based systems,

while others remain heavily personalist? Once implemented, do executive
constraints actually play an effective role in promoting autocratic stability?
To understand patterns of regime institutionalization, this book studies the
emergence of constitutional term limits and succession procedures, as well as
elite power-sharing within presidential cabinets. Anne Meng argues that
institutions credibly constrain leaders only when they change the underlying
distribution of power between leaders and elites by providing elites with
access to the state. She also shows that initially weak leaders who
institutionalize are less likely to face coup attempts and are able to remain in
office for longer periods than weak leaders who do not. Drawing on an
original time-series dataset of 46 countries in Sub-Saharan Africa from 1960
to 2010, formal theory, and case studies, this book ultimately illustrates how
some dictatorships evolve from personalist strongman rule to institutionalized
regimes.
Anne Meng is Assistant Professor in the Department of Politics, University of

Virginia. Her research centers on authoritarian politics and institutions.
Professor Meng’s work has been published in the British Journal of Political
Science, The Journal of Theoretical Politics, Columbia Law Review, and
Studies in Comparative International Development, and has won the “Best
Paper Award” from the Democracy and Autocracy section at APSA.
Published online by Cambridge University Press

po l i t i c a l e c o no m y o f in s t i t u tio n s an d d e ci s i o n s
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 and Joseph P. Ferrie, Southern Paternalism and the Rise of the
American Welfare State: Economics, Politics, and Institutions, 1865–1965
Lee J. Alston, Thrainn Eggertsson and Douglass C. North, eds., Empirical Studies
in Institutional Change
James E. Alt and Kenneth Shepsle, eds., Perspectives on Positive Political
Economy
Josephine T. Andrews, When Majorities Fail: The Russian Parliament, 1990–1993
Jeffrey S. Banks and Eric A. Hanushek, eds., Modern Political Economy: Old
Topics, New Directions
Yoram Barzel, Economic Analysis of Property Rights, 2nd 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
Common-Pool Resources in Long-Term Perspective
Adam Dean, From Conflict to Coalition: Profit-Sharing Institutions and the
Political Economy of Trade
Mark Dincecco, Political Transformations and Public Finances: Europe,
1650–1913
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, 1876–1929
Ron Harris, Industrializing English Law: Entrepreneurship and Business
Organization, 1720–1844
Anna L. Harvey, Votes Without Leverage: Women in American Electoral Politics,
1920–1970

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, and 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?
Chrysostomos 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,
1970–2000
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, 1700–1860
Michael L. Ross, Timber Booms and Institutional Breakdown in Southeast Asia
Shanker Satyanath, Globalization, Politics, and Financial Turmoil: Asia’s
Banking Crisis
Norman Schofield, Architects of Political Change: Constitutional Quandaries and
Social Choice Theory
Norman Schofield and Itai Sened, Multiparty Democracy: Elections and
Legislative Politics
Alberto Simpser, Why Governments and Parties Manipulate Elections: Theory,
Practice, and Implications
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, 1688–1789
Charles Stewart III, Budget Reform Politics: The Design of the Appropriations
Process in the House of Representatives, 1865–1921
George Tsebelis and Jeannette Money, Bicameralism
Georg Vanberg, The Politics of Constitutional Review in Germany
Nicolas van de Walle, African Economies and the Politics of Permanent Crisis,
1979–1999
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

CONSTRAINING
DICTATORSHIP
From Personalized Rule to Institutionalized Regimes
ANNE MENG
University of Virginia

University Printing House, Cambridge cb2 8bs, United Kingdom
One Liberty Plaza, 20th Floor, New York, ny 10006, USA
477 Williamstown Road, Port Melbourne, vic 3207, Australia
314–321, 3rd Floor, Plot 3, Splendor Forum, Jasola District Centre,
New Delhi – 110025, India
79 Anson Road, #06–04/06, Singapore 079906
Cambridge University Press is part of the University of Cambridge.

It furthers the University’s mission by disseminating knowledge in the pursuit of
education, learning, and research at the highest international levels of excellence.
www.cambridge.org
Information on this title: www.cambridge.org/9781108834896
doi: 10.1017/9781108877497
© Anne Meng 2020
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.
First published 2020
Printed in the United Kingdom by TJ International Ltd. Padstow Cornwall
A catalogue record for this publication is available from the British Library.
isbn 978-1-108-83489-6 Hardback
isbn 978-1-108-79247-9 Paperback
Cambridge University Press 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.


regimes.


Series Editors
Founding Editors

and the Economy
Economy
Scope of the State
Power
Decision Making

Electoral System
1650–1913
African Society
Policy in Africa
South America
Medieval Trade
Accountability
Mexico, 1876–1929
1920–1970

Politics in France
Government
Performance
Collective Action
1970–2000

Banking Crisis
1979–1999

CONSTRAINING
DICTATORSHIP
ANNE MENG


www.cambridge.org
doi: 10.1017/9781108877497
© Anne Meng 2020
isbn 978-1-108-83489-6 Hardback


regimes.


Series Editors
Founding Editors

and the Economy
Economy
Scope of the State
Power
Decision Making

Electoral System
1650–1913
African Society
Policy in Africa
South America
Medieval Trade
Accountability
Mexico, 1876–1929
1920–1970

Politics in France
Government
Performance
Collective Action
1970–2000

Banking Crisis
1979–1999

CONSTRAINING
DICTATORSHIP
ANNE MENG


www.cambridge.org
doi: 10.1017/9781108877497
© Anne Meng 2020
isbn 978-1-108-83489-6 Hardback


regimes.


Series Editors
Founding Editors

and the Economy
Economy
Scope of the State
Power
Decision Making

Electoral System
1650–1913
African Society
Policy in Africa
South America
Medieval Trade
Accountability
Mexico, 1876–1929
1920–1970

Politics in France
Government
Performance
Collective Action
1970–2000

Banking Crisis
1979–1999

CONSTRAINING
DICTATORSHIP
ANNE MENG


www.cambridge.org
doi: 10.1017/9781108877497
© Anne Meng 2020
isbn 978-1-108-83489-6 Hardback

Contents
List of Figures page xii

List of Tables xiv
Acknowledgments xvi
List of Abbreviations xix
1 Introduction 1
1.1 Two Puzzles About Authoritarian Institutions 1
1.2 Examining Regime Outcomes: Personalist Rule and
Institutionalized Systems 5
1.3 Sometimes Window Panes, But Often Window
Dressing 9
1.4 Autocratic Regime Institutionalization: A Deeper
Look 12
1.5 Causes of Regime Institutionalization 16
1.6 Consequences of Regime Institutionalization 19
1.7 This Book’s Contributions: Authoritarian Regimes 23
1.8 Why Africa? This Book’s Contributions to African
Politics 27
1.9 Plan of the Book 31
2 Why Do Leaders Institutionalize? 34
2.1 Institutions as Commitment Devices 36
2.2 A Theoretical Model 41
2.3 Solving the Baseline Model 44
2.4 Model Extension: Uncertainty About the Future 50
2.5 Conclusion 54
3 Two Illustrative Cases 64
3.1 The Formal Model Summarized 66
3.2 Cameroon Under Ahidjo: High Levels of Regime
Institutionalization 69
ix
Contents
3.3 Cô te D’Ivoire Under Houphouët: Low Levels of Regime

3.4 Empirical Analysis of Power Sharing in Presidential
Cabinets 89
3.5 Conclusion 91
4 How Should Institutionalization Be Measured? 93
4.1 Conceptualizing Authoritarian Regime
4.2 Measuring Regime Institutionalization 101
4.3 Executive Constraints in Sub-Saharan Africa:
General Trends and Patterns 111
4.4 Regime Typologies 116
4.5 Comparisons with Other Existing Datasets 123
4.6 Conclusion 126
5 What Are the Causes of Regime Institutionalization? 134
5.1 Observable Implications of the Theoretical
Argument 135
5.2 Justifying the Model’s Assumptions 136
5.3 Cross-sectional Design: Data and Measurement 140
5.4 Cross-sectional Design: Empirical Strategy and
Findings 147
5.5 Differences-in-differences Design: The End of the Cold
War 157
5.6 Conclusion 161
6 What Are the Consequences of Institutionalization
on Autocratic Stability? 186
6.1 Are Constitutional Rules Actually Followed In
Dictatorships? 187
6.2 Empirical Disagreement: Do Institutions Promote
Regime Stability? 189
6.3 Problems with Existing Approaches 190
6.4 Estimating the Effects of Endogenous
6.5 Empirical Analysis and Findings 195
6.6 Conclusion 199
7 What Are the Consequences of Institutionalization
on Leadership Succession? 205
7.1 Do Dictators Care About Leadership Succession? 207
7.2 How Does Succession Planning Promote Regime
Stability? 209
x
Contents
7.3 Designating Informal Successors 212

7.4 Disaggregating Autocratic Constitutions: Succession
Policies Versus Term Limits 213
7.5 Endogenous Institutionalization Reconsidered 214
7.6 Data and Descriptive Statistics 216
7.7 Descriptive Statistics 218
7.8 Empirical Analysis 219
7.9 Conclusion 221
8 Conclusion 226
8.1 Taking Stock: Authoritarian Institutions
Reconsidered 226
8.2 Empirical Contribution: Reexamining Regime
8.3 Theoretical Contribution: The Origins and
Consequences of Regime Institutionalization 232
8.4 Looking Ahead: Implications for Institutional Design,
Democratization, and Democratic Backsliding 237
References 239
Index 249
xi
Figures
1.1 Number of failed and successful coups in autocracies page 8

1.2 Number of violent and peaceful leadership transitions in
autocracies 8
1.3 Proportion of autocratic countries with nominally
democratic institutions 10
1.4 Number of years ruling parties remain in power past
the founding leader 11
1.5 Proportion of autocratic countries in sub-Saharan Africa
with executive constraints 15
1.6 The effect of regime institutionalization on leader
and regime outcomes 22
2.1 Game tree: baseline model 44
2.2 Graphing equilibrium results 48
2.3 Distribution of p2 51
2.4 Game tree: model extension 52
3.1 Yearly gross domestic product (GDP) of Cameroon
and Côte d’Ivoire 65
3.2 Ethnic favoritism in president cabinets 91
4.1.a Cross-sectional differences in institutionalization 112
4.1.b Cross-sectional differences in institutionalization 113
4.2 Over time changes in institutionalization 115
4.3 Comparison between institutionalization and XCONST
variable 124
4.4 Comparison between institutionalization and V-Dem
measure of legislative constraint 125
4.5 Comparison between institutionalization and V-Dem
measure of judicial constraint 126
5.1 Founding leaders versus subsequent leaders 149
5.2 Strong nationalist leaders versus presidents allied
with outgoing colonial powers 150
xii
List of Figures
5.3 Coup leaders versus noncoup leaders 152

5.4 Rebel group leaders versus nonrebel group leaders 157
5.5 Were leaders with access to oil revenues less likely to
institutionalize after the end of the Cold War? 160
6.1 What is the effect of institutionalization on regime stability? 191
6.2 What is the effect of institutionalization on leader tenure
(founders)? 196
6.3 What is the effect of institutionalization on leader tenure
(coup leaders)? 197
6.4 What is the effect of institutionalization on the frequency
of coup attempts? 198
6.5 What is the effect of institutionalization on coup
vulnerability? 199
7.1 What is the effect of regime institutionalization
on leadership succession? 215
7.2 Marginal effects of having a succession policy
and designated successor on likelihood of peaceful
transition 220
xiii
Tables
Appendix 3.1 Summary statistics page 92

4.1 Previous positions of autocratic leaders
(conditional on peaceful leader entry) 108
4.2 Previous positions of autocratic leaders
(conditional on irregular leader entry) 109
4.3 Correlation matrix of seven dimensions of
institutionalization 116
4.4 Comparison between institutionalization
and regime type 121
Appendix 4.1 List of countries and leaders included in this
study 127
Appendix 4.2 Summary of institutionalization scores by
country 132
5.1 How long into their tenure do leaders
create formal executive constraints? 137
5.2 Do constitutional rules persist, once created? 139
Appendix 5.1 Founding leaders 163
Appendix 5.2 Strong nationalist leaders 165
Appendix 5.3 Strong nationalist leaders controlling for
participation in war 167
Appendix 5.4 Coup leaders 169
Appendix 5.5 Coup leaders (Cold War sample) 171
Appendix 5.6 Coup leaders (post-1991 sample) 173
Appendix 5.7 Variation within coup leaders 175
Appendix 5.8 High-ranking coup leaders only 176
Appendix 5.9 Rebel groups 178
Appendix 5.10 Rebel groups (Cold War sample) 180
Appendix 5.11 Rebel groups (post-1991 sample) 182
Appendix 5.12 Were autocratic regimes less institutionalized
during the Cold War? 184
xiv
List of Tables
Appendix 5.13 Differences-in-differences test 185

6.1 Summary statistics 188
Appendix 6.1 Summary statistics: length of leader tenure 200
Appendix 6.2 Summary statistics: frequency of coups 200
Appendix 6.3 What is the effect of institutionalization
on leader tenure (leader strength proxy:
founder versus successor)? 201
Appendix 6.4 What is the effect of institutionalization on
leader tenure (leader strength proxy: coup
leaders versus noncoup leaders)? 202
Appendix 6.5 What is the effect of institutionalization on coup
vulnerability (leader strength proxy: founder versus
successors)? 203
Appendix 6.6 What is the effect of institutionalization on coup
vulnerability (leader strength proxy: coup
leaders versus noncoup leaders)? 204
Appendix 7.1 Summary statistics 222
Appendix 7.2 Does succession planning increase the likelihood
of peaceful leadership transitions? 222
Appendix 7.3 Global sample 223
Appendix 7.4 Robustness check: only units with
constitutions 224
Appendix 7.5 Robustness check: time trends 225
xv
Acknowledgments
This project could not have come together without the support and
wisdom of a wonderful academic community. I spent eleven formative
years at UC Berkeley – both as an undergraduate and as a Ph.D. student.
I am grateful that even as a college student (one out of over 30,000
students) my professors cultivated my research interests, pushed me to
think harder and more critically, and encouraged me to pursue academia.
I took my first game theory class with Robert Powell in my junior year of
college, and it completely reoriented the ways in which I thought about
strategic interaction in politics. A seminar on Chinese politics with
Kevin O’Brien sparked my desire to understand durable forms of
authoritarianism, and a writing seminar with Steve Fish motivated me to
always write with the goal of publishing. In my junior year of college,
I begged Peter Lorentzen to let me into his Ph.D. seminar on political
economy – he did, and the exposure to graduate training and research
solidified my desire to pursue a Ph.D. Early on in graduate school, I took
a seminar with Leonardo Arriola, and his sharp insights paired with an
infectious enthusiasm to pursue new and exciting research embodied the
kind of scholar I wanted to be. These seemingly small moments early on
made lasting impacts on me, and I often think fondly of my time spent at
Berkeley.
I want to express special thanks to my dissertation cochairs, Leonardo
Arriola and Robert Powell. They truly were a dream-team of advisors. Leo
and Bob poked holes in my arguments, asked difficult questions (that
sometimes took months to adequately address), and pushed me to be the
best scholar I could be. At the same time, they championed my research,
offered their expertise when I was stuck, and provided invaluable advice
each step of the way. I am so grateful for their friendship and guidance.
Much of who I am as a scholar today has been shaped by their mentoring.
I would also like to thank other faculty members and my peers at
Berkeley who generously offered their time, insights, moral support, and
xvi
Acknowledgments
friendship: Andrew Bertoli, Tara Buss, Christopher Chambers-Ju,

Pradeep Chhibber, David Collier, Ruth Collier, Laurel Eckhouse,
Frederico Finan, Sean Gailmard, Stephen Goggin, Nina Hagel, Ron
Hassner, Ryan Hubert, Jason Klocek, Peter Lorentzen, Aila Matanock,
Michaela Mattes, Katherine Michel, Joel Middleton, Jack Paine, Brian
Palmer-Rubin, Janna King Rezaee, Eric Schickler, Fiona Shen-Bayh, Jason
Wittenberg, and Noam Yuchtman.
Although many of the ideas presented in this book first took root while
I was at Berkeley, this manuscript was written while I was an assistant
professor at the University of Virginia and finalized while I was on
sabbatical at New York University and the University of Chicago. My
colleagues at the University of Virginia (UVA) have been welcoming and
supportive since day one, and I would like to thank Lawrie Balfour, Dale
Copeland, Paul Freedman, Dan Gingerich, Murad Idris, Gabi Kruks-
Wisner, David Leblang, Carol Mershon, Sid Milkis, Phil Potter, Rachel
Augustine Potter, Jen Rubenstein, Todd Sechser, David Waldner, Denise
Walsh, and Mila Versteeg for their feedback, mentoring, and friendship.
I had an excellent team of graduate and undergraduate research assistants
for this project – I thank Simonas Cepenas, Ben Helms, Sarah Genovese,
Trent Lefkowitz, Natalie Lohr, Ben Stievater, Chen Wang, Hannah
Waxelbaum, and Roland Wuu for their work.
I would also like to extend my thanks to many other scholars outside of
UVA who helped me along the way. In particular, Jennifer Gandhi, Scott
Gehlbach, Rachel Beatty Riedl, and Milan Svolik read the entire
manuscript and participated in a full-day book workshop. Their
feedback and guidance were immensely helpful in shaping and revising
the project. Monika Nalepa was my writing buddy when I was revising
this manuscript, and I am grateful for her friendship and mentoring. I also
thank Anna Grzymala-Busse, Gretchen Helmke, Ellen Lust, Eddy
Malesky, Gwyneth McClendon, John Patty, Tom Pepinsky, Dan Slater,
and Nic van de Walle for providing helpful feedback and encouragement
on this project.
When I became an assistant professor, I found an amazing community
of fellow assistant professors. It is a joy to celebrate publications and
promotions together and also reassuring to know that we are not alone
in weathering challenges. I want to thank Jack Paine for his friendship and
unadulterated enthusiasm for research, which has not waned since we
were in grad school. In the comparative authoritarianism community,
I thank Quintin Beazer, Brett Carter, Erin Baggott Carter, Charles
Crabtree, Florian Hollenbach, Holger Kern, Andrew Little, Alex Lee,
Ora John Reuter, David Szakonyi, Emily Sellars, and Scott Tyson for
making conferences so much fun. I am also grateful for my friends in the
xvii
Acknowledgments
African politics community, including Jaimie Bleck, Ruth Carlitz, Mai

Hassan, Lauren Honig, Kathleen Klaus, Eric Kramon, Nathan Noah, Jeff
Paller, and Liz Sperber. These friendships I’ve developed are one of the
best parts of this job.
Outside of my academic family, I am incredibly grateful for my loving
and loyal group of friends. I would like to thank Kevin Lin, Kate Poirer,
Peggy Wong, and Candice Yu for being my oldest friends and for always
giving me a home in San Francisco and New York City. When I moved to
Charlottesville, I had the good fortune of forming a tight-knit group of
friends who quickly helped me make Virginia my new home. I would like
to thank Zach Bethune, Christine Bethune, Ambar LaForgia, Adam Leive,
Lauren Reichman, Sebastian Tello-Trillo, and Katelyn Wood for their
camaraderie.
I started doing comedy while I was in graduate school, and life has not
been the same since. I thank my friends in the improv community –
particularly at Endgames Improv in San Francisco, Bent Theatre in
Charlottesville, Coalition Theater in Richmond, and the Upright
Citizens Brigade in New York City – as well my friends in the
Richmond, Charlottesville, and New York stand-up scene. Thanks for
making me laugh harder than I knew I could and for creating weird and
crazy content with me (inevitably in the basement of some theater).
Finally, I want to express my deep gratitude for Jamie Douglas, who
whole-heartedly supports my every endeavor with love and enthusiasm,
from sitting in the front row of all my comedy shows to designing the
beautiful cover of this book. He champions me every day, and I am lucky
to be surrounded with such wonderful loved ones. Without them, my life
would be much less joyful.
xviii
Abbreviations
BDC Bloc Democratique Camerouanis

CCM Chama Cha Mapindui
CCP Chinese Communist Party
CNU Cameroonian National Union
FRELIMO Mozambique Liberation Front
KANU Kenya African National Union
MPLA Moviemento Popular de Libertacao de Angola
PDCI Parti Démocratique de la Côte d’Ivoire
PDG Parti democratique de Guinee
PPN Nigerian Progressive Party
PRI Partido Revolucionario Institucional
RENAMO Mozambican National Resistance
SAA Syndicat Agricole Africain
TANU Tanganyika African National Union
UC Union Camerounaise
UNIP United National Independence Party
UNITA National Union for the Total Independence of Angola
UPC Union des Populations du Cameroun
xix
1
Introduction
“The simplest political system is that which depends on one individual.

It is also the least stable.”
Samuel Huntington, Political Order in Changing Societies (1968, 18)
1 . 1 t w o p uzzl e s a bo u t au t hori t ari an

i n s t it u t i o n s
As newly independent states, Tanzania and Guinea seemed to be on the
same trajectory of durable authoritarian rule. Tanzania, under the found-
ing presidency of Julius Nyerere, was a single-party state, led by the
Tanganyika African National Union (TANU) party. TANU politicians
filled the National Assembly, which met on a regular basis. Presidential
and legislative elections have been held every five years, like clockwork
since 1965, as stated in the constitution.
This stability has lasted for decades. Nyerere, the first postindepen-
dence leader, retired after the 1985 presidential elections. Power was
swiftly handed over to Nyerere’s handpicked successor, Ali Hassan
Mwinyi, and the ruling party continued to control over the government.
The same regime remains in power today. The ruling TANU/CCM party
continues to dominate Tanzanian politics1 – even surviving the introduc-
tion of multiparty elections in 1995. Since independence, the country has
undergone four peaceful leadership transitions and is one of the longest
reigning autocracies in Africa.
Like Tanzania, Guinea had a ruling party, legislature, and regularly
held elections as a newly independent state. Under the founding presi-
dency of Ahmed Sekou Toure, Guinea was a single-party regime, led by
1
TANU was renamed Chama Cha Mapindui (CCM) in 1977, following the merger of
Tanzania and Zanzibar, although the party remained largely the same.
1
https://doi.org/10.1017/9781108877497.001 Published online by Cambridge University Press
1 Introduction
the Parti democratique de Guinee (PDG) party. PDG politicians filled the
National Assembly, which conducted two regular sessions every year.
Presidential elections were held regularly in 1961, 1968, 1974, and
1982, and National Assembly elections were held in 1963, 1968, 1974,
and 1980, as stated in the constitution.
Yet these institutions did not provide long-term stability in Guinea. In
1984, Sekou Toure died of a heart attack after being airlifted to Cleveland,
Ohio for emergency heart surgery while on a trip to Saudi Arabia. Before
succession plans could be finalized, the military seized power in a coup
d’état and the leader of the coup, Colonel Lansana Conte, claimed the
presidency. The PDG was immediately disbanded, the National Assembly
was dissolved, and the constitution was abolished. In short, the regime
died with its leader.
Why did regime outcomes in Tanzania and Guinea diverge so drasti-
cally? Why was the authoritarian system in Guinea unable to survive the
death of the leader, even with a full set of nominally democratic institu-
tions in place? Sekou Toure had a ruling party, a legislature, and regularly
held elections. He was even a socialist who aimed to replicate the Soviet
state. Nonetheless, the regime fell in Guinea, and these institutions them-
selves were swiftly wiped out after the death of the leader.
These vignettes raise the first puzzle of the book: what explains differ-
ences in authoritarian regime outcomes, if not differences in quasi-
democratic institutions?
One possible consideration is that we need to look beyond the most
common types of quasi-democratic institutions – such as parties and
legislatures, which are quite prevalent across authoritarian regimes –
and consider more subtle forms of variation. Indeed, Guinea and
Tanzania did differ in one important institution: adoption of executive
constraints.
Since independence, presidents in Tanzania have had a number of
institutional constraints on their authority. During the tenure of the
founding president Nyerere, term limits and detailed leadership
succession procedures became enshrined in the constitution, and these
rules remain in place today. According to the constitution, presidents are
limited to two terms in office (Article 40), and in the case of the president’s
death or incapacitation, the vice president is the designated successor
(Article 37). The presidential cabinet, which is filled with TANU party
elites, exists as a genuine power-sharing organization rather than a hallow
endorsement device. Since independence, all presidents have maintained
fully functional cabinets, and all key cabinet positions, such as the vice
presidency and defense ministry, have been appointed to elites on
a regular basis. Critically, appointments for the position of vice president
2
1.1 Two Puzzles About Authoritarian Institutions
are infrequently shuffled, which endorses an elite in this position as a clear

and stable successor in accordance with constitutional rules regarding
succession.
By contrast, Sekou Toure faced very few constraints on his presidential
authority. The constitution of Guinea did not specify clear succession
procedures, promoting the image that Sekou Toure was an irreplaceable
leader, nor did it include term limits.2 Within the presidential cabinet,
Sekou Toure maintained clear dominance. The roles of prime minister and
defense minister were eliminated over half the time he was in office. When
a defense minister was appointed, elites who filled this position were
shuffled frequently. In fact, the average tenure of this appointment
under Sekou Toure was under three years.
Moreover, although the regime in Guinea under Sekou Toure had
a ruling party, legislature, elections, and a constitution, these institutions
did not function to tie the leader’s hands. In fact, Sekou Toure exploited
these institutions to amplify his own power. The ruling party, the PDG,
was used primarily as a mouthpiece to promote the leader’s own ideology
and policies, rather than as a forum for elite power sharing (Adamolekun
1976; Camara 2005). Although the National Assembly of Guinea met
regularly twice a year, its only formal function was to endorse legislation
and budgetary requests that were put forth by Sekou Toure. As described
by Jackson and Rosberg (1982): “Most laws originate simply and swiftly
in the decrees and edicts of the ruler. As the supreme authority in the land,
not only do his opinions prevail over all others, but they ‘become laws as
they are uttered’” (212). In this case, institutions clearly did not constrain
the leader.
In sum, although Tanzania and Guinea had similar looking parties and
legislatures on the surface, these two cases had very different patterns of
executive constraints that shaped the power of the president.
This raises a second puzzle of the book: why do some authoritarian
rulers adopt executive constraints while others do not?
This book will offer new insights on both of these puzzles. The primary
thesis of this study is that autocratic regime institutionalization – the
creation of rules and procedures that tie the leader’s hands by empowering
other elites – is key to understanding patterns of regime durability in
dictatorships. Concrete examples of such measures include constitutional
rules specifying the leadership succession order or term limits, in addition
to the appointment of elites to high-ranking cabinet positions, such as the
vice presidency. I will demonstrate that these institutions provide explicit
2
In fact, the constitution stated explicitly that the president may be reelected without
mention of term restrictions.
3
1 Introduction
constraints on executive power by providing high-level state access to

other elites, therefore empowering them with resources and their own
independent influence.
This argument stands in contrast with the conventional wisdom that
nominally democratic institutions, such as parties, legislatures, or elec-
tions, drive authoritarian stability. A key assumption in much existing
scholarship is that these institutions generally have the organizational
capacity to constrain leaders and facilitate elite power sharing. As this
chapter will show, parties and legislatures have become extremely com-
monplace in dictatorships, yet most are organizationally weak and overly
reliant on the influence of particular leaders. Rather than assuming that
the existence of parties or legislatures can effectively constrain leaders, this
book examines the creation of explicit executive constraints within con-
stitutions and power-sharing appointments in presidential cabinets. While
it is certainly true that not all autocratic institutions are merely instances
of window dressing, it is also important to recognize that not all institu-
tions successfully constrain leaders.
Importantly, this book addresses the key question of how certain types
of institutions constrain leaders. After all, a leader who can create an
institution can also disassemble it as well. How do institutions have any
bite in dictatorships? I argue that institutions can credibly constrain
leaders only when they change the underlying distribution of power
between leaders and elites. When an elite is appointed to a key cabinet
position, such as vice president or the minister of defense, he is given
access to power and resources that allow him to consolidate his own base
of support. The appointment of elites to these key cabinet positions
creates a focal point around these individuals and identifies them as
credible challengers to the incumbent. Over time, the delegation of
authority shifts the distribution of power away from the incumbent by
identifying alternative leaders that elites can rally around if the president
were to renege on distributive promises. Institutions that empower and
identify specific challengers help to solve elite coordination problems,
therefore better allowing them to hold incumbents accountable.
Institutionalization limits executive power by creating conditions that
actually threaten the leader.
My theory underscores the point that the existence of a democratic
façade is not of primary importance. Rather, institutions constrain when
they change the underlying distribution of power within the ruling coali-
tion. When a leader institutionalizes the regime, she hands the (figurative)
sword to someone else while pointing it at herself. This helps to explain
why nominally democratic institutions cannot necessarily explain why
some regimes are institutionalized systems while others remain personalist
4
1.2 Examining Regime Outcomes
dictatorships. This is especially true when parties or legislatures are empty

vehicles that simply amplify the authority of an incumbent, rather than
constraining them.3 Institutions matter, not because they establish de jure
rules, but when they affect de facto political power.
The theme of how political order becomes established and institutio-
nalized has long been a fundamental question in the study of comparative
politics. In a seminal study, Huntington (1968) first emphasized the con-
cept of political institutionalization, arguing that the strength of societies
depends on the strength of political organizations and procedures.
Organizational durability depends on the extent to which the functions
and procedures of these groups become institutionalized – the process by
which the institutions themselves acquire “value” and “stability.”
Importantly, he highlights the need to separate institutions from leaders:
so long as an organization still has its first set of leaders, so long as a procedure is
still performed by those who first performed it, its adaptability is still in doubt. The
more often the organization has surmounted the problem of peaceful succession
and replaced one set of leaders by another, the more highly institutionalized it is.
(Huntington 1968, 14)
This book approaches Huntington’s “organization” as the authoritar-

ian regime itself, and examines how executive power in dictatorships can
become institutionalized, such that the regime does not depend on any
particular set of leaders to survive. How does an authoritarian regime
evolve from a government run by “big men” to a system run by rules?
1.2 e xamining regime outcomes: personalist

r u le a n d i n s ti t u t i o n al i z e d s y s t em s
The world of authoritarian regimes varies considerably in the extent to
which politics is governed by set rules, procedures, and norms or con-
trolled by a single personalist strongman.4 Consider, for example, the
well-known cases of highly institutionalized dictatorship in twentieth-
century China and Mexico. The People’s Republic of China, which was
established in 1949 by Mao Zedong and the Communist Party of China
(CCP), is characterized as a hierarchical system with established norms
3
I do not claim that all parties and legislatures are window-dressing institutions that
do not constrain leaders. Some autocracies, such as in China, the former Soviet
Union, or Mexico under the PRI, have well-organized parties and legislatures that
do not merely rubber-stamp legislation. However, in many autocracies, these institu-
tions are incredibly weak and do not serve to empower specific elites.
4
This book uses the terms “dictatorship,” “authoritarian regime,” and “autocracy”
synonymously.
5
1 Introduction
and procedures that govern leadership promotion. The state constitution

is considered the highest law – its authority stands above the leader and
the ruling party. Since 1949, the regime has undergone four peaceful
leadership transitions and remains in power today.
Mexico under the rule of the Partido Revolucionario Institucional
(PRI) can be characterized similarly as a regime run by established rules
and norms. Stable authoritarian rule was established in Mexico in the
years following the end of the Mexican Revolution. In 1929, President
Plutarco Elías Calles founded the ruling party as a means of institutiona-
lizing elite power sharing that had been established in the resolution of the
revolution.5 Under these agreements, presidents serve one six-year term in
office and never seek reelection. The incumbent also handpicks their
successor, who then becomes the next president. Elite politics in Mexico
would run like clockwork according to these rules for the next 70 years.
The PRI regime remained in power until 2000, when it lost the presiden-
tial election to an opposition party, the National Action Party (PAN).
In both of these cases, the regime lasted beyond the dictatorship of
a single individual to become a system run by rules. Importantly, the
process of leadership succession was routinized, allowing for the conti-
nuity of the regime beyond the founding leader. Yet, such stable outcomes
are not always the story in the world of dictatorships.
Now consider the Democratic Republic of Congo under the dictator-
ship of Mobutu Sese Seko or the Dominican Republic under Rafel
Trujillo – regimes where a highly personalist leader ruled without con-
straints on his power. Mobutu seized power in the Democratic Republic
of Congo through a coup, five years after independence was granted in
1960. During his rule, Mobutu centralized power around himself, rather
than sharing it with other elites. He named himself the head of all impor-
tant political institutions including the minister of defense and single-
handedly decided all appointments and promotions within the regime,
often purging elites at will. Mobutu remained in office for 28 years until he
was deposed by Laurent-Désiré Kabila during the First Congo War in
1997.
The Trujillo regime in the Dominican Republic was a similar story.
Rafael Trujillo came to power through a coup in 1930. Upon taking office,
he concentrated his personal authority by declaring martial law and kill-
ing regime opponents.6 By the end of his rule, Trujillo had more public
5
The party was initially called the Partido Nacional Revolucionario (PNR). It was
eventually renamed the PRI in 1946.
6
It is estimated that 500,000 people were killed by the regime’s secret police during the
Trujillo era.
6
1.2 Examining Regime Outcomes
statues of himself on display in the Dominican Republic than any other

world leader at the time. After 31 years in power, Trujillo was assassi-
nated in 1961. Three years later, a democratically elected leader took
office but was deposed in a coup four months later.
In both of these cases, the regime failed to survive past a single strong-
man leader. Yet it is important to remember that single leaders can some-
times remain in office for relatively long periods of time, and this perceived
longevity speaks nothing to the institutional quality of the regime.
Mobutu and Trujillo both remained in power for three decades. During
those periods, the regime retained a façade of stability through the leaders’
iron-tight personalist grip on power. However, as Huntington cautioned,
this “simple political system” that depended on one individual was, in
reality, the least stable form of autocracy. The regime simply could not
survive past its founding leader.
These broad patterns extend beyond a few individual cases. Despite
a surge in scholarship on authoritarian stability, the world of dictator-
ships is filled with Mobutus and Trujillos. Leaders often take power (and
lose power) via coups, which occur with tremendous frequency. In fact,
coup d’états make up the vast majority of nonconstitutional exits from
office for dictators (Singh 2014; Svolik 2012). From 1950 to 2014,
a total of 235 failed coup attempts and 236 successful coups were carried
out in dictatorships. In 1963 and 1966, 12 successful coups were carried
out in a single year. In 1991, 10 coups were attempted but failed, in
addition to four successful coups that were carried out (Powell and
Thyne 2011). Figure 1.1 displays the number of failed and successful
coups that have been carried out in dictatorships between 1950 and
2014.
Beyond persistent coup threats to incumbents, leadership transitions
are often violent and disruptive, and many regimes fail to survive past the
departure of individual leaders. Figure 1.2 displays the number of peaceful
and violent leadership transitions over time. From this graph it is easy to
see that violent leadership transitions are extremely common. From 1946
to 2008, almost half (44 percent) of all autocratic leadership transitions
did not occur peacefully.7 Even when dictators manage to remain in office
until voluntarily retirement or a natural death, elites often wait in the
wings, eager to usurp power forcefully, as in the case of Guinea after
Sekou Toure. The continuity of the average authoritarian regime when
faced with a leadership transition is far from guaranteed. In sum, the
7
I define a peaceful transition as one where the outgoing leader exits power via regular
means and the incoming leaders enters power via regular means – to be defined more
precisely later (Goemans, Gleditsch, and Chiozza 2009).
7
1 Introduction
15
10
0
1950 1960 1970 1980 1990 2000 2010
year
successful coups failed coups
Figure 1.1 Number of failed and successful coups in autocracies
10
Peaceful transitions
5
Number of transitions
0
Violent transitions
5
End of Cold War
10
1945
1950
1955
1960
1965
1970
1975
1980
1985
1990
1995
2000
2005
2010
Year
Figure 1.2 Number of violent and peaceful leadership transitions in autocracies
8
1.3 Sometimes Window Panes, But Often Window Dressing
stability of authoritarian regimes varies widely across countries and over

time. Differences in stability stem both from threats to leaders while they
are in power as well as the durability of the regime in light of leadership
transitions. While some dictatorships cease to exist after the death of
a single personalist leader, other regimes develop into stable and institu-
tionalized systems.
1.3 s ometimes window panes, but often window

dressing
To explain variation in regime outcomes, recent studies of dictator-
ship have focused on the role of nominally democratic institutions8 –
such as parties and legislatures – in order to promote authoritarian
durability.9 The general consensus is that institutions matter, even in
autocracies. This finding has been hugely important in advancing
theories of authoritarian rule – earlier works on dictatorships had
completely written off parties, legislatures, and elections as shams. As
Gandhi (2008) notes, prior work tended to assume that the presence
of authoritarian institutions was little more than “mere window
dressing” (xv). The recent “institutional turn” in comparative author-
itarianism has rightfully renewed attention to the role of formal
institutions in autocracies by highlighting ways in which leaders can
benefit strategically from these institutions.10
Despite these important advancements in the literature on authoritarian
institutions, the presence of parties, legislatures, and elections cannot explain
variation in regime outcomes simply because most contemporary dictator-
ships employ a wide range of institutions. From 1946 to 2008, autocratic
leaders maintained a ruling party 87 percent of the time. During that same
period, authoritarian regimes had legislatures 85 percent of the time
(Cheibub, Gandhi, and Vreeland 2010). Figure 1.3 displays the proportion
of autocratic countries with ruling parties and legislatures over this period. It is
clear that the vast majority of these countries have these institutions in place.
Autocratic constitutions and elections have been just as common. From
1946 to 2008, 93 percent of all autocracies had constitutions (Elkins,
Ginsburg, and Melton 2014). During that same period, a total of 2,122
8
This book uses the terms “quasi-democratic” or “nominally democratic” institu-
tions in autocracies and “authoritarian institutions” synonymously.
9
See Bracanti (2014), Gandhi and Lust-Okar (2009), Geddes, Wright, and Frantz
(2018), Gehlbach, Sonin, and Svolik (2016), Lagace and Gandhi (2015), Magaloni
and Kricheli (2010), and Pepinsky (2014) for recent surveys of the literature on
authoritarian institutions.
10
“Institutional turn” phrase borrowed from Pepinsky (2014).
9
1 Introduction
1
.9
.8
.7
.6
.5
.4
.3
.2
.1
0
1940 1950 1960 1970 1980 1990 2000 2010
year
% autocracies w/ ruling party % autocracies w/ legislature

% autocracies w/ constitution
Figure 1.3 Proportion of autocratic countries with nominally democratic institutions
elections were held in 124 countries – 707 presidential elections and 1,415
legislative elections (Hyde and Marinov 2012).11
These numbers are not simply being driven by a post-Cold War pro-
liferation of institutions. From 1946 to 1990, 84 percent of authoritarian
regimes had parties, 80 percent had legislatures, and 92 percent had
constitutions. Moreover, 1,165 elections were held during that time per-
iod – 338 presidential elections and 827 legislative elections. The typical
post-Second World War autocracy has had parties, legislatures, elections,
and constitutions while in power. In other words, the presence of author-
itarian institutions is simply unremarkable and there really is minimal
variation in the existence of institutions in modern autocracies.
Moreover, most ruling parties fail to outlive the death of the founder.
61 percent of ruling parties do not survive more than a year past the
founding leader’s death or departure from office, as illustrated in
Figure 1.4. Even in cases where the first leader experienced a nonviolent
exit from power, only 58 percent of ruling parties outlive the leader.
Furthermore, 43 percent of ruling parties that are coded as part of domi-
nant-party regimes fail to survive a year past the departure of the first
11
The percentage for ruling parties, legislatures, and constitutions are calculated as the
percentage of country–year observations that had the institution. The election
numbers are presented as counts.
10
1.3 Sometimes Window Panes, But Often Window Dressing
mean +1 s.d.
90 90
80
70
60
Frequency
50
40
30
20
11
10 6 7
5 5 4 2 4 2
0
0 10 20 30 40 50
Number of years party remained in power past founding leader
Figure 1.4 Number of years ruling parties remain in power past the founding leader
leader (Meng 2019). In sum, the existence of many ruling parties seems to
rely heavily on the influence of a single leader.
We then have a contradiction. On one hand, existing theories argue
that these institutions are supposed to help regimes survive. Yet, the data
shows that authoritarian institutions are incredibly common and fre-
quently do not outlive individual leaders. The average authoritarian
regime often cannot survive independently of particular leaders and
regime instability has persisted.
I argue that this seeming contradiction exists because scholars have
been focusing on the existence of institutions, rather than the content of
these institutions – in particular whether the leader is constrained. Simply
having a ruling party or legislature does not necessarily mean that the rules
or procedures governing these organizations are institutionalized. In fact,
the appearance of democratic-like institutions, such as ruling parties,
often obscures the true lack of constraints on the executive.
This book examines the causes and consequences of autocratic regime
institutionalization. The first half of the book presents a theory of how
institutionalized regimes emerge by explaining the conditions under
which autocratic leaders choose to implement constraints on their author-
ity after coming into power. In examining the origins of institutionalized
regimes, this book provides a clear mechanism for the question of how
institutions constrain leaders in dictatorship. After all, if autocratic lea-
ders can create institutions, then they can also dismantle them as well.
11
1 Introduction
How then, do rules and procedures have any bite within inherently weakly
institutionalized environments? The second half of the book shifts from
explaining the causes of regime institutionalization to examining the
consequences of institutionalization on key outcomes, such as leader
tenure, coup vulnerability, and leadership succession.
1.4 autocratic regime i nstitutionalization:

a deeper look
Since regime institutionalization is a key concept of this book, I first
provide a brief discussion of this term. I define autocratic regime institu-
tionalization as the creation of rules and structures that govern the dis-
tribution of power and resources within the ruling coalition. Importantly,
within authoritarian systems, institutionalization depersonalizes the ways
in which the regime is run by constraining the leader’s ability to make
arbitrary decisions in the future. Institutionalized regimes are autono-
mous organizations, capable of functioning regardless of which leader is
in power. One of the key features of an authoritarian state is that power is
concentrated in the hands of a small group of elites – and often – in the
hands of a single dictator. Since institutions are especially prone to pre-
dation by autocratic leaders, institutionalization strengthens regimes by
implementing rules, procedures, and structures that promote organiza-
tional autonomy and permanence. When we think about the durability of
an autocratic regime, the extent to which there are structures and proce-
dures in place to guard against personalist rule is critically important.
Institutionalization can occur at many levels of government and within
any kind of organization. Scholars of party-system institutionalization,
for instance, argue that parties can be more or less institutionalized
depending on the extent to which competition occurs, whether the orga-
nizations have stable roots in society and whether the parties are perceived
to be legitimate (Mainwaring and Scully 1995). When regimes, organiza-
tions, or procedures become embedded and valued within society, the
interests of the masses become tied to the organizations and rules them-
selves. As a result, it becomes difficult to for elites to single-handedly
revoke or alter existing institutions.
So why does this book focus on elite politics and the institutionalization
of power at the very top, rather than the process by which these rules
become embedded into societal interests? This study focuses on institu-
tionalization at the very highest echelons of power because the ability to
solve the problem of leadership succession is a first-order challenge facing
all dictatorships. Regimes cannot become institutionalized without first
constraining the leader – it is a necessary first step in order to transform
12
1.4 Autocratic Regime Institutionalization
a regime of one man into a self-sustaining organization. Before institutio-

nalization at the societal level can occur, rules and procedures must first
gain stability and permanence at the elite level. This study therefore
focuses on the institutionalization of elite politics, notably the creation
of procedures that regulate leadership transitions and mechanisms of elite
appointments within presidential cabinets.
To measure regime institutionalization, I present original cross-
sectional time-series data on executive constraints in all countries within
sub-Saharan Africa from 1960 to 2010. Sub-Saharan Africa has a long
and dynamic history of personalist regimes, military dictatorships, and
party-based rule, providing a rich setting in which to examine variation in
regime institutionalization. I collect historical data on state constitutions
and presidential cabinets, for which I have comprehensive yearly records
for 46 countries in sub-Saharan Africa. From these historical records,
I document the creation of executive constraints, hierarchical positions,
and implementation of rules and procedures that structured the distribu-
tion of power. Since institutionalized regimes are those that can outlive
individual leaders, my measures focus on the creation of an autonomous
government, including the development of formal leadership succession
policies. I document whether term limits and formal succession policies
exist within the constitution. A regime with constitutional rules specifying
how leaders will be replaced and the length of their tenure represents
a more highly institutionalized organization because it has internal
mechanisms to perpetuate itself beyond the lifespan of a single leader.
Beyond formal constitutional rules, I also consider the development of
norms surrounding regular presidential cabinet appointments as measures
of informal institutionalization. In particular I focus on whether top
ministerial posts – the positions of vice president, prime minister, and
the minister of defense – are filled on a regular basis. The appointment of
elites to key “power positions” demonstrates a willingness of the leader to
delegate authority and share power. The appointment of key posts to
other elites is not a trivial matter when it comes to presidential cabinets
within sub-Saharan Africa. Presidents frequently eliminated cabinet posi-
tions or appointed themselves to top cabinet posts because they were
hesitant to share power with other elites.12 Cases in which presidents
appointed top cabinet positions to other elites represent regimes with
higher levels of institutionalization because decision-making authority is
12
Jose Eduardo dos Santos, for instance, the first president of Angola, named himself
as his own vice president for a number of years while he was in power. Hastings
Banda, the first president of Malawi, appointed himself as the minister of defense
during his entire tenure, from 1964 until 1993.
13
1 Introduction
not concentrated entirely within the hands of the leader. Furthermore, by

delegating responsibility and authority to other elites within the cabinet,
ministerial appointments allow the regime to function independently of
the leader.
This appointment of elites to key cabinet positions, especially over minis-
tries that control valuable resources, is not a small or meaningless handout.
The goal of capturing power, especially within sub-Saharan Africa, is to
control the state. Having access to the state provides elites with the ability
to control the economy. This was especially true during the decades immedi-
ately following independence due to the agricultural and subsistence nature of
most African economies. The government was and continues to be the main
source of funding for development projects, employment, and the distribu-
tion of scarce resources to localities. Having authority over a piece of the
central government represents a great deal of power for elites.
By examining norms surrounding regular cabinet appointments, this
book argues that institutionalization can occur when informal institutions
constrain leaders13 – the process is not solely restricted to the domain of
formal constitutional rules. When the appointment of cabinet officials
becomes regularized, for instance, this creates the expectation that the
president will delegate cabinet positions to other elites in the future. The
appointment of elites to the presidential cabinet represents a dimension of
institutionalization that is informal, yet plays a key role in constraining
executive power. Regime institutionalization can be achieved even in the
absence of ruling parties or legislatures.
Figure 1.5 displays the proportion of autocratic countries in Africa
with executive constraints from 1960 to 2010. When we conceptualize
regime institutionalization in this way, the frequency with which dictators
are constrained suddenly looks very different – especially when compared
with Figure 1.3. In the first decade after independence, less than a quarter
of countries within sub-Saharan Africa had any constitutional constraints
in place. This contrasts sharply with statistics of the frequency of ruling
parties or legislatures during the same period, which were in place upward
of 80 percent of the time. Even in the two decades after the end of the Cold
War, only about 50 to 80 percent of countries in Africa had formal
executive constraints, compared with over 90 percent of countries that
have parties, legislatures, and constitutions during this period.
This book provides an alternative perspective on democratization in
Africa. Almost three decades after the introduction of multiparty
13
Following existing studies, I define informal institutions as unwritten rules that are
“created, communicated, and enforced outside of officially sanctioned channels”
(Helmke and Levitsky 2004, 727).
14
1.4 Autocratic Regime Institutionalization
1
.9
.8
.7
.6
.5
.4
.3
.2
.1
0
1960 1970 1980 1990 2000 2010
year
Successor policy Term limits

VP/PM appointment Defense appointment
Figure 1.5 Proportion of autocratic countries in sub-Saharan Africa with executive

constraints
elections, true democracy continues to elude most African nations.

Despite the high expectations of these seemingly democratic institutions,
elites still continue to control the system in many countries. By contrast,
this book contends that the real story of Africa in the 1990s was not
democratization; it was institutionalization. As Figure 1.5 illustrates, one
of the most dramatic changes that occurred at the end of the Cold War is
that authoritarian regimes in Africa became much more institutionalized,
along constitutional dimensions as well as through cabinet appointments.
Many leaders went from being unconstrained in their rule to accepting
formal limits on their authority in constitutions. Presidents went from
serving as their own vice president and defense minister to delegating these
powers to other elites. This book will show that the most dramatic change
that occurred in Africa in the 1990s was not that ordinary citizens could
vote in multiparty elections14 – it was that elites recalibrated the political
system amongst themselves to entrench their stability.15 Credible elite
power-sharing set the stage for more durable authoritarian rule, even
with the introduction of nominally democratic institutions.
14
Although the introduction of multiparty elections was indeed an important move-
ment toward political liberalization. Prior to the end of the Cold War, single-party
elections dominated the political landscape in most African regimes.
15
This argument echoes Albertus and Menaldo’s (2018) study of elite-driven
democratization.
15
1 Introduction
1.5 causes of regime i nstitutionalization
1.5.1 Why Do Some Leaders Adopt Executive Constraints?

The first half of the book tackles the question of how institutionalized
forms of dictatorship emerge. Why do leaders choose to institutionalize
their regimes after coming into power if institutionalization ties their own
hands and under what conditions would they do so?
Building on insights from prior scholarship I argue that regime institu-
tionalization is a mechanism that allows autocratic leaders to create
a semiautonomous structure that can enforce joint rule. I present a game-
theoretic model in which regime institutionalization shifts the distribution
of power against the leader in the future period of a two-period bargaining
game. Institutionalization alleviates commitment problems in elite bar-
gaining by empowering elites, therefore providing them with the ability to
hold autocrats accountable for promises made about future rent
distribution.
1.5.2 How Do Institutions Constrain?

A key question that emerges in this discussion is how institutions provide
commitment power in dictatorships. After all, one of the defining char-
acteristics of an authoritarian regime is that no independent authority can
guarantee promises to divide spoils. Authoritarian regimes are, by
definition, weakly institutionalized environments. Moreover, a leader
who can create an institution can also disassemble it as well. How do
institutions have any bite in dictatorships?
I argue that institutions can credibly constrain leaders only when they
change the underlying distribution of power between leaders and elites.
Institutions matter, not because they establish de jure rules, but when they
affect de facto political power. In the context of this book, the establish-
ment of executive constraints empower elites by providing them with
access to the state. When an elite is given a key cabinet position, such as
vice president or the minister of defense, he is given access to power and
resources that allows him to consolidate his own base of support. Elites
who are appointed to positions of authority within the regime then
become focal points for other elites. They become obvious potential
challengers to the incumbent if she were to renege on promises to dis-
tribute rent. This is particularly true if a particular appointee, such as the
vice president, is designated in the constitution as the legal successor to
the incumbent. In such a case, a particular elite is publicly declared the
number two authority within the regime, allowing other elites to
16
1.5 Causes of Regime Institutionalization
coordinate around them. Even when leaders have the ability to choose
who they appoint to these key positions – as they often do – the simple act
of delegating authority shifts the underlying distribution of power
between the leader and her appointees. In the model, this mechanism is
formalized as a shift in the future distribution of power against the leader.
When a president institutionalizes the regime, she voluntarily helps to
solve the elite collective action problem by identifying and empowering
particular individuals who become more capable of unseating her in the
future.
If regime institutionalization seems to weaken leaders, why would any
autocrat choose to tie their own hands after coming into power? The
model shows that autocratic leaders are most likely to place constraints
on their own authority when they enter power vulnerable and highly
susceptible to being deposed. Because per-period transfers are often insuf-
ficient to buy quiescence from exceptionally strong elites, weak leaders
remain in power only by delegating authority to elites as an accountability
mechanism that guarantees their access to future rents. On the other hand,
exceptionally strong leaders who have already consolidated power by the
time they come into office have no incentives to create institutionalized
mechanisms for rent distribution because they face very low likelihoods of
being deposed. A weak autocratic leader is therefore better off taking
actions to tie her own hands because doing so lengthens her time in office.
It is important to clarify that these constraints are essentially forced on
weak incumbents; they reluctantly institutionalize under duress when
faced with elites who can credibly remove them.
My theory stresses the path dependent nature of regime institutionali-
zation. How a leader enters power and the extent to which they have
already consolidated their authority when they enter office determines, in
large part, whether regime institutionalization will occur. In the model,
even if the leader receives a particularly weak draw of power in the initial
period, and anticipates being much stronger in the future, commitment
problems that arise in the present swamp future distributive considera-
tions. Leaders make decisions about institutions at the start of their
tenure, and these institutional decisions shape the rest of their rule.16
This theory suggests an ironic twist of fate. Initially strong leaders are
never incentivized to build credible ruling organizations because they are
able to remain in power without making institutional commitments to
other elites. Yet personalist strategies of rule are ultimately destabilizing in
16
In this sense, the initial period of a leader’s tenure can also be considered a critical
juncture.
17
1 Introduction
the long run, especially upon the death of the ruler. Conversely, initially
weak autocrats who lack a strong basis of support must pursue the
counterintuitive strategy of committing to give power away when they
are most vulnerable. Doing so allows such leaders to buy support from
elites who would otherwise jump at the opportunity to depose them. Yet
at the same time, these self-interested actions generate stable power-
sharing institutions, setting the stage for durable authoritarian rule.
Empirical patterns of institutionalization in sub-Saharan Africa sup-
port this argument. Founding presidents who came into office immedi-
ately following independence – especially those who were influential
leaders of mass independence movements – benefitted from extreme
legitimacy and high levels of popular support. Such leaders, like Félix
Houphouët-Boigny of Côte d’Ivoire or Kwame Nkrumah of Ghana, were
seen as the “founding fathers” of the newly independent African states.
Both leaders, like most other founding presidents, did not appoint a vice
president during their tenure and lacked constitutional constraints on
their power.
Presidents who came into power via coups, such as Mobutu Sese
Seko of Zaire or Gnassingbe Eyadema of Togo, were also much less
likely to institutionalize their regimes along formal and informal
dimensions. One of the most precarious risks to any civilian incum-
bent is the threat of a military takeover. When the leader is the head
of the military, however, this threat is diminished. Autocrats who
successfully come into power through coups demonstrate that they
have control over the coercive apparatus that allowed them to launch
a successful coup in the first place. Since they retain control over the
military, coup leaders do not need to institutionalize the regime in
order to remain in power.
The end of the Cold War was also a watershed event that changed
the dynamics of regime institutionalization. Beginning in the 1990s,
international norms abruptly shifted toward favoring democratization
and political liberalization. Leaders who had ruled as single strong-
man dictators were now facing condemnation in light of changing
international norms, therefore shifting the distribution of power in
favor of elites. As a result, levels of formal and informal institutiona-
lization increased sharply. This discontinuous jump in the levels of
institutionalization after 1990, as illustrated in Figure 1.5, is very
evident when graphed over time (especially when compared with
minimal changes in the existence of parties, legislatures, and constitu-
tions, as illustrated in Figure 1.3). In fact, this sharp increase in
institutionalization also explains why leadership transitions have
become increasingly peaceful in the post-Cold War era.
18
1.6 Consequences of Regime Institutionalization
1.6 consequences of regime

i n st i t u t ion a l i z a ti o n
1.6.1 Do Executive Constraints Promote Autocratic Stability?

After taking into account the endogenous emergence of executive con-
straints, the book reexamines the debate over the relationship between
institutionalization and autocratic durability. Does the creation of execu-
tive constraints have stabilizing properties for autocratic leaders and
regimes? The second half of this book examines these questions for
a number of key regime outcomes: length of leader tenure, coup vulner-
ability, and leadership succession.
Despite the general consensus that institutionalized forms of dictator-
ship tend to be more stable, the empirical record of the relationship
between institutions and regime stability has been surprisingly mixed. In
a seminal study, Geddes (1999a) finds that single-party regimes remain in
power for longer periods of time compared with personalist regimes. Since
then, many scholars have expanded on this argument that ruling parties
can prolong the lifespan of authoritarian regimes because they serve
a number of stabilizing functions.
At the same time the empirical finding in Geddes (1999a) has been
challenged by other scholars who do not find a statistically significant
relationship between strong parties and regime longevity. Smith (2005),
for instance, shows that the significant effect of single-party regimes is
driven primarily by two outliers: Mexico and the Soviet Union. Using
a different dataset, Gandhi (2008) does not find evidence of a statistically
significant relationship between the presence of parties and legislatures
and autocratic survival. She argues that we should not observe
a significant effect because leaders endogenously respond to threats with
“appropriate” degrees of institutionalization.
How do we make sense of these divergent empirical findings? Should
we observe a significant effect of institutionalization on stability? This
book heeds Pepinsky’s (2014) call to address the “nagging problem” that
remains within the scholarship on authoritarian institutions. He notes
that “[if] institutions under authoritarian rule are vulnerable to manip-
ulation because political actors believe that institutional manipulation can
shape political outcomes in their favor, then it is also true that factors that
explain the origins of (and changes in) dominant parties also directly
affect those political outcomes” (631). When we consider the effects of
institutionalization on regime stability, we must consider that leaders
make decisions to institutionalize based on underlying threats to their
authority.
19
1 Introduction
The first half of this book makes the case that leaders make strategic
decisions about whether to institutionalize their regimes, depending on
their strength, vis-à-vis other elites. Because institutionalization is an
endogenous process, executive constraints should only have an observable
effect when we condition on initial leader strength. Strong leaders who
have already consolidated power when they take office do not need to rely
on power-sharing institutions to remain in power. Their rule is secure with
or without institutional mechanisms. We therefore should observe little or
no effect of institutionalization on regime durability for strong leaders.
For leaders who are weak, however, building an institutionalized
regime does promote stability. Leaders who are highly vulnerable to
being deposed by their fellow elites require executive constraints in
order to make credible commitments to share power and resources.
Institutionalization should therefore lengthen the time horizons and lessen
coup risk, but primarily for weak leaders.
Since initially strong leaders are likely to be able to remain in power
regardless of whether they institutionalize or not, a regression model that
regresses institutionalization on regime stability without taking into
account differential effects based on leader type will likely result in either
a diminished effect of institutionalization or a null effect, both of which
would be inaccurate.
To address this endogenous institutional choice, I incorporate an
interaction term between regime institutionalization and leader strength
which accounts for differential effects of institutionalization based on
whether the leader was initially strong or weak. Using founding presi-
dents and postcoup leaders as proxies for strong leaders, I find when we
condition on initial leader strength, institutionalization does indeed have
a stabilizing effect for weak leaders but little or no effect for strong
leaders.
1.6.2 Are Institutionalized Regimes More Likely To Undergo Peaceful

Leadership Transitions?
Finally, I also consider whether highly institutionalized regimes are more
likely to undergo peaceful leadership transitions. The central challenge
facing the survival of all authoritarian regimes is how the regime can
transform from a dictatorship of one man to a self-sustaining system
that is governed by rules. Dictatorships inherently lack mechanisms of
electoral competition and leadership succession. Incumbents often remain
in office for long periods of time, making it difficult to establish norms of
leader turnover. Elections – even when they exist – are largely meant to
sustain, rather than disrupt, the incumbent’s regime. As a result, the
20
1.6 Consequences of Regime Institutionalization
process of peacefully transferring power from one leader to another is

often quite precarious in dictatorships.
I argue that the creation of constitutional succession rules and the
designation of a de facto successor play a critical role in regulating the
process of authoritarian succession. Take the case of the first presidential
succession from Jomo Kenyatta to Daniel arap Moi in Kenya in 1978. The
Kenyan constitution includes a provision that reads: “If a President dies,
or a vacancy otherwise occurs during a President’s period of office, the
Vice President becomes interim President for up to 90 days while
a successor is elected.” Near the end of Kenyatta’s rule, a faction within
the ruling Kenya African National Union (KANU) party tried to contest
the authority of then-vice president Moi on the grounds that he was not
a member of the dominant ethnic group. Moi and his supporters were able
to dispute these claims by relying on the constitutional procedure govern-
ing presidential succession (Karimi and Ochieng 1980; Tamarkin 1979;
Widner 1992).
Importantly, however, I argue that only certain types of rules are
effective in promoting leadership succession. Constitutional rules that
identify a clear line of succession are the most effective in regulating the
transfer of power because such policies establish certainty and predict-
ability about the actors who are expected to come into office. This pro-
vides elites with a stake in maintaining the existing regime in order to reap
rewards from the succession order. Term limits, on the other hand, are
much less effective in facilitating peaceful leadership transitions because
they do not empower specific elites in the way that succession policies do
by designating an heir apparent. Because term limits do not identify an
alternative leader that elites can coordinate around, term limits do not
resolve the collective action problem elites face in holding the incumbent
accountable.
This distinction between term limits and succession procedures is an
excellent illustration of the following property that mentioned at the start
of the chapter: institutions matter, not because they establish de jure rules,
but when they affect de facto political power. On paper, it would seem like
term limits are an extremely powerful tool to prevent executive overreach.
In practice, however, it is a less effective institutional constraint compared
with succession procedures because term limits alone do not change de
facto political power.
It is also important to note that the peaceful transition of power is an
unintended consequence of strategic institutionalization by leaders.
Incumbents plan for succession not out of concern for the stability of the
regime after they die, but because they do not want to be preemptively
deposed by elites who will compete for the presidency in the absence of
21
1 Introduction
a designated successor. Although incumbents institutionalize in order to

stabilize their own rule, these self-interested actions benefit their succes-
sor, who, once appointed, has every incentive to remain loyal to the
current regime in the hope of becoming the next incumbent. In turn,
these self-reinforcing succession plans promote regime stability in the
long run.
In an analysis of all postindependence leadership transitions in sub-
Saharan Africa, I show that regimes with constitutional succession pro-
cedures and a designated successor are significantly more likely to
undergo peaceful leadership successions. A regime that has constitu-
tional leadership succession rules is about 60 percent more likely to
undergo a peaceful transition compared with one without any formal
rules regulating succession. The designation of an informal successor has
a similar effect.
Altogether these results suggest that institutionalization does
indeed matter for regime longevity, as long as we condition on the
endogenous emergence of institutions. In fact, an important lesson of
this book is that parties, legislatures, constitutions, and elections can
matter greatly, but the effectiveness of these institutions depends
crucially on the strength and level of institutionalization of these
institutions themselves.
Figure 1.6 summarizes the theoretical argument made in Part II of the
book. Initially strong leaders do not institutionalize their regimes because
they do not need to do so to stay in power. Such regimes appear to be long
lasting and stable while the founding leader is still alive. However, upon
the death of the founding leader, the regime undergoes conflict if succes-
sion plans have not been formulated or institutionalized. Initially weak
leaders who do institutionalize are able to stabilize their tenure, compared
with initially weak leaders who do not institutionalize. Furthermore,
institutionalized regimes are significantly more likely to undergo peaceful
leadership transitions. Somewhat ironically, regimes that originate with
weak leaders are much more likely to become durable and long lasting
autocracies.
Strong No Violent leadership

Stable regime
leader institutionalization transition
No Unstable Violent leadership

Institutionalization regime transition
Weak
leader
Peaceful leadership
Institutionalization Stable regime transition
Figure 1.6 The effect of regime institutionalization on leader and regime outcomes
22
1.7 This Book’s Contributions: Authoritarian Regimes
1.7 t his book’s contributions: authoritarian

regimes
This book makes a number of contributions to studies of authoritarian
regimes. First, it highlights – both theoretically and empirically – the
importance of examining the content rather than existence of authoritar-
ian institutions. By providing data on the details of these institutions, I am
able to examine the extent to which autocratic leaders truly tie their hands
via formal or informal rules and procedures. Despite the wide prevalence
of parties or legislatures, there are important differences in the
institutional makeup of dictatorships when we examine the content of
executive constraints.
This study reveals these differences by presenting new evidence of how
regime institutionalization varies along key dimensions of executive con-
straints. In doing so, I operationalize a key variable to which many scholars
have referred but often do not carefully define nor measure systematically.
These measurement decisions have important substantive implications,
both for our understanding of the frequency of institutionalized forms of
dictatorship but also for our ability to accurately test arguments centering
on the relationship between institutional strength and regime durability.
The literature on authoritarian parties, for instance, has made the case that
ruling parties may play an important role in regulating leadership succes-
sion. However, this argument has had limited empirical verification due to
data limitations. By contrast, I show that constitutional succession rules and
the designation of a successor (which are sometimes, but not always,
embedded in party-based regimes) positively associate with regulated
transitions. Ruling parties, without taking into account the quality of the
institutional arrangements, simply have no effect on their own.
This measurement contribution builds on Geddes’ (1999a) path-
breaking work on regime typologies. In a seminal study, she classifies all
autocracies into regime types: party, personal, military, or hybrids of these
categories. This classification scheme was one of the earliest studies to
codify differences in the institutional quality of dictatorship and stimu-
lated a large body of work that followed. However, the regime typologies
framework is subject to a number of measurement concerns. First, placing
regimes into time-invariant categories obscures institutional change over
time.17 This is a significant limitation because leaders within the same
regime often make very different decisions about the extent to which they
17
Although it is possible for countries to be coded as different regimes over time, most
are not. Furthermore, this coding scheme does not allow for variation within
regimes. See Chapter 4 for the full discussion.
23
1 Introduction
are willing to defer to institutions. Take China as an example. The Chinese

Communist Party (CCP) currently rules China as a strong and autono-
mous party machine. The CCP has an extensive organizational hierarchy
and well-defined norms regarding promotion within the party ranks. By
contrast, China under the rule of its founding leader, Mao Zedong, was
significantly less institutionalized than it is today. Mao arbitrarily purged
and promoted party officials at will and launched political campaigns,
such as the Hundred Flowers Movement and the Cultural Revolution, in
order to eliminate his rivals. According to the regime typology framework,
China is coded as a single-party regime since 1949, despite this institu-
tional variation over time.
Furthermore, regime typologies are composed of mutually exclusive
categories that are often insufficiently nuanced to capture differences
between countries coded within the same category (Svolik 2012). China
under the rule of the Communist Party is very different compared with
Guinea under the (superficial) rule of the PDG, though both are coded as
single-party regimes. Rather than creating mutually exclusive categories,
this book creates disaggregated indicators of regime institutionalization
that capture changes in institutional quality over time.
Moreover, the measures presented in this book are objectively coded in
a transparent way that can be verified and replicated. Although Geddes
(2003) outlines a clear set of guidelines that were employed to categorize
regimes into different categories, a number of these criteria used to code
regime types require the researcher to make subjective decisions about
how to code the regime.18 It is also not clear what the individual responses
to these criteria were, making it difficult to interpret regime categories that
are an aggregation of these individual criteria. By contrast, the measures
presented in this book are all disaggregated and were constructed by
simply observing whether certain constitutional rules or cabinet appoint-
ments existed.19 The coding scheme therefore does not require the
researcher to make subjective judgment calls and is easily replicable by
other researchers.
In fact, I demonstrate that regime institutionalization is not synon-
ymous with the presence of nominally democratic institutions. Very
18
Examples of subjective criteria include: “Does the party have functioning local level
organizations that do something reasonably important, such as distribute seeds or
credit or organize local government?” or “Has rule of law been maintained?”. The
possibility of measurement error based on subjective coding rules is also heightened
by the fact that the dataset spans multiple regions and time periods and often rely on
information from various country experts.
19
My approach is similar to the measurement strategies used in Gandhi (2008) and
Svolik (2012).
24
1.7 This Book’s Contributions: Authoritarian Regimes
often data on the presence of institutions has substituted for data on the
content of institutions. However, comparisons of my institutionalization
measures with existing datasets that focus on the existence of autocratic
institutions verify that simply having parties or legislatures does not mean
that executive constraints on the leader exist. Moreover, many regimes
that have been coded as dominant-party regime types are also not very
institutionalized. Even aggregate Polity scores of executive constraints
(“XCONST”) display massive inconsistencies when compared to the
presence of constitutional term limits and succession procedures. In
sum, existing variables of autocratic institutions are poor predictors of
regime institutionalization and do not reflect the extent to which the
leader is constrained.
This book also contributes to the literature on comparative political
institutions by developing a theory of why dictatorships vary in institu-
tional strength. I build on a small but growing literature of institutional
change in developing or transitional states (Gryzmala-Busse 2007;
Helmke 2017; Levitsky and Murillo 2009; Miller 2020; Opalo 2019).
This book also highlights the role of informal institutions in regulating
political power (Gryzmala-Busse 2010; Helmke and Levitsky 2004; 2006;
Lauth 2000; Mershon 1994) – a topic that has received very limited
attention within the autocratic context (Tsai 2007). It is also in conversa-
tion with studies that examine how and when institutions become self-
reinforcing in otherwise weakly institutionalized settings (Ginsburg and
Simpser 2013; Greif and Laitin 2004; Mahoney and Thelen 2010; Nalepa
2010; Negretto 2013; Pierson 2000; Przeworski 2015; Thelen 2004).
However, while most of these studies explain why and how institutions
evolve in newly democratic or quasi-democratic settings, I focus my
attention on explaining institutional change within dictatorships.
As a second key contribution, this book ties the origins of institutions
to their consequences in a comprehensive account of the causes and
consequences of autocratic regime institutionalization. Scholars have
only recently begun to examine why some leaders create institutional
arrangements rather than relying on private transfers to maintain elite
support. Early studies either took the existence of a dominant party as
exogenous (as noted explicitly in Magaloni 2008)20 or identified multiple
strategies of rule (repression, informal cooptation, or institutions) without
20
She states: “A key question that emerges from this discussion is why not all dictators
create political parties if these play such powerful roles at minimizing their risk of
being overthrown by members of the ruling coalition. My account does not address
the question of origins – how successful and credible political parties get established
in the first place” (11).
25
1 Introduction
specifying the conditions under which leaders would choose to rule with
or without institutions (Haber 2006). As Reuter (2017) notes, “we know
much more in political science about the equilibrium characteristics of
dominant parties . . . than we do about how these equilibria come to be
established in the first place” (13).
A growing number of studies are now considering how institutional
equilibria emerge in dictatorships. Gandhi (2008) and Gandhi and
Przeworski (2007) first highlight the notion that institutions emerge
endogenously when they benefit leaders. Brownlee (2007) argues that
intra-elite conflict during the early years of the regime necessitates the
creation of dominant parties, and Slater (2010) argues that elites create
strong states and durable parties when they face “endemic threats” to
their existences. Other studies have also highlighted the role of external
threats (Paine 2019; Smith 2005) and revolutionary conflict (Huntington
1968; Levitsky and Way 2012; 2013) in creating conditions that facilitate
the development of strong party institutions or elite power-sharing. Three
recent studies have dealt with this question of institutional creation most
directly and point to the key mechanism of commitment problems
between leaders and elites that motivates the creation of power-sharing
institutions (Boix and Svolik 2013; Gehlbach and Keefer 2011; Reuter
2017).
This book differs from these existing accounts by examining the origins
of executive constraints in constitutions and presidential cabinets rather
than the origins of the parties and legislatures. In fact, historical evidence
often suggests that many ruling parties are not strategically created, but
instead, inherited by leaders. Within my sample of African countries, for
instance, virtually all leaders came into power with a preexisting party in
place immediately following independence. These ruling parties were
created prior to decolonization not as elite power-sharing devices, but
rather as pro-independence organizations (Bienen 1970; 1978; Zolberg
1966; 1969) or parties that were used to participate in preindependence
elections (Collier 1982). Rather than arguing that leaders make strategic
decisions to create institutions, this book argues that leaders make strate-
gic decisions to institutionalize their regimes after coming into power.
In providing a comprehensive account of both the causes and conse-
quences of institutionalization, this book is able to take into account
endogenous decisions to institutionalize when estimating the effects of
these institutions. Scholarship examining the effects of institutions on
regime durability has generally developed in isolation from studies of
the origins of these institutions. Yet the consequences of institutions are
intimately related to where these institutions come from. This book con-
siders how the strategic origins of institutionalization have an effect on the
26
1.8 This Book’s Contributions to African Politics
consequences of such rules and procedures. In doing so, I aim to separate

the effects of institutions on regime outcomes from the effects of the
underlying power distributions that drove leaders to create these institu-
tions in the first place.
1.8 why africa? this book’s contributions

to af rican politics
To study regime institutionalization, this book takes a broad look within
a particular region: I examine all 46 countries within sub-Saharan Africa
from the end of independence through 2010. To my knowledge, this is
one of the few studies of authoritarian politics within sub-Saharan
Africa21 – and one of the few that is comprehensive in the coverage of
countries within the region.
It is a bit surprising that sub-Saharan Africa as a region has been
underrepresented in the study of authoritarian politics. Every country
within Africa was an authoritarian state following independence in
the 1960s.22 The majority of countries within Africa remain author-
itarian today. Countries like Tanzania, Cameroon, and Mozambique
have been governed by the same group of elites that took power after
independence was granted. These ruling elites were able to retain
power despite the emergence of democratic pressures that surfaced
at the end of the Cold War, and these regimes remain clearly author-
itarian today. Even a country such as Botswana that is considered by
many to have fair elections has been governed by the same ruling
party since independence in 1966. In short, the politics of dictatorship
has been, and in many cases continues to be, the politics of sub-
Saharan Africa.
Beyond current authoritarian regimes in Africa, this book is also
important for understanding the politics of countries that are newly
democratized, such as Malawi or Kenya, or those that are in the process
of democratizing, such as Liberia. Pressures to democratize did not arise in
Africa until the 1990s and transitions away from authoritarianism have
occurred very recently, with many still ongoing. The legacies of author-
itarian rule often persist in these new or fledgling democracies and
21
A few exceptions include Carter and Hassan (2019), Hassan (2020), and Opalo
(2019). Other studies have examined the effect of authoritarian legacies on demo-
cratization in Africa, without focusing explicitly on the politics of authoritarian rule
(Bleck and van de Walle 2018; Bratton and van de Walle 1997; Cheeseman 2015;
LeBas 2013; Posner and Young (2007); Riedl 2014).
22
Only the island of Mauritius has been considered a democracy since it gained
independence in 1968.
27
1 Introduction
frequently hinder the consolidation of democratization. This study is in

conversation with studies of democratization in Africa by examining the
authoritarian backdrop in which political reforms take place (Arriola
2013; Bleck and van de Walle 2018; Bratton and van de Walle 1997;
Cheeseman 2015; LeBas 2013; Riedl 2014). Studies of newly democra-
tized African countries indicate that party systems and the strength of
opposition parties in the democratic era are shaped by the activities of
these parties from the authoritarian era (LeBas 2013; Riedl 2014). Leaders
often use cabinet positions to buy off opposition politicians, therefore
hampering the creation of strong opposition coalitions. Importantly, they
are able to do so without explicitly violating any electoral rules (Arriola,
DeVaro, and Meng 2017; Buckles and Gandhi 2017). Incumbency advan-
tages gained during the authoritarian period allow incumbents to subvert
genuine democratic competition while still allowing for multiparty elec-
tions. This book is also in conversation with Roessler (2011, 2016), which
examines which types of elites (out-groups versus in-groups) leaders
choose to share power with in order to balance between threats of coups
and civil wars. By contrast, this study examines the conditions that moti-
vate leaders to delegate authority to other elites in the first place. In sum,
the study of dictatorships in Africa both past and present is essential for
our understanding of politics in Africa today.
In studying Africa through the lens of authoritarian politics, this book
challenges a number of conventional wisdoms that have prevailed in the
study of African politics. Scholars have historically argued that regimes in
Africa have often been dominated by “big men” (Bienen 1970; 1978;
Decalo 1976; 1990; Jackson and Rosberg 1982; Zolberg 1966; 1969)
and that leaders rely primarily on informal patronage-based rule to stay in
power (Arriola 2009; Francois, Rainer, and Trebbi 2015). In contrast
with these dominant perspectives, I provide empirical evidence that
some African autocrats do indeed rule through institutionalized mechan-
isms that regulate and depersonalize power, even during the most author-
itarian decades prior to the end of the Cold War. Institutions are not
uniformly weak in Africa, and variation exists both across countries and
over time.
1.8.1 Reinterpreting Contemporary African Politics

Following the end of the Cold War in 1990, many African countries began
to introduce multiparty elections (Bratton and van de Walle 1997).
Although scholars acknowledged that these elections were often marked
by electoral fraud, intimidation of opposition forces, or media censorship,
the general consensus in the academic literature (and in the policymaking
28
1.8 This Book’s Contributions to African Politics
world) was that most African countries were in the process of democratiz-
ing. However, despite the presence of multiparty elections in much of sub-
Saharan Africa for the last 30 years, many incumbents and ruling parties
have maintained a solid grip on power (Bleck and van de Walle 2018). In
short, authoritarian regimes continue to persist, albeit under the façade of
shallow democratic institutions. Yet scholarly discussions of African pol-
itics remain centered on trying to understand problems of democratiza-
tion, rather than trying to understand how autocracies have evolved and
adapted to post-Cold War conditions.
This book proposes a radical reinterpretation of contemporary African
politics: that the institutionalization of regimes beginning in the 1990s
reinforced authoritarian rule, rather than ushering in real democracy. In
order to really make sense of African politics today, we need to stop trying
to understand these regimes as problematic democracies, and analyze
them instead as institutionalized autocracies.
As my data on executive constraints demonstrate, following the end of
the Cold War, regimes in Africa became much more institutionalized as
leaders established power-sharing mechanisms in order to maintain sup-
port from fellow elites. Rather than offering citizens true access to power,
incumbents and elites consolidated their stranglehold on political power,
all while keeping the appearance of governing through democratic-
seeming institutions. The majority of African countries have now intro-
duced term limits and succession policies in their constitutions, in addition
to regularizing the appointment of elites to key cabinet positions.
However, as I argue, the institutionalization of these regimes has resulted
in authoritarian stability, rather than regime weakness or transition to
democracy. By diffusing power among many ruling elites, leaders and incum-
bent regimes actually solidified their grip on power by maintaining elite
support through these institutionalized pacts of long-term rent distribution.
Along with the introduction of multiparty elections – which in reality did little
to threaten incumbents since most opposition forces remained weak – the
introduction of power-sharing institutions allowed authoritarian leaders to
remain in power by limiting their absolute authority. In the post-Cold War
world, constrained authoritarianism is stable authoritarianism, and this
mode of governance persists across much of sub-Saharan Africa.
There are also important advantages to theorizing about authoritarian
stability by examining African dictatorships. There have been many semi-
nal studies of authoritarian institutions and strategies of rule in twentieth-
century China and Mexico23 – countries that do indeed have exceptionally
23
One cannot say the same, for instance, of Djibouti or Mauritania.
29
1 Introduction
strong institutions and have been very long-lived. It is undoubtedly impor-

tant to understand these cases: China has the largest population in the
world, and Mexico under the rule of the PRI was one of the longest-lived
modern autocracies. However, the emphasis on these countries has likely
biased our existing theories. As a result of a large number of studies that
focused on a limited set of countries, theories were often built around
cases that resemble outliers rather than the modal authoritarian country.
This is also especially problematic when arguments that were constructed
around especially successful cases are then generalized using broad large-
N variables (such as the presence of parties) when these data cannot
adequately capture institutional quality.
By contrast, sub-Saharan Africa as a region provides a uniquely good
opportunity in which to theorize about authoritarian politics more
broadly due to variation in institutional quality across regimes and over
time. Authoritarian rule in Africa has ranged from stable and highly
institutionalized regimes, such as in Kenya or Mozambique, to military
forms of government, such as in Ghana or Benin, to personalist one-man
dictatorships, such as Mobutu Sese Seko of Zaire or Idi Amin of Uganda.
In other words, this book does not focus solely on regimes that are stable
and long-lived. Moreover, states within sub-Saharan Africa share
a number of economic and historical similarities, allowing me to hold
a number of macro-conditions constant.
The postcolonial nature of African states also provides an ideal
setting in which to study the emergence of institutionalized systems.
The enterprise of governing in sub-Saharan Africa after decoloniza-
tion was primarily one of state building. The institutions that had
been set up by colonial administrations were primarily used to extract
rather than to govern. When African leaders finally came to power,
they had to reconstitute institutions of authority, and this provided
the opportunity for these leaders to shape presidential power. The
postcolonial experience in Africa, like many other postcolonial
nations, allows researchers to examine how contemporary institutions
are created in new states.
I focus my analysis on regimes within sub-Saharan Africa but maintain
a broad comparative approach by including all countries within the region
in my study, rather than focusing on a single case study. This design allows
me to collect and analyze country-level data on executive constraints
without relying on broad measures that do not provide enough detail or
accuracy. At the same time, I am able to retain a comparative approach by
studying 46 countries over a time span of 50 years, which is more amen-
able for making generalizable arguments about authoritarian regimes
more broadly.
30
1.9 Plan of the Book
In addition, this book takes a multimethod approach by combining

game theory, time-series cross-sectional data analysis, and illustrative case
evidence. There has been a recent push toward studies that link formal and
empirical analysis. The combination of different approaches helps scholars
triangulate between various types of evidence that have different compara-
tive advantages (Aldrich, Alt, and Lupia 2007; Granato and Scioli 2004).
The advantage of applied game theory is that formal models “make argu-
ments more transparent both to those making them and to those to whom
the arguments are made” (Powell 1999, 29). Using a model, I am able to
identify clear tradeoffs leaders are faced with when deciding whether to
institutionalize the regime, as well as consider the effects of other competing
factors. I am also able to clearly identify assumptions that model is making
and validate these assumptions using empirical evidence. By collecting time-
series cross-sectional data on executive constraints, I am able to test the
theoretical arguments produced by my model. While the model’s aim is to
establish internal validity, statistical analyses help to establish external
validity. In addition, by taking measurement very seriously, this book is
able to examine the consequences of institutionalization on a wide range of
outcomes that has had limited empirical verification due to limitations on
existing data. Finally, this book also uses illustrative case evidence to
provide examples of mechanisms and specific cases that exemplify broader
trends that are found in the data. In doing so, I can provide a more concrete
sense of the different types of strategies autocratic leaders used to retain
power over the last five decades within sub-Saharan Africa. Altogether, it is
my hope that the combination of multiple types of inquiry and different
kinds of evidence help to build a thorough and compelling case.
1.9 plan of t he book

The aim of this book is to explain how executive constraints on autocratic
leaders emerge and examine the effects of institutionalization on regime
durability. I alternate between treating regime institutionalization as the
dependent variable and independent variable. Chapters 2 and 3 focus on
institutionalization as the dependent variable by explaining how executive
constraints emerge and how they become self-sustaining. Chapter 4 lays
the groundwork for the empirical analysis of this book. It discusses how
I conceptualize regime institutionalization, describes my dataset, and
compares my measures against other existing datasets on authoritarian
institutions. Regime institutionalization remains the dependent variable
in Chapter 5, which provides empirical evidence of the relationship
between leader strength and regime institutionalization. Chapters 6 and
7 treat regime institutionalization as the independent variable by
31
1 Introduction
examining whether executive constraints promote autocratic stability and

peaceful leadership succession. While the first three chapters of the book
are largely theoretical, Chapters 4 through 7 are primarily empirical.
Chapter 2: Why Do Leaders Institutionalize?

This chapter presents a theory of why and when leaders institutionalize their
regimes after coming into power. I argue that leaders institutionalize their
regimes in order to make credible commitments about future rent distribution
to other elites. I present a formal model in which institutionalization shifts the
future distribution of power away from the leader in a two-period bargaining
game. The model shows that autocrats who enter power in a position of
strength relative to other elites will not institutionalize the regime because
they never face commitment problems in bargaining. Weaker leaders without
such guarantees of stability are more likely to pursue a strategy of institutio-
nalization because doing so provides benefits of stable rule.
Chapter 3: Two Illustrative Cases

This chapter presents two illustrative case studies: Cameroon (a highly
institutionalized regime) and Côte d’Ivoire (a weakly institutionalized
regime). The first president of Cameroon, Ahmadou Ahidjo, entered
power extremely weak. Ahidjo was not an influential independence leader –
he was a close ally of the French colonial administration who inherited his
position of power upon independence. To compensate for extremely low
levels of popularity, Ahidjo used cabinet appointments as a means to buy
support from other elites. By contrast, Félix Houphouët-Boigny, the first
president of Côte d’Ivoire, entered power already extremely influential. He
had been a renowned independence fighter for decades and there were no
other elites of similar political stature in Côte d’Ivoire by the time he took
office. Houphouët remained in power for three decades as a personalist
strongman ruler, but the regime collapsed soon after his death due to the
absence of institutionalized structures.
Chapter 4: How Should Institutionalization Be Measured?

This chapter lays the empirical groundwork for the central concept of the
book. I discuss how I conceptualize and operationalize autocratic regime
institutionalization, presenting the dataset and explaining in detail the coding
strategy. I show that other commonly used datasets of authoritarian institu-
tions, such as regime typologies and Polity scores, do not accurately measure
the extent to which leaders are constrained.
32
1.9 Plan of the Book
Chapter 5: What Are the Causes of Regime Institutionalization?

This chapter provides a cross-national test of the theory that was presented
in Chapter 2. I show that initial leader strength systematically determines
patterns of regime institutionalization. Leaders who were influential inde-
pendence movement leaders or those who came to power via a coup tend
to institutionalize less because these types of leaders enter office with
distinct advantages over other elites. I also show that the end of the Cold
War is associated with a sharp increase in institutionalization as African
leaders lost access to military and economic aid from external sponsors.
Chapter 6: What Are the Consequences of Institutionalization

on Autocratic Stability?
This chapter examines the effects of institutionalization on autocratic stabi-
lity by examining leader tenure and coup threats. I find that executive
constraints lengthen the tenure of weak leaders but have no effect on leaders
who are strong. Similarly, I find that weak leaders who institutionalize are
significantly less likely to face coup attempts, but there is not a strong effect of
institutionalization on coup vulnerability for strong leaders. These results
show that institutionalization does indeed matter for regime durability, as
long as we condition on the endogenous emergence of institutions.
Chapter 7: What Are the Consequences of Institutionalization

on Leadership Succession?
This chapter provides evidence that regime institutionalization matters
greatly for the most critical regime outcome: peaceful leadership succession.
I argue that institutionalized succession procedures regulate the process of
peaceful leadership transitions. I show that regimes with formal succession
rules written into the constitution and leaders who designate a clear succes-
sor are significantly more likely to undergo successful leadership transitions.
Chapter 8: Conclusion
The final chapter concludes by summarizing five key claims made in this
book. I discuss each of the claims and highlight the empirical and theore-
tical contributions of these findings. I then consider the implications of
autocratic regime institutionalization for future studies of institutional
design, democratization, and democratic backsliding.
33
2
Why Do Leaders Institutionalize?
“Montesquieu observed that, at the birth of new polities, leaders mold

institutions, whereas afterwards institutions mold leaders.”
Robert Putnam, Making Democracy Work (1993, 26)
A central question this book addresses is how institutionalized forms of

autocracy emerge. Why do some leaders implement executive constraints
after coming into power while others do not?
On one hand, institutionalization provides some compelling benefits
for leaders. The introduction of this book previewed some key findings
from Chapters 6 and 7, namely that regimes that are institutionalized are
significantly less vulnerable to being overthrown by coups. Moreover,
institutionalized succession procedures allow leaders to remain in power
for longer periods and protect them from being preemptively deposed over
succession conflicts. If executive constraints promote leader and regime
stability, then why don’t all leaders institutionalize their regimes?
Now consider the flip side of this question. I defined regime institutio-
nalization as the creation of rules and procedures that structure the dis-
tribution of power and resources. This necessarily entails constraining the
personal authority of the leader. If the process of institutionalization
involves voluntarily creating rules and procedures that place immovable
restriction on one’s own authority, why would any dictator ever under-
take such measures?
In fact, many autocratic leaders have avoided creating such limitations on
their power. The Democratic Republic of Congo under the dictatorship of
Mobutu Sese Seko, for instance, lacked formal institutionalized rules, and
elites were routinely purged at the will of the leader. Power was concentrated
around Mobutu alone, from the time he seized power via a coup and
throughout his 28-year rule (Jackson and Roseberg 1982). Uganda under
the dictatorship of Idi Amin Dada was characterized as having a “pattern of
random and continued violence” and an “uncontrolled reign of terror”
34
2 Why Do Leaders Institutionalize?
(Decalo 1976; Committee on Foreign Relations 1978). After taking power

through a coup in 1971, Amin purged the army and polices forces, often
using extermination squads. Within a year of taking power, Amin had an
estimated 10,000 people killed and around 1,500 political prisoners impri-
soned. These massive purges created large personnel gaps in state institu-
tions. Amin filled these vacancies “by a series of bizarre promotions from the
ranks that completely bypassed the few middle-level [non-coethnics] who
were not purged” (Decalo 1976, 213).
Why do some autocratic leaders institutionalize their regimes, while others
do not? In this chapter, I will argue that regime institutionalization allows
autocratic leaders to create a semiautonomous structure that can enforce
joint rule. I present a formal model in which an autocrat chooses a level of
regime institutionalization at the start of a two-period bargaining game. In
the model, institutionalization shifts the future distribution of power in favor
of elites, therefore alleviating commitment problems in bargaining by enhan-
cing the ability of elites to overthrow the leader in the future period.
A main finding from the model is that autocratic leaders are likely to
place constraints on their own authority when they enter power weak and
ficient to buy quiescence from exceptionally strong elites, initially weak
leaders remain in power by making themselves even weaker by providing
other elites access to the state in return for their support. Even if the leader
receives a particularly weak draw of power in the first period and is, in the
future, quite strong relative to elites, the need to alleviate commitment
problems in the first period swamps future distributive considerations. As
Montesquieu observed, leaders make decisions about institutions at the
start of their tenure, and these institutional decisions shape the rest of their
rule.
Importantly, this chapter addresses the key question of how certain types
bite in dictatorships? I argue that institutions can credibly constrain leaders
only when they change the underlying distribution of power between
leaders and elites. When an elite is given a key cabinet position, such as
resources that allow him to consolidate his own base of support. Over time,
these positions shift the distribution of power away from the president by
allowing them to better hold incumbents accountable. This mechanism
demonstrates how institutions become self-enforcing.
35
The model also illuminates a counterintuitive argument for power

sharing. The model shows how initially weak autocrats can better secure
their hold on power by giving it away to the very elites who are most
capable of unseating them. By building institutions that empower poten-
tial challengers, the leader hands the (figurative) sword to someone else
while pointing it at herself.
The theory underscores the point that the existence of a democratic
tion. This helps to explain why the presence of nominally democratic
institutions cannot necessarily explain why some regimes are institutio-
nalized systems while others remain personalist dictatorships. This is
especially true when parties or legislatures are empty vehicles that simply
amplify the authority of an incumbent, rather than constraining them.
affect de facto political power.
2.1 i nsti tutions as commitment devices
2.1.1 The Fundamental Problem of Autocratic Rule

One of the main findings from recent scholarship on dictatorships is that
autocratic leaders are plagued by a fundamental paradox: extremely
powerful governments cannot credibly commit to share power with or
distribute rents to elites, opposition groups, or larger segments of society.
This is the challenge that Svolik (2012) refers to as the “problem of
authoritarian power-sharing.” This credibility problem affects many
dimensions of leaders’ rule, and existing scholarship has explored the
various ways in which commitment problems arise.
Commitment problems stemming from absolute power hamper auto-
crats’ abilities to stave off threats that emerge from outside the regime.
Scholars have noted that commitment problems make it difficult for
leaders to credibly promise opposition parties or nonregime elites that
they will receive a steady stream of rents into the future (Blaydes 2010;
Gandhi 2008; Gandhi and Przeworski 2007; Lust 2006). Commitment
problems also prevent leaders from convincing the masses that they will
receive future redistribution when the threat of revolutionary is transitory
(Acemoglu and Robinson 2001; Acemoglu and Robinson 2006).
However, existing research shows that the most critical threats leaders
face generally come from their own inner circle. It is well established that
the majority of autocratic leaders lose power through elite coups, rather
than revolutions. From 1946 to 2008, 68 percent of nonconstitutional exits
36
2.1 Institutions as Commitment Devices
from office of autocrats resulted from coups, compared with only 11 percent
that resulted from popular uprisings (Svolik 2012). For dictators to survive
they must maintain support from their closest allies. My account therefore
focuses on how leaders maintain support from their own regime elites.
Credibility problems affect autocrats’ abilities to maintain support
from inside of the regime (Cox 2016; Gehlbach and Keefer 2011; Haber
2008; Magaloni 2008; North and Weingast 1989; Stasavage 2003). As
Magaloni notes, dictators need to commit to “not abuse their ‘loyal
friends,’ [but] this commitment is hard to establish” (2). Studies that
focus on intra-elite commitment problems argue that power sharing
between leaders and elites is sustained only when elites have mechanisms
that allow them to check predation from leaders. In order for elites to be
able to effectively hold leaders accountable, they must be able to coordi-
nate a viable threat of rebellion (Boix and Svolik 2013; Myerson 2008).
Yet, a viable threat of rebellion from elites is often far from guaranteed in
an autocratic setting.
Most existing accounts of the commitment problem within the leader’s
inner circle argue that a lack of information hinders elite collective action
(Albertus and Menaldo 2012; Boix and Svolik 2013; Gehlbach and Keefer
2011; Myerson 2008). These studies argue that secrecy within authoritar-
ian governments prevent elites from being able to coordinate and com-
municate with each other. The lack of transparency also hinders elites’
abilities to monitor leader compliance. Furthermore, in regimes where
rules and norms are not clearly established, it is difficult for elites to agree
on what exact actions constitute a transgression. In sum, these studies
argue that information problems form the key barrier for elite collective
action, therefore preventing them from effectively keeping the despot’s
absolute power in check.
2.1.2 Institutional Solutions

To address commitment problems stemming from asymmetric informa-
tion, existing scholarship has generally focused on the creation of nomin-
ally democratic institutions that can serve as “forums” for elite
coordination. These studies argue that parties and legislatures promote
transparency and allow elites to interact, fostering coordination among
them (Boix and Svolik 2013; Myerson 2008). Constitutions establish clear
guidelines about what constitutes rule breaking, therefore eliminating
ambiguity that might prevent elites from knowing when to coordinate
on a rebellion (Albertus and Menaldo 2012).
Yet, if parties, legislatures, and constitutions do indeed solve intra-elite
commitment problems, then the introduction of these institutions should be
37
strongly correlated with regime durability. However, as I argue in the intro-

duction of this book, nominally democratic institutions are extremely pre-
valent, even in unstable dictatorships, and therefore cannot fully explain
variation in regime durability. A number of existing studies also show that
the presence of ruling parties is not strongly correlated with regime durability,
and the majority of these parties do not last beyond the tenure of a single
leader. These findings cast doubt on the claim that intra-elite commitment
problems stem primarily from information problems and that these problems
can be solved through the creation of nominally democratic institutions.
By contrast, I argue that intra-elite commitment problems arise due to
dynamic power shifts between leaders and their allies.1 Elites rebel when
they predict that power will shift adversely in the future, so that they will
become weaker if they wait to act. As Albertus and Menaldo (2012) note,
regime elites tend to be at the “apex” of their power at the start of the
regime when they are well organized (292). When leaders are in a state of
vulnerability, elites would prefer to depose the dictator in the current
period when they have the advantage, rather than wait for the distribution
of power to shift against them in the future.
My characterization of the commitment problem as a result of dynamic
power shifts contrasts with existing models of authoritarian politics which
assume that elites cannot coordinate due to static information problems.
Instead, my approach is related to a broad class of formal models that
examine shifting power as a cause of international war (Fearon 1995;
Powell 2006) and models of democratization that result from commit-
ment problems (Acemoglu and Robinson 2006). However, these existing
models do not propose solutions for leaders to relinquish or undercut their
power in order to prevent conflict from occurring in equilibrium.2
If autocratic leaders face intra-elite commitment problems that result
from shifting power, what types of institutional solutions can they imple-
ment to stabilize their rule? I propose a mechanism where leaders take
actions to endogenously shift the future distribution of power by arming
elites with access to state resources. Such access can take the form of
presidential cabinet appointments, such as to the office of the vice president
or minister of defense. Credible executive constraints can also take the form
of formal rules that codify this delegation of power, such as procedures
regarding promotion or leadership succession. Credible executive con-
straints emerge when elites are given real access to power within the state.
1
In fact, as the model will show, commitment problems that arise due to dynamic
power shifts can occur even with full information.
2
In Acemoglu and Robinson’s model of democratization, elites cannot voluntarily
give up power in order to prevent the masses from rebelling.
38
Appointing elites to key cabinet positions, especially over ministries

that control valuable resources, endows them with the ability to shape
policy and target material resources to their supporters and constituents.
This access allows elites to consolidate their own independent power base,
thus shifting the distribution of power away from the leader. The appoint-
ment of elites to powerful positions within government and formalization
of the hierarchy of power in the constitution thus provides elites with de
facto power necessary to sustain long-run promises about rent distribu-
tion. These measures change the future distribution of power, countering
commitment problems that would have otherwise arisen if power had
been allowed to shift unchecked.
A few existing studies of interstate conflict also model leader decisions
to endogenously change the distribution of power (Debs 2014; Powell and
dal Bo 2013). However, these models focus on leaders who take actions to
strengthen or consolidate their power by investing in military capabilities,
rather than weaken themselves by binding their hands. By contrast, I stress
a mechanism in which leaders voluntarily weaken themselves to counter
changes in the future distribution of power that could trigger conflict.
Finally, I also build on studies in comparative political economy that
consider how domestic institutions can tie the leader’s hands and provide
credible commitment. A number of studies argue that the English Crown
was able to credibly commit not to alter property rights only after trans-
ferring control over taxation to parliament (Cox 2016; North and
Weingast 1989). Similarly, in his examination of public debt in early
modern Europe, (Stasavage 2003) notes that it was common for rulers
to “delegate authority with the express intent of improving their credibil-
ity. So for example, a ruler might give a group of officials the right to
manage public revenues so as to ensure full debt repayment” (3). Haber,
Maurer, and Razo (2003) demonstrate that a similar solution was used to
achieve economic growth in revolutionary Mexico by granting key eco-
nomic actors direct access to decision-making inside the state. However,
these studies do not focus on the conditions under which leaders under-
take these institutional choices. Building on these ideas, I specify when
autocratic leaders voluntarily tie their hands in order to provide credible
commitment power.
To sum up, this chapters claims that commitment problems arise when
elites lack the de facto ability to counter future shifts in power. Only after
leaders provide elites with direct access to key government offices can
credible commitments be sustained and self-enforcing. This mechanism
explains why simply creating a focal point by establishing constitutional
rules or allowing elites to communicate within a ruling party are insufficient
in solving intra-elite commitment problems that arise due to shifting power.
39
It is important to note that while this chapter examines why certain types
of executive constraints emerge, I do not seek to explain the creation of
parties or legislatures in dictatorships. In fact, historical evidence often
suggests that many ruling parties are not strategically created, but instead,
inherited by leaders. Within my sample of African countries, for instance,
virtually all leaders came into power with a preexisting party in place
immediately following independence. These ruling parties were created
prior to independence not as elite power-sharing devices, but rather as pro-
independence organizations (Bienen 1970; 1978; Zolberg 1966; 1969) or
parties that were used to participate in preindependence elections (Collier
1982). Similarly, autocratic legislatures do not tend to predate political
parties (Cheibub, Gandhi, and Vreeland 2010). Together, this evidence
suggests that autocratic leaders often do not have total control over the
set of institutions that are available to them when they take power.
Rather than arguing that leaders make strategic decisions to create
institutions, this book argues that leaders make strategic decisions to
institutionalize their regimes after coming into power. Importantly, insti-
tutionalization can occur with or without the presence of formal institu-
tions such as parties or legislatures (though these institutions are very
commonly found). Political parties were banned in Swaziland in 1973,
and this law remains intact today. Despite the absence of parties,
Swaziland has stable prime minister appointments and constitutional
succession rules. Parties were banned in Ethiopia from 1960 through the
early 1980s, but the prime minister and defense minister positions were
filled and remained stable during those years. Equatorial Guinea under the
rule of Teodoro Obiang Nguema Mbasogo did not have a ruling party for
the first six years the leader was in power, but the cabinet had stable vice
president and defense minister appointments. Meanwhile, Niger under
the presidency of Hamani Diori did have a ruling party, the Nigerian
Progressive Party (PPN), but no institutionalized occurred; the regime
did not have formal leadership succession policies and key cabinet
appointments were kept empty.
If the creation of formal organizations such as parties is often histori-
cally determined, do dictators really have control over the content of state
constitutions? Historical accounts of how constitutions were drafted in
the newly independent African states corroborate the claim that leaders –
especially founding presidents – were given extraordinary leeway and
authority to construct their constitutions as they desired. Zolberg’s
account of the constitutional process in the newly independent Ivory
Coast highlights the fact that founding president, Felix Houphouet-
Boigny, had total control over the contents of the document.
40
2.2 A Theoretical Model
When work began on [drafting the constitution], most opposition leaders were
either in exile or had rejoined the [ruling Parti Démocratique de la Côte d’Ivoire
(PDCI) party]. There were no organs of public opinion in which the projects might
be discussed. The constitution of 1959 was therefore framed by a government and
an assembly dominated by a single party, itself controlled by one man. The
decision-making process itself reveals that Houphouet-Boigny and his lieutenants,
accountable only to one another, were unusually free to institutionalize their own
political preferences . . . . The constitution was . . . approved with almost no debate
in the Assembly on March 26, 1959. (Zolberg 1966, 250–251)
Leaders who came after the founding presidents also had control over the
content of state constitutions. Denis Sassou Nguesso, who came to power in
the Republic of Congo in 1979, issued a presidential decree in 1984 that
created a constitutional council. This consultative body (that was, in prac-
tice, accountable only to Sassou Nguesso) determined the constitutionality
of treaties and laws before their adoption by the National People’s
Assembly. At the same time, he then shut down the constitutional chamber
of the Supreme Court (Radu and Sommerville 1989). It is important to note
that even though these leaders may have the opportunity to remove the
constitutional constraints their predecessors implemented, they will not
necessarily do so if it is not in their own best interest.
2.2 a theoretical model

I will now formalize my theory and identify the conditions under which
autocratic leaders will institutionalize their regimes. I first present
a baseline model where the current and future distribution of power is
common knowledge. I show that leaders who enter power weak are more
likely to create constraints in order to remain in power. I then present
a model extension that introduces uncertainty about the future distribu-
tion of power and show that conflict is possible in equilibrium when there
is imperfect information about the leader’s strength in the future.
2.2.1 Model Set Up

Formally, imagine a two-player, two-period bargaining game in which an
Autocrat (A) and a regime Elite (A) divide a set of benefits or “pies”
normalized to size 1.3 I will refer to the Autocrat using a female pronoun
and the Elite using a male pronoun.
3
As I argue in Section 2 of the chapter, intra-elite commitment problems arise due to
shifts in power, rather than information problems that prevent elite coordination.
Since this model does not focus on elite collective action problems, I treat the
coalition of elites as a single player (E) in the model.
41
In the first period, A offers x1 to E, who can accept the division or reject
it. If E accepts A’s offer in that period, then A and E receive payoffs of
1 x1 and x1 , respectively, and A remains in power. The game continues
onto the second period, and A makes a new offer.
If E rejects A’s offer in the first period, then conflict occurs. Elite
defections are known to be one of the primary drivers of authoritar-
ian breakdown, and the removal of support can be extremely danger-
ous for incumbents (Haggard and Kaufman 1995; Reuter and Gandhi
2011). Research on military coups has shown that they rarely succeed
without substantial civilian support. Militaries often oust govern-
ments during periods of crisis when citizens express discontent with
the civilian leader’s incompetence or mismanagement of the economy
(Geddes 2009).
If conflict occurs, A will be deposed with probability pt . We
assume that p1 > p2 so that power shifts against elites in the second
period. Elites are often strongest at the start of the regime when they
are well organized, before the leader has a chance to consolidate
power (Albertus and Menaldo 2012). Conflict ends strategic decision-
making in the game, and the winner receives all future benefits. If
fighting occurs in period 1, then the winner consumes the pie for both
periods. If fighting occurs in period 2, then the winner consumes the
pie for the final period. However, fighting is costly and destroys
a fraction of the pie. If conflict occurs, then only a fraction σ2 ð0; 1Þ
of the pie remains.
At the start of the game A decides whether to institutionalize her regime
by establishing executive constraints. Since credible executive constraints
provide elites with access to the state, we model it as a shift in the future
distribution of power away from the leader. Executive constraints are
represented by the parameter g2 ½0; 1 p2 . If A selects g > 0, then
the second period distribution of power will be p2 þ g. Any positive
value of g will shift the entire distribution of power away from A in
the second period. A’s offer in period 2 is affected by the institutional
decision she makes at the start of the game.4
This setup reflects the fact that most leaders establish constraints near
the start of their tenure. Leaders frequently draft new constitutions after
4
We assume that A does not value g inherently. She does not consume g; it only affects
the extent to which the distribution of p2 shifts. In other words, A only cares about
the results of g, rather than the inherent level of g. This assumption reflects the idea
that leaders do not have a preference ordering about the strategies they use to rule.
Instead, I assume that they care only about maximizing rents and time in office,
rather than the continuation of the regime after their death.
42
taking power, and in the case of postindependence regimes, many amend

existing constitutions adopted from the colonial period at the onset of
their rule (Albertus and Menaldo 2012; Zolberg 1969). As I demonstrate
in Chapter 5, the vast majority of constitutional succession rules and term
limits are created within the first four years of a leader’s tenure.
I also assume that the regime is not institutionalized at the start of the
game and that leaders can only establish constraints, rather than remove
or weaken constraints on executive power. I relax these assumptions in an
extension of the model that is presented in Appendix C at the end of this
chapter. In the modified version of the game I allow for leaders to set
negative values of g, which essentially allows them to remove existing
constraints. However, since we are focusing primarily on the creation of
institutionalized regimes, the baseline model restricts our attention to
positive levels of g.
The game proceeds as follows:
1. At the start of the game, A selects g2 ½0; 1 p2 and E observes this
choice.
2. A offers x1 2 ½0; 1.
3. E accepts or rejects the offer of x1 .
a. If E rejects the offer, conflict occurs. If conflict occurs, A is
deposed with probability p1 and remains in power with prob-
ability 1 p1 . Fighting is costly and only a fraction σ of the pie
remains after fighting. The winner of the fight consumes the
remainder of the pie for both periods, and the loser gets nothing
for both periods.
b. If E accepts the offer, then E receives x1 and A receives 1 x1 .
4. If conflict did not occur in period 1, the game moves on to period 2. A
offers x2 2 ½0; 1.
a. If E rejects the offer then conflict occurs. If conflict occurs, A is
deposed with probability p2 þ g and remains in power with prob-
ability 1 p2 g. Fighting is costly and only a fraction σ of the
pie remains after fighting. The winner of the fight consumes the
remainder of the pie for the second period, and the loser gets
nothing for the second period.
The game ends.
Figure 2.1 presents the game tree.
43
2σ(1 − p1), 2σp1 (1 − x1) + σ(1 − p2 − g),x1 + σ(p2 + g)
reject
reject
1 − p2 1 1
A g A x1 E accept A x2 E accept 2 − (x1 + x2),

x1 + x2
0 0 0
Figure 2.1 Game tree: baseline model
2.3 s olving the baseline model

In the following section we first establish the conditions under which
A will not institutionalize. We then derive equilibrium levels of institutio-
nalization and show that A will prefer to institutionalize when faced with
a commitment problem in period 1. The equilibrium solution concept is
Subgame Perfect Nash equilibrium. The full proofs for the baseline model
are presented in Appendix A at the end of this chapter.
2.3.1 No Institutionalization
When will the leader decide not to institutionalize the regime? In this
section, we show that an autocrat who initially enters power strong will
not face a commitment problem in period 1. She therefore does not need to
establish constraints in order to make an offer that is acceptable to the
elite.
Assume that a commitment problem never exists. If that is the case,
then A can always make an offer xt that can always be accepted in both
periods. In period 1, A makes E indifferent between accepting and reject-
ing an offer by satisfying the following condition:
EUE ðrejectÞ ≤ EUE ðacceptÞ

(1)
2σp1 ≤ x1 þ VE
VE denotes the continuation value of accepting the offer and moving

onto period 2 for E. In the second period, A will hold E down to his
reservation price by offering the expected utility of rejecting (since there
are no future offers to condition on). VE is therefore equal to E’s period 2
expected utility of fighting. Plugging VE into equation (1) allows us to
solve for x1 . Formally, A will make the following offer in period 1:
x1 ¼ maxf0; 2σp1 σp2 g (2)
Whether A will always be able to make this offer depends on her relative
strength in period 1. The largest possible per-period offer A can make is
44
2.3 Solving the Baseline Model
equal to the entire size of the pie, which is normalized to 1.5 Since A cannot
commit to future offers, each per-period offer cannot exceed 1.
Proposition 2.3.1: When p1 ≤ 2σ 1 ^ , then A can always make
þ p22 ≡ p
an offer x1 that can induce an acceptance by E. For all p1 2 ½0; p ^ ,
there exists an xt ≤ 1 such that E½UE ðrejectÞ ≤ E½UE ðacceptÞ.
Proposition 4.1 tells us that if A is strong when she first enters power,
she will not institutionalize. Recall that p1 is the probability that A will be
deposed, therefore when p1 is sufficiently low, A enters power in
a position of strength. Because the probability that E can successfully
depose A is very low, A will be able to make an offer that will match E’s
expected utility of rejecting, and commitment problems will not occur. In
this scenario, A will not institutionalize in equilibrium because she does
not need to in order to sustain peaceful bargaining.
How does the threshold of peaceful bargaining without institutionali-
zation, denoted by p ^ , change relative to the future distribution of power?
Proposition 2.3.2: As A’s period 2 strength decreases, the range for
peaceful bargaining without institutionalization increases.
∂p ^
Formally, ∂p2
> 0.
Put together, Propositions 4.1 and 4.2 produce some interesting counter-
intuitive results. When A enters power strong, she will not institutionalize
because she can make an offer that will satisfy E. In this case, peaceful
bargaining can be sustained without institutionalization, where E has no
guarantees over future rent distribution. However, keeping A’s initial
strength constant, as p2 increases, E’s continuation value also increases.
When E can rely on future de facto power, this puts less pressure on the
period 1 offer. In this case, peaceful bargaining can also be sustained with-
out institutionalization due to a different mechanism. When E becomes
more powerful in period 2, he will not need to fight in period 1 because he is
guaranteed future rent distribution through increasing de facto power.
2.3.2 Institutionalization
Now let’s assume that p1 > p^ . An offer x1 large enough to induce an
acceptance from E cannot be made without some positive level of institu-
tionalization g > 0.
5
If p1 is very small relative to p2 , 2σp1 σp2 can actually be a negative number.
Because offers are restricted to be within ½0; 1 we must restrict x1 to nonnegative
numbers.
45
2.3.2.1 Finding the Equilibrium Level of Institutionalization. In period

1, E will accept an offer only if the following condition is satisfied:

(3)
2σp1 ≤ x1 þ VE
^ , A will always offer x1 ¼ 1.

Lemma 2.3.1: If p1 > p
If peaceful bargaining cannot be sustained without institutionalization,
A will always prefer to set x1 as large as possible in order to take pressure
off g. Not only is x1 a per-period offer with no lasting consequences for
the second period of the game, g is also a less efficient mechanism for
increasing E’s continuation value, compared with x1 .
To find the equilibrium level of institutionalization, g , we first observe
that E’s continuation value is once again equal to his expected utility of
rejecting the offer in period 2. VE is therefore equal to σðp2 þ gÞ. Plugging
VE into equation (3) allows us to solve for the equilibrium level of
institutionalization:
1
g ¼ 2p1 p2 (4)
σ
As long as A sets g ¼ g , she will be able to make an offer x1 ¼ 1 that will
satisfy E in period 1.6 In the appendix we show that A will always be able
to set g ¼ g .7
2.3.2.2 Comparative Statics. We take comparative statics of g with

respect to key parameters of interest.
Proposition 2.3.3: As A’s period 1 level of strength decreases, the

∂g
equilibrium level of institutionalization increases. Formally, ∂p1
> 0.
As p1 increases, A faces a more intense commitment problem in period
1, which increases the need for institutionalization. Interestingly, how-
ever, we find the opposite relationship between the equilibrium level of
institutionalization and the future distribution of power, p2 .
6
Changing the functional form of g does not alter the results substantively. For
instance, if g was inefficient, say some input ^ g would only shift the distribution by
gÞ < ^
f ð^g Þ such that f ð^ g , A will not be more or less likely to institutionalize. See
Appendix A at the end of this chapter for discussion.
7
Recall that g is bounded above by 1 p2 , therefore A can establish the equilibrium
level of institutionalization only if g ≤ 1 p2 .
46

∂g
equilibrium level of institutionalization decreases. Formally, ∂p 2
< 0.
If A is weaker in period 2 this means E has higher levels of de facto
power in the future. This expected high draw of p2 alleviates the need for
institutions to ensure peaceful bargaining.
Finally, we can also consider how the equilibrium level of institutiona-
lization g changes with respect to the cost of fighting, σ.
Proposition 2.3.5: As the cost of fighting decreases, the equilibrium

level of institutionalization increases. Formally, ∂g
∂σ
> 0.
Recall that σ is the portion of the pie that remains after a period of
conflict. Increasing levels of σ suggests that conflict is getting less costly. As
conflict gets less destructive, the period 1 payoff of rejecting an offer
increases because a larger portion of the pie is preserved in the case of
conflict. Under these circumstances, it becomes harder to buy E off, there-
fore requiring higher levels of institutionalization.
2.3.2.3 Determining A’s Equilibrium Behavior. We have derived the

equilibrium level of institutionalization, g , required for peaceful bargain-
ing – but will A always prefer to implement g ¼ g ? We consider the
tradeoffs A faces when she decides whether to institutionalize.
Proposition 2.3.6: (Benefits of Institutionalization). If A sets g ¼ g ,
then conflict does not occur in equilibrium.
Proposition 2.3.7: (Costs of Institutionalization). A’s second period

consumption is decreasing in g. Formally, ∂ð1x
∂g
2Þ
< 0.
Institutionalization comes with costs and benefits for A. On one hand, if
A sets g ¼ g then conflict will not occur in either period of the game. A is
therefore able to pocket the surplus saved from not fighting in her period 2
consumption.8 On the other hand, as g increases, A’s second period
consumption decreases. This is because g shifts the second period distribu-
tion of power in favor of E. The higher g is, the larger x2 must be in order to
induce an acceptance from E.
In the appendix, we demonstrate that A’s expected utility from institu-
tionalizing is larger than her expected utility of fighting in period 1. In
other words, EUA ðg ¼ g Þ ≥ EUA ðg ¼ 0Þ. When faced with a commitment
problem, A will always prefer to institutionalize.
8
Recall that if a commitment problem occurs in period 1, then A will set x1 ¼ 1 and
consume nothing in the first period.
47
g*
g*
No Institution−
No Institution− institutionalization alization
institutionalization alization
^p = 0.86 1 ^p = 0.71 1
P1 P1
Note: σ = .9, p2 = .6 Note: σ = .9, p2 = .3
Figure 2.2 Graphing equilibrium results
Proposition 2.3.8: The equilibrium of the game can be character-

ized as following:
1. No Institutionalization: If p1 ≤ p ^ , A will set g ¼ 0. In period 1,

A will offer x1 ¼ x1 , and in period 2, A will offer x2 ¼ σp2 . In
each round, E will accept each offer if
EUE ðacceptÞ ≥ EUE ðrejectÞ and reject otherwise.
2. Institutionalization: If p1 > p^ , A will set g ¼ g . In period 1, A
will offer x1 ¼ maxf0; x1 g, and in period 2, A will offer
x2 ¼ σp2 . In each round, E will accept each offer if
Figure 2.2 presents the equilibrium results graphically. The graphs
show that institutionalization occurs only when the autocrat enters
power weak. Leaders do not institutionalize when the probability of
being deposed in the first period is sufficiently low. The figure also illus-
trates that dictatorships become more institutionalized as the probability
of deposing the leader in period 1 increases.
2.3.3 Discussion
The model makes a number of important predictions, which we summar-
ize in this section. First, the model shows that there are two different types
of autocratic rule, which differ based on the leader’s relative strength
when she first comes into office. Leaders who enter office initially strong
are never incentivized to institutionalize because their initial likelihood of
being deposed is very low. Such leaders prefer not to empower elites by
providing them access to the state because they are always able to make
a per-period transfer that elites will accept. As a result, peaceful bargain-
ing can be sustained in a no-institutionalization equilibrium where
48
initially strong autocrats remain in power but do not provide elites with
access to the state.
On the other hand, leaders who enter office initially weak are incenti-
vized to institutionalize because they cannot sustain peaceful bargaining
without shifting the future distribution of power in favor of elites.
Empowering elites in the second period relaxes demands on the first
period transfer by raising the elite continuation value, which allows
initially weak autocrats to made credible future promises to elites.
Although doing so weakens the leader in the future, she is willing to
institutionalize in order to remain in power for both periods of the
game. In this case, peaceful bargaining can also be sustained in an insti-
tutionalization equilibrium where initially weak autocrats remain in
power only by providing elites with access to the state.
These two different types of autocratic rule have one very important
feature in common: both initially strong leaders and initially weak leaders
who institutionalize remain in power for both periods of the game. In
other words, conditional on the weak type institutionalizing, the leader
will be able to remain in power for the same length of time as the initially
strong type.
This feature highlights an important empirical point, which we will
return to in Chapter 6: scholars should not assume that regimes that are
long-lived have strong institutions. Leader tenure does not serve as
a useful proxy for the quality of institutions due to the strategic nature
of institutionalization. Initially weak leaders are incentivized to institu-
tionalize in order to remain in power because they would otherwise face
commitment problems in bargaining and risk being deposed by elites.
When weak leaders institutionalize, they can indeed remain in power for
longer periods of time.
Initially strong leaders, on the other hand, can remain in power regard-
less of whether they have strong institutions or not, because they do not
face commitment problems in bargaining. In fact, as the model shows,
leaders who enter power strong do not institutionalize yet are able to
remain in power for both periods of the game. We should therefore expect
to see a systematic relationship between leader tenure and institutional
strength only when we condition on initial leader strength. This may help
to explain why existing empirical studies, which do not condition on
leader strength have found inconsistent relationships between the pre-
sence of strong institutions and leader tenure (for instance, see Gandhi
2008; Lucardi 2017; Smith 2005).
These results provide some interesting contrasts with findings from
existing models. In the Acemoglu and Robinson (2006) model, a main
finding is that democratization is most likely to occur if the poor pose
49
a credible threat of rebellion infrequently (loosely speaking, when the

leader is frequently strong). By contrast, in my model, constrained rule
is most likely to occur when the leader enters power weak.9 What
accounts for this difference? In the Acemoglu and Robinson model, the
poor can stage a revolution only when nature draws a low cost of rebellion
(in the language of the A & R model, when μ ¼ μH ). Furthermore, if the
poor stage a revolution, it is guaranteed to succeed. In other words, when
the poor choose to rebel, they are guaranteed a postrevolutionary income
of 1 μ in every future period. Because of this, in a world where the poor
are very unlikely to hold a credible threat of rebellion (in the language of
the A & R model, when q is low), that makes periods where they can stage
a rebellion extremely valuable. Therefore, when the poor pose a credible
threat of rebellion infrequently, they would prefer to revolt whenever they
can because the probability of being able to do so is very low in the future.
By contrast, in my model, elites can always remove support of the
autocrat; they are not, by assumption, constrained to rebel only in periods
where the autocrat is weak. Furthermore, when elites initiate conflict, they
are not guaranteed to win. Therefore an elite who has a temporarily good
draw of pt does not feel compelled to rebel against the autocrat as long as
the autocrat can make an offer xt that can satisfy the elite.10
2.4 model e xtension: uncertainty about

the f uture
In the baseline model, the distribution of power is common knowledge.
Both players observe p1 and p2 at the start of the game. I now relax this
assumption and introduce uncertainty about the future distribution of
power, p2 . I show that, consistent with the baseline model, leaders will
institutionalize when they enter power initially weak. However, in the
modified version of the game, conflict is possible in equilibrium when
leaders cannot establish a high enough level of institutionalization in order
to satisfy elites.
At the start of the modified game, p1 is common knowledge but p2 is
not observed by either player. Both players know the distribution from
which p2 will be drawn, but neither knows what the precise value of p2
will be. If the game moves onto the second period, Nature selects p2 and
9
Note that in the Acemoglu and Robinson model, elites are the analogous player as
the autocrat (A) in my model, and the poor are the analogous player as the elite (E)
in my model.
10
Unlike in the Acemoglu and Robinson model, where the probability that elites will
be unseated jumps discontinuously from 0 to 1 if the poor initiate a revolution, pt is
a continuous parameter in my model.
50
2.4 Model Extension: Imperfect Information
both players observe this draw before bargaining occurs. p2 is uniformly

distributed on ½pm μ; pm þ μ, so that pm represents the mean draw of
p2 .11 At the start of the game, A decides whether to institutionalize in light
of uncertainty surrounding what the future distribution of power will be.
While it is reasonable to expect that leaders and elites have accurate
information about their relative strength in the current period, uncertainty
surrounding the future distribution of power can easily arise in dictator-
ships. Positive or negative economic shocks, changes in foreign aid, or
shifts in international norms can shift the distribution of power in favor of
or against the leader in future periods. Thus, leader strength can be
affected by external shocks that are difficult to predict when they first
take power.
At the start of the modified game, A decides whether to institutiona-
lize. This decision is represented by the parameter g2 ½0; 1 ðpm þ μÞ. If
A selects g > 0, then p2 will be drawn from a modified uniform distribu-
tion of ½pm μ þ g; pm þ μ þ g. Any positive value of g will shift the
entire distribution of power away from A in the second period. If A sets
g ¼ 0, then p2 will be drawn from the original uniform distribution,
p2 2 ½pm μ; pm þ μ.
Figure 2.3 illustrates how g shifts the future distribution of power.
The modified game proceeds as follows:
1. At the start of the game, A selects g2 ½0; 1 ðpm þ μÞ and
E observes this choice.
2. A offers x1 2 ½0; 1.
µ µ
pm − µ pm pm + µ
0 1
µ µ
pm − µ + g pm + g pm + µ + g
0 1
g ∈[0,1 − (pm + µ)]
Figure 2.3 Distribution of p2
11
We assume that μ is sufficiently small, such that pm μ > 0 and pm μ < 1.
51

for both periods.
4. If conflict did not occur in period 1, the game moves on to period 2.
Nature selects p2 2 ½pm μ þ g; pm þ μ þ g. Both players observe this
draw.
5. A offers x2 2 ½0; 1.
The game ends.
Figure 2.4 presents the game tree of the modified game.
When we introduce uncertainty about the future distribution of
power, one important difference emergences: conflict is possible in
equilibrium. In the baseline model, autocrats are able to implement
equilibrium levels of institutionalization in order to ensure peaceful
bargaining. As we demonstrated, weak autocrats will always institu-
tionalize when faced with a commitment problem, therefore conflict
never occurs in equilibrium.
We now demonstrate that when the players face uncertainty about the
future distribution of power, p2 , A will not always be able to establish
sufficiently high levels of institutionalization to ensure peaceful bargaining.
2σ(1 − p1), 2σp1 (1 − x1) + σ(1 − p2), x1 + σp2

reject
reject
1 − (pm + µ) 1 pm + µ + g 1
A g A x1 E accept N p2 A x2 E accept 2 − (x1 + x2),

x1 + x2
0 0 pm − µ + g 0
Figure 2.4 Game tree: model extension
52
As a result, fighting can occur in period 1. The full proofs for the modified
game are presented in Appendix B at the end of the chapter.
Similar to the baseline model, when p1 is sufficiently low, A will not
institutionalize because she is always able to make an offer x1 that E will
accept. However, when A enters power weak, a commitment problem will
arise in period 1 if g ¼ 0. We can establish that there exists some g > 0
that would allow A to make an offer x1 ¼ 1 that can satisfy E in period 1.
For this game, g ¼ 2p1 pm 1σ .
Will A always be able to set x1 ¼ 1? In the modified version of the
game, g is bounded above by 1 ðpm þ μÞ. Therefore g is feasible only if
g ≤ 1 ðpm þ μÞ.
Proposition 2.4.1: As long as p1 is sufficiently small, A will be able
to set g ¼ g in order to allow a peaceful bargain to go through in
period 1. Formally, as long as p1 ≤ 12 þ 2σ1
μ2 ≡ p, there exists a
g ¼ g such that E½UE ðrejectÞ ≤ E½UE ðacceptÞ.
A key implication of Proposition 2.4.1 is that conflict will occur in
equilibrium if p1 is not sufficiently small. When p1 > p, even if
A were to set g to the highest possible level, she still cannot prevent
conflict from occurring in period 1 because the level of institutiona-
lization (g ) necessary to allow all bargains to go through peacefully
is larger than the maximum possible value of g. We establish in
the appendix that A still always prefers to institutionalize when
faced with a commitment problem, however, now she cannot always
do so.
Proposition 2.4.2: The equilibrium of the modified game can be
characterized as following:
^ , A will set g ¼ 0. In period 1,
1. No Institutionalization: If p1 ≤ p

A will offer x1 ¼ x1 , and in period 2, A will offer x2 ¼ σp2 .
In each round, E will accept each offer if
^ and p1 ≤ p, A will set g ¼ g . In
2. Institutionalization: If p1 > p
period 1, A will offer x1 ¼ maxf0; x1 g, and in period 2, A will

offer x2 ¼ σp2 . In each round, E will accept each offer if

3. Conflict: If p1 > p ^ and p1 > p, A will set g to any
g2 ½0; 1 ðpm þ μÞ. In period 1, A will offer any x1 2 ½0; 1,
and in period 2, A will offer x2 ¼ σp2 . In each round, E will
accept each offer if EUE ðacceptÞ ≥ EUE ðrejectÞ and reject
otherwise.
53
The introduction of imperfect information regarding the future distribu-

tion of power produces an important prediction: conflict is possible in
equilibrium. If A is unable to implement a high enough level of institutio-
nalization, then E will always reject the period 1 offer, and A will be unable
to prevent fighting from occurring in equilibrium. Unlike the baseline model,
where strong and weak leaders can remain in power for similar lengths of
time, the modified version of the game predicts that some dictatorships will
be short-lived.
2 . 5 co n c l u s i o n
This chapter sought to understand why we see differences in the organiza-
tional capacity of authoritarian regimes by examining the creation of
constraints limiting executive power. Constraints bind only when they
provide elites with access to important government positions, empowering
them to hold the incumbent accountable to promises about rent distribu-
tion. Through a formal model I demonstrated that strong autocrats who
enter power with a low probability of being deposed are less likely to
institutionalize. Initially weak autocrats without such guarantees of stabi-
lity are more likely to pursue institutionalization in order to maintain
support from elites. Importantly, rather than assuming that institutions
constrain leaders, I show how institutions can provide credible commit-
ment power by shifting the future distribution of power in favor of elites.
The next chapter of the book illustrates the primary findings of the model
through two illustrative case studies of Cameroon and Côte d’Ivoire.
Appendix A Proofs: Baseline Model
Proof of Proposition 2.3.1. A must make an offer x1 ¼ max

f0; 2σp1 σp2 g in order to induce an acceptance from E. However, A
faces a budget constraint of 1, the size of the entire pie. Under what
conditions does the optimal offer required not exceed the size of the entire
pie?
x1 ≤ 1
2σp1 σp2 ≤ 1 (5)
1 p2
p1 ≤ þ ≡ p ^
2σ 2
As long as the first draw of p1 is sufficiently small, A will always be able to
make an offer x1 that can induce an acceptance from E.
54
2.5 Conclusion
^ also
Proof of Proposition 2.3.2. It is easy to see that as p2 increases, p
increases:
^
∂p 1
¼ >0 (6)
∂p2 2
Proof of Lemma 2.3.1. A must choose values of x1 and g such that the
following equation is satisfied:
2σp1 ≤ x1 þ VE (7)
What is VE ? We know that one of two things must happen in period 2. It is

possible A cannot make an offer that satisfies E, and E decides to reject the
offer. If this happens, then E’s continuation value is equal to his expected
utility of fighting. The only other possible outcome is that A can make an
offer that satisfies E in period 2. However, A will always try to make the
cheapest possible offer to E, which is exactly his expected utility of fight-
ing. Therefore we know that VE is simply equal to E’s expected utility of
fighting in period 2. Plugging in E’s expected utility of fighting into the
equation produces the following inequality:
2σp1 ≤ x1 þ σEVðp2 Þ
2σp1 ≤ x1 þ σðp2 þ gÞ (8)
2σp1 σp2 ≤ x1 þ σg
Since each unit of g is weighted by σ, it is more efficient to increase x1 in

order to satisfy the inequality, rather than g.
To solve for g , we first observe that A must ensure that VE is large
enough in order to satisfy the following equation:
2σp1 ≤ 1 þ VE (9)
To determine the continuation value, note that if the game moves peace-
fully onto the second period, there will never be fighting. Even if p2 ¼ 1
and σ ¼ 1, A can always offer x2 ¼ 1. Therefore we know that VE is
simply equal to E’s expected utility of rejecting in period 2, since A will
always try to make the cheapest possible offer to E.
VE ¼ EUE ðrejectÞ
¼ σEVðp2 Þ (10)
¼ σðp2 þ gÞ
We now plug E’s continuation value into the equation and solve for g.
55
2σp1 1 ≤ VE
2σp1 1 ≤ σðp2 þ gÞ (11)
1
g ¼ 2p1 p2
σ
We can establish that changing the functional form of g does not alter the
results substantively. Instead of assuming that wðgÞ ¼ g, let’s assume that
institutions are extremely efficient, such that hðe gÞ > e g (in other words,
hð Þ is concave). How would e g compare with g ? We know that
g ¼ 2p1 p2 1σ ¼ hðe gÞ > e
g : Therefore, g > e g . Unsurprisingly,
when institutions are efficient, lower levels of institutionalization are
required to sustain peaceful bargaining.
Interestingly, however, this does not change the threshold, p ^ of insti-
tutionalization, nor does it make A more or less willing to institutionalize,
compared with when wðgÞ ¼ g. First, note that g does not affect the
calculation of the threshold, p ^ . Second, recall that A does not value g
inherently. She does not consume g, it only affects the extent to which the
distribution of p2 shifts. In other words, A only cares about the results of g,
rather than the inherent level of g. Therefore, even if g was inefficient, say
gÞ < ^
if f ð^ g , A will still always be willing to institutionalize.
Proof. We show that A will always be able to set g ¼ g . Recall that g is
bounded above by 1 p2 , therefore A can establish the equilibrium level
of constraints if g ≤ 1 p2 .
g ≤ 1 p2
1
2p1 p2 ≤ 1 p2 (12)
σ
1 1
p1 ≤ þ
2 2σ
Since σ2 ½0; 1, equation (12) is always true.
Proof of Proposition 2.3.3. It is easy to see that as p1 increases, g also
increases:
∂g
¼2>0 (13)
∂p1
Proof of Proposition 2.3.4. It is easy to see that as p2 increases, g decreases:

∂g
¼ 1 < 0 (14)
∂p2
Proof of Proposition 2.3.5. It is easy to see that σ increases as g increases.
56
2.5 Conclusion
∂g 1
¼ 2>0 (15)
∂σ σ
Proof of Proposition 2.3.6. This proof follows directly from the construc-
tion of g , which is the minimal level of g that guarantees that the follow-
ing condition is true: EUE ðrejectÞ ≤ EUE ðacceptÞ. If g ≥ g , then E will
always accept in period 1. If the game makes it to period 2, then conflict
will never occur because p2 < 1 by assumption, therefore A will always be
able to make an offer x2 that can induce an acceptance by E.
Proof of Proposition 2.3.7. A’s second period consumption is simply
1 x2 , since she can always make an offer x2 that can induce an accep-
tance by E. It is easy to see that 1 x2 is decreasing in g:
∂ð1 x2 Þ
¼ σ < 0 (16)
∂g
Proof. We show that A will therefore always prefer to institutionalize if

^ . First we observe that A will set g ¼ g in equilibrium only if the
p1 > p
following condition holds:
EUA ðg ¼ g Þ ≥ EUA ðg ¼ 0Þ (17)
The expected utility of A not institutionalizing is equal to the expected

utility of E rejecting the offer x1 . This can be expressed as follows:
EUA ðg ¼ 0Þ ¼ EUA ðE rejectsÞ

(18)
¼ 2σð1 p1 Þ
If A chooses to institutionalize, she will set g ¼ g and x1 ¼ 1. As long as A

sets g ¼ g , E will accept the period 1 offer.
EUA ðg ¼ g Þ ¼ EUA ðE acceptsÞ

¼ ð1 x1 Þ þ VA (19)
¼ 0 þ VA
A gets to pocket the portion of the pie that she doesn’t offer to E, therefore
her continuation value is the size of the pie minus the expected value of the
period 2 offer, x2 .
VA ¼ 1 EVðx2 Þ (20)
In period 2, A will make the cheapest possible offer to E, so in expectation,

x2 will be equal to the expected value of fighting for E.
57
EVðx2 Þ ¼ σðp2 þ g Þ (21)
Plugging g into equation (21) produces the following:
EVðx2 Þ ¼ 2σp1 1 (22)
We plug this back into A’s continuation value:
VA ¼ 1 EVðx2 Þ
(23)
¼ 2 2σp1
To verify that A will always prefer to institutionalize, we check whether

A’s expected utility of setting g ¼ g is larger than A’s expected utility of
setting g ¼ 0:
EUA ðg ¼ g Þ ≥ EUA ðg ¼ 0Þ
2 2σp1 ≥ 2σð1 p1 Þ (24)
1≥σ
Equation (24) is always true by assumption. A will therefore always prefer

^.
to institutionalize if p1 > p
Proof of Proposition 2.3.8. We break the proof of Proposition 2.3.8 into
two parts. First we establish the No Institutionalization equilibrium.
^ then A can always
Proposition 2.3.1 has already established that if p1 ≤ p
make an offer x1 that can induce an acceptance from E in period 1. Therefore
in the No Institutionalization equilibrium, A’s best response is to set g ¼ 0.
In the second period of the game, A’s strict best response is to offer
x2 ¼ σp2 because doing so allows her to pocket the surplus saved from not
fighting while offering the smallest possible amount that will induce an
acceptance from E. E’s best response is to accept an offer that is at least as
good as his expected utility of fighting in the second period.
Moving to the first period of the game, A will always choose to set
x1 ¼ x1 to ensure peaceful bargaining, rather than choosing to fight
because her expected utility from fighting is strictly less.
A’s expected utility from peaceful bargaining is equal to
2 ðx1 þ x2 Þ ¼ 2 2σp1 . A’s expected utility from fighting in period 1
is equal to 2σð1 p1 Þ. We can show that A’s expected utility from peace-
ful bargaining is higher than her expected utility from fighting in period 1:
EUA ðx1 < x1 Þ < EUA ðx1 ¼ x1 Þ

2σð1 p1 Þ < 2 2σp1 (25)
σ<1
58
2.5 Conclusion
Since σ ≤ 1 by assumption, equation (25) is always true. Therefore A’s

strict best response is to offer x1 ¼ x1 in period 1. Once again, E’s best
response is to accept x1 because it is, by construction, the smallest possible
offer that can induce an acceptance by E in period 1. We have therefore
established a unique equilibrium when p1 ≤ p ^.
Now we establish the Institutionalization equilibrium. Here, we assume
that p1 > p ^ (otherwise we would be in the No Institutionalization equili-
brium) and that a peaceful offer cannot be made in period 1 without setting
g ¼ g .
If the game reaches a second period of bargaining, then A can always
make an offer that will satisfy E. A’s strict best response in period 2 is to
offer x2 ¼ σp2 because doing so allows her to pocket the surplus saved
from not fighting while offering the smallest possible amount that will
induce an acceptance from E. E’s best response is to accept an offer that is
at least as good as his expected utility of fighting in the second period.
Moving to period 1, A can make an offer x1 ¼ 1 that will ensure
peaceful bargaining as long as g ≥ g , or she will not be able to make any
offer that will satisfy E if g < g . We show that if A can make an offer
x1 ¼ 1, given that g ¼ g , she will choose to do so, rather than choosing to
fight.
We have already established from Lemma 2.3.1 that if p1 > p ^ , then A will
always offer x1 ¼ 1. If she chooses to do this, then her total expected utility
over the two periods is simply the expected utility of 1 x2 , since she receives
nothing in period 1. We show that EUA ðx1 < 1Þ < EUA ðx1 ¼ 1Þ.
A’s expected utility from fighting in period 1 is equal to 2σð1 p1 Þ. To
calculate EUA ð1 x2 Þ, we first establish x2 given that g ¼ g . We know
that x2 ¼ EVðp2 Þσ.
σðp2 þ g Þ
EVðx2 Þ ¼
1
¼ σp2 þ σ 2p1 p2 (26)
σ
¼ 2σp1 1
A’s two-period expected utility from peaceful bargaining is equal to

1 EVðx2 Þ ¼ 2 2p1 σ. We show that this is strictly larger than A’s
expected utility from fighting in period 2.
EUA ðx1 < 1Þ < EUA ðx1 ¼ 1Þ

2σð1 p1 Þ < 2 2σp1 (27)
σ<1
Once again, since σ ≤ 1 by assumption, equation (27) is always true.

Therefore, it is always the case that setting x1 ¼ 1 produces a larger
59
expected utility for A than fighting in period 1. A’s strict best response,
given that g ¼ g is to offer x1 ¼ 1 in period 1. E’s best response is to
accept x1 ¼ 1 because by construction, g ensures that E’s expected utility
of accepting x1 ¼ 1 is greater than or equal to his expected utility of
fighting in period 1.
We now move to the very start of the game, where A decides what to set
g. We have already established that the following is always true:
EUA ðinstitutionalizeÞ ≥ EUA ðnot institutionalizeÞ. Therefore if p1 > p^,

then A’s best response is to set g ¼ g at the onset of the game.
Appendix B Proofs: Model Extension

We demonstrate that, similar to the baseline model, when A enters power
sufficiently strong, she will not establish constraints because she will always
be able to make an offer x1 that E will accept. Unlike the baseline model,
when we introduce uncertainty over the future distribution of power p2 ,
conflict is possible in equilibrium when A cannot establish a high enough
level of constraints because g > maxðgÞ.
First we establish that when A enters power sufficiently strong, she does
not face a commitment problem in bargaining, and therefore does not
institutionalize. Assume that a commitment problem never exists. If that is
the case, then A can always make an offer xt that can always be accepted in
both periods. In period 1, A makes E indifferent between accepting and
rejecting an offer by satisfying the following condition:

(28)
2σp1 ≤ x1 þ VE
In the second period, A will hold E down to his reservation price by

offering the expected utility of rejecting. VE is therefore equal to E’s period 2
expected utility of fighting. E’s expected utility of fighting is equal to σpm .
Plugging VE into the equation above allows us to solve for x1 . Formally,
A will make the following offer in period 1:
x1 ¼ maxf0; 2σp1 σpm g (29)
strength in period 1. Since A cannot commit to future offers, the largest
possible period 1 offer she can make is x1 ¼ 1. We plug in x1 ¼ 1 and
solve for p1 . When p1 ≤ 2σ
1 ^ , then A can always make an offer x1
þ p2m ≡ p
that can induce an acceptance by E.
60
2.5 Conclusion
^ , A does not face a commit-

Similar to the baseline model, when p1 ≤ p
ment problem in bargaining, and therefore does not institutionalize.
However, when p1 > p ^ , A cannot make a period 1 offer that E will
accept if g ¼ 0. A needs to set g > 0 in order to prevent conflict from
occurring in period 1.
In period 1, E will accept an offer only if the following condition is
satisfied:

(30)
2σp1 ≤ 1 þ VE

that E’s continuation value is, once again, equal to his expected utility of
rejecting the offer in period 2. VE is therefore equal to σðpm þ gÞ. Plugging
VE into the equation above allows us to solve for the equilibrium level of
1
g ¼ 2p1 pm (31)
σ
satisfy E in period 1. However, we show that A will not always be able to
set g at a high enough level if g > maxðgÞ.
Proof of Proposition 2.4.1. We established above that as long as A sets
g ¼ g , she will be able to make an offer, x1 , that E will accept and conflict
will not occur in period 1. However, g is bounded above by 1 ðpm þ μÞ
because it is drawn from the following distribution: g2 ½0; 1 ðpm þ μÞ.
To see why g is bounded above by 1 ðpm þ μÞ, refer to Figure 2.3. g
shifts the entire distribution of p2 upward, and the largest value the upper
bound of p2 can take is 1. Therefore the largest possible value g can take is
the distance between 1 and the upper bound of the original distribution of
p2 , pm þ μ.
A can set g ¼ g only when g ≤ 1 ðpm þ μÞ.
g ≤ 1 ðpm þ μÞ
1
2p1 pm ≤ 1 ðpm þ μÞ (32)
σ
1 1 μ
p1 ≤ þ ≡p
2 2σ 2
A can set g ¼ g only when p1 ≤ p. When p1 > p, even if A were to set g to
the highest possible level, it would not be high enough to prevent conflict
in period 1.
61
Proposition 2.4.2. Proposition 2.4.2 establishes the following equili-

brium: No Institutionalization, Institutionalization, and Conflict. First,
we note that the No Institutionalization and Institutionalization equili-
brium follow an identical logic as the No Institutionalization and
Institutionalization equilibrium that was established in the baseline
model version of the game. The reader can therefore refer to the proof
of Proposition 2.3.8 for these equilibria.
We now address the Conflict equilibrium. If the game reaches a second
period of bargaining, then A can always make an offer that will satisfy E,
since p2 ≤ pm þ μ, which is strictly less than the total size of the pie. A’s
strict best response in period 2 is to offer x2 ¼ σp2 because doing so allows
her to pocket the surplus saved from not fighting while offering the
smallest possible amount that will induce an acceptance from E. E’s best
response is to accept an offer that is at least as good as his expected utility
of fighting in the second period.
Moving to period 1, we have already established through Proposition
2.3.1 that A will not be able to set g high enough to induce an acceptance
from E in period 1. In other words, conflict is inevitable in equilibrium if
p1 > p. Since this is the case, any offer x1 2 ½0; 1 is a best response, since
no offer will be able to satisfy E in period 1. Therefore in the conflict
equilibrium, A can offer any x1 2 ½0; 1. E’s best response is to accept x1
only if EUE ðacceptÞ ≥ EUE ðrejectÞ. Since this equation will never be satis-
fied in the conflict equilibrium, E will always reject the offer.
Moving back to the start of the game, since p1 > p, then

g > 1 ðpm þ μÞ. In other words, no level of institutionalization will
be high enough to prevent conflict. Since this is the case, any level of
institutionalization g 2 ½0; 1 ðpm þ μÞ is a best response, since no level
of g is high enough to prevent conflict. A can therefore select any
g 2 ½0; 1 ðpm þ μÞ.
Appendix C Model Extension: Removal of Institutionalization

In the baseline model, we assumed that the regime is not institutionalized
at the start of the game. The autocrat’s decision is limited to positive
values of g. Now we relax the assumption that the regime is not
institutionalized when the autocrat first comes into power, and allow
the autocrat to de-institutionalize.
In this extension of the model we allow for A to remove constraints by
allowing A to set negative levels of g. Negative levels of g shifts power in
favor of A in the second period. In other words, A can remove existing
constraints by setting negative levels of g and decreasing her average
62
2.5 Conclusion
probability of being deposed in the second period. In the modified version

of the game, A can select g from a distribution of g 2 ½p2 ; 1 p2 , and
everything else remains the same as the baseline model.
First, we establish that A will not choose to de-institutionalize unless
p1 < p^ . Recall from the baseline model that if p1 > p ^ , she will need to
institutionalize in order to maintain peaceful bargaining. We therefore
^ . In period 1, A needs to make
restrict our attention to cases where p1 < p
sure the following condition holds:

(33)
2σp1 ≤ x1 þ VE
When p1 < p ^ , then A can set x1 ¼ 1 and g ¼ 0. However, if p1 < p ^ , then

the left hand side of equation (33) will be strictly less than the right hand
side of the equation. In that case, then A can offer x1 < 1 or set g < 0.
Observe that A will prefer to offer x1 < 1 over setting negative levels of g,
since g is weighted by σ. In other words, A will prefer to keep scaling down
x1 , rather than setting g < 0 because it is more efficient. However, if A sets
x1 ¼ 0 and the LHS of the equation is still strictly less than the RHS of the
equation, then A will want to remove institutions by setting negative levels
of g. When we plug x1 ¼ 0 into equation (33), we see that g ¼ 2p1 p2 .
Finally, we note that, unlike in the baseline model, incomplete removal
can occur in equilibrium. Recall that in the baseline model where g > 0, A
had to set g ¼ g when p > p ^ , otherwise E would reject the period 1 offer
and conflict will occur in equilibrium. However, when A sets g < 0, E no
longer has veto power. In fact, A is removing constraints precisely because
she can satisfy equation (33) even when x1 ¼ 0. Recall that g is bounded
below by p2 , since it is drawn from the distribution g 2 ½p2 ; 1 p2 . In
cases where negative levels of g are actually more negative than
minðgÞ ¼ p2 , A cannot remove constraints as much as she wants to.
However, unlike in the case of g > 0, incomplete removal will not trigger
conflict. Therefore if g < 0, then A will set g ¼ minfp2 ; g g.
63
2


take such measures?
34

rule.
35


36

37

1
2
38

commitment power.
39
40

2.2.1 Model Set Up

3
41
(Geddes 2009).
of the pie remains.
4
42

choice.
2. A offers x1 2 ½0; 1.
for both periods.
offers x2 2 ½0; 1.
The game ends.
43
2σ(1 − p1), 2σp1 (1 − x1) + σ(1 − p2 − g),x1 + σ(p2 + g)
reject
reject
1 − p2 1 1

x1 + x2
0 0 0

elite.

(1)
2σp1 ≤ x1 þ VE

x1 ¼ maxf0; 2σp1 σp2 g (2)
44
þ p22 ≡ p
∂p ^
Formally, ∂p2
> 0.
5
numbers.
45


(3)
2σp1 ≤ x1 þ VE

1
g ¼ 2p1 p2 (4)
σ
to set g ¼ g .7


∂g
> 0.
6
gÞ < ^
7
46

∂g
< 0.

∂σ
> 0.


∂g
2Þ
< 0.
8
47
g*
g*
No Institution−
^p = 0.86 1 ^p = 0.71 1
P1 P1
Note: σ = .9, p2 = .6 Note: σ = .9, p2 = .3

ized as following:

2.3.3 Discussion
48
strong type.
2008; Lucardi 2017; Smith 2005).
49


the f uture
to satisfy elites.
9
in my model.
10
50

take power.
2. A offers x1 2 ½0; 1.
µ µ
0 1
µ µ
pm − µ + g pm + g pm + µ + g
0 1
g ∈[0,1 − (pm + µ)]
11
51

for both periods.
draw.
5. A offers x2 2 ½0; 1.
The game ends.
2σ(1 − p1), 2σp1 (1 − x1) + σ(1 − p2), x1 + σp2

reject
reject
1 − (pm + µ) 1 pm + µ + g 1

x1 + x2
0 0 pm − µ + g 0
52
do so.



otherwise.
53

be short-lived.

pie?
x1 ≤ 1
2σp1 σp2 ≤ 1 (5)
1 p2
p1 ≤ þ ≡ p ^
2σ 2
54
2.5 Conclusion
^ also
increases:
^
∂p 1
¼ >0 (6)
∂p2 2
2σp1 ≤ x1 þ VE (7)


2σp1 ≤ 1 þ VE (9)
¼ σEVðp2 Þ (10)
¼ σðp2 þ gÞ
55
2σp1 1 ≤ VE
2σp1 1 ≤ σðp2 þ gÞ (11)
1
g ¼ 2p1 p2
σ
g ¼ 2p1 p2 1σ ¼ hðe gÞ > e
gÞ < ^
g ≤ 1 p2
1
2p1 p2 ≤ 1 p2 (12)
σ
1 1
p1 ≤ þ
2 2σ
increases:
∂g
¼2>0 (13)
∂p1

∂g
¼ 1 < 0 (14)
∂p2
56
2.5 Conclusion
∂g 1
¼ 2>0 (15)
∂σ σ
∂ð1 x2 Þ
¼ σ < 0 (16)
∂g

p1 > p


(18)
¼ 2σð1 p1 Þ


¼ ð1 x1 Þ þ VA (19)
¼ 0 þ VA
VA ¼ 1 EVðx2 Þ (20)

57
EVðx2 Þ ¼ 2σp1 1 (22)
VA ¼ 1 EVðx2 Þ
(23)
¼ 2 2σp1

setting g ¼ 0:
2 2σp1 ≥ 2σð1 p1 Þ (24)
1≥σ

^.
^ then A can always

2σð1 p1 Þ < 2 2σp1 (25)
σ<1
58
2.5 Conclusion

g ¼ g .
fight.
σðp2 þ g Þ
EVðx2 Þ ¼
1
¼ σp2 þ σ 2p1 p2 (26)
σ
¼ 2σp1 1


2σð1 p1 Þ < 2 2σp1 (27)
σ<1

59



(28)
2σp1 ≤ x1 þ VE

þ p2m ≡ p
60
2.5 Conclusion

satisfied:

(30)
2σp1 ≤ 1 þ VE

1
g ¼ 2p1 pm (31)
σ
p2 , pm þ μ.
1
2p1 pm ≤ 1 ðpm þ μÞ (32)
σ
1 1 μ
p1 ≤ þ ≡p
2 2σ 2
in period 1.
61


g 2 ½0; 1 ðpm þ μÞ.

62
2.5 Conclusion


(33)
2σp1 ≤ x1 þ VE

63
2


take such measures?
34

rule.
35


36

37

1
2
38

commitment power.
39
40

2.2.1 Model Set Up

3
41
(Geddes 2009).
of the pie remains.
4
42

choice.
2. A offers x1 2 ½0; 1.
for both periods.
offers x2 2 ½0; 1.
The game ends.
43
2σ(1 − p1), 2σp1 (1 − x1) + σ(1 − p2 − g),x1 + σ(p2 + g)
reject
reject
1 − p2 1 1

x1 + x2
0 0 0

elite.

(1)
2σp1 ≤ x1 þ VE

x1 ¼ maxf0; 2σp1 σp2 g (2)
44
þ p22 ≡ p
∂p ^
Formally, ∂p2
> 0.
5
numbers.
45


(3)
2σp1 ≤ x1 þ VE

1
g ¼ 2p1 p2 (4)
σ
to set g ¼ g .7


∂g
> 0.
6
gÞ < ^
7
46

∂g
< 0.

∂σ
> 0.


∂g
2Þ
< 0.
8
47
g*
g*
No Institution−
^p = 0.86 1 ^p = 0.71 1
P1 P1
Note: σ = .9, p2 = .6 Note: σ = .9, p2 = .3

ized as following:

2.3.3 Discussion
48
strong type.
2008; Lucardi 2017; Smith 2005).
49


the f uture
to satisfy elites.
9
in my model.
10
50

take power.
2. A offers x1 2 ½0; 1.
µ µ
0 1
µ µ
pm − µ + g pm + g pm + µ + g
0 1
g ∈[0,1 − (pm + µ)]
11
51

for both periods.
draw.
5. A offers x2 2 ½0; 1.
The game ends.
2σ(1 − p1), 2σp1 (1 − x1) + σ(1 − p2), x1 + σp2

reject
reject
1 − (pm + µ) 1 pm + µ + g 1

x1 + x2
0 0 pm − µ + g 0
52
do so.



otherwise.
53

be short-lived.

pie?
x1 ≤ 1
2σp1 σp2 ≤ 1 (5)
1 p2
p1 ≤ þ ≡ p ^
2σ 2
54
2.5 Conclusion
^ also
increases:
^
∂p 1
¼ >0 (6)
∂p2 2
2σp1 ≤ x1 þ VE (7)


2σp1 ≤ 1 þ VE (9)
¼ σEVðp2 Þ (10)
¼ σðp2 þ gÞ
55
2σp1 1 ≤ VE
2σp1 1 ≤ σðp2 þ gÞ (11)
1
g ¼ 2p1 p2
σ
g ¼ 2p1 p2 1σ ¼ hðe gÞ > e
gÞ < ^
g ≤ 1 p2
1
2p1 p2 ≤ 1 p2 (12)
σ
1 1
p1 ≤ þ
2 2σ
increases:
∂g
¼2>0 (13)
∂p1

∂g
¼ 1 < 0 (14)
∂p2
56
2.5 Conclusion
∂g 1
¼ 2>0 (15)
∂σ σ
∂ð1 x2 Þ
¼ σ < 0 (16)
∂g

p1 > p


(18)
¼ 2σð1 p1 Þ


¼ ð1 x1 Þ þ VA (19)
¼ 0 þ VA
VA ¼ 1 EVðx2 Þ (20)

57
EVðx2 Þ ¼ 2σp1 1 (22)
VA ¼ 1 EVðx2 Þ
(23)
¼ 2 2σp1

setting g ¼ 0:
2 2σp1 ≥ 2σð1 p1 Þ (24)
1≥σ

^.
^ then A can always

2σð1 p1 Þ < 2 2σp1 (25)
σ<1
58
2.5 Conclusion

g ¼ g .
fight.
σðp2 þ g Þ
EVðx2 Þ ¼
1
¼ σp2 þ σ 2p1 p2 (26)
σ
¼ 2σp1 1


2σð1 p1 Þ < 2 2σp1 (27)
σ<1

59



(28)
2σp1 ≤ x1 þ VE

þ p2m ≡ p
60
2.5 Conclusion

satisfied:

(30)
2σp1 ≤ 1 þ VE

1
g ¼ 2p1 pm (31)
σ
p2 , pm þ μ.
1
2p1 pm ≤ 1 ðpm þ μÞ (32)
σ
1 1 μ
p1 ≤ þ ≡p
2 2σ 2
in period 1.
61


g 2 ½0; 1 ðpm þ μÞ.

62
2.5 Conclusion


(33)
2σp1 ≤ x1 þ VE

63
2


take such measures?
34

rule.
35


36

37

1
2
38

commitment power.
39
40

2.2.1 Model Set Up

3
41
(Geddes 2009).
of the pie remains.
4
42

choice.
2. A offers x1 2 ½0; 1.
for both periods.
offers x2 2 ½0; 1.
The game ends.
43
2σ(1 − p1), 2σp1 (1 − x1) + σ(1 − p2 − g),x1 + σ(p2 + g)
reject
reject
1 − p2 1 1

x1 + x2
0 0 0

elite.

(1)
2σp1 ≤ x1 þ VE

x1 ¼ maxf0; 2σp1 σp2 g (2)
44
þ p22 ≡ p
∂p ^
Formally, ∂p2
> 0.
5
numbers.
45


(3)
2σp1 ≤ x1 þ VE

1
g ¼ 2p1 p2 (4)
σ
to set g ¼ g .7


∂g
> 0.
6
gÞ < ^
7
46

∂g
< 0.

∂σ
> 0.


∂g
2Þ
< 0.
8
47
g*
g*
No Institution−
^p = 0.86 1 ^p = 0.71 1
P1 P1
Note: σ = .9, p2 = .6 Note: σ = .9, p2 = .3

ized as following:

2.3.3 Discussion
48
strong type.
2008; Lucardi 2017; Smith 2005).
49


the f uture
to satisfy elites.
9
in my model.
10
50

take power.
2. A offers x1 2 ½0; 1.
µ µ
0 1
µ µ
pm − µ + g pm + g pm + µ + g
0 1
g ∈[0,1 − (pm + µ)]
11
51

for both periods.
draw.
5. A offers x2 2 ½0; 1.
The game ends.
2σ(1 − p1), 2σp1 (1 − x1) + σ(1 − p2), x1 + σp2

reject
reject
1 − (pm + µ) 1 pm + µ + g 1

x1 + x2
0 0 pm − µ + g 0
52
do so.



otherwise.
53

be short-lived.

pie?
x1 ≤ 1
2σp1 σp2 ≤ 1 (5)
1 p2
p1 ≤ þ ≡ p ^
2σ 2
54
2.5 Conclusion
^ also
increases:
^
∂p 1
¼ >0 (6)
∂p2 2
2σp1 ≤ x1 þ VE (7)


2σp1 ≤ 1 þ VE (9)
¼ σEVðp2 Þ (10)
¼ σðp2 þ gÞ
55
2σp1 1 ≤ VE
2σp1 1 ≤ σðp2 þ gÞ (11)
1
g ¼ 2p1 p2
σ
g ¼ 2p1 p2 1σ ¼ hðe gÞ > e
gÞ < ^
g ≤ 1 p2
1
2p1 p2 ≤ 1 p2 (12)
σ
1 1
p1 ≤ þ
2 2σ
increases:
∂g
¼2>0 (13)
∂p1

∂g
¼ 1 < 0 (14)
∂p2
56
2.5 Conclusion
∂g 1
¼ 2>0 (15)
∂σ σ
∂ð1 x2 Þ
¼ σ < 0 (16)
∂g

p1 > p


(18)
¼ 2σð1 p1 Þ


¼ ð1 x1 Þ þ VA (19)
¼ 0 þ VA
VA ¼ 1 EVðx2 Þ (20)

57
EVðx2 Þ ¼ 2σp1 1 (22)
VA ¼ 1 EVðx2 Þ
(23)
¼ 2 2σp1

setting g ¼ 0:
2 2σp1 ≥ 2σð1 p1 Þ (24)
1≥σ

^.
^ then A can always

2σð1 p1 Þ < 2 2σp1 (25)
σ<1
58
2.5 Conclusion

g ¼ g .
fight.
σðp2 þ g Þ
EVðx2 Þ ¼
1
¼ σp2 þ σ 2p1 p2 (26)
σ
¼ 2σp1 1


2σð1 p1 Þ < 2 2σp1 (27)
σ<1

59



(28)
2σp1 ≤ x1 þ VE

þ p2m ≡ p
60
2.5 Conclusion

satisfied:

(30)
2σp1 ≤ 1 þ VE

1
g ¼ 2p1 pm (31)
σ
p2 , pm þ μ.
1
2p1 pm ≤ 1 ðpm þ μÞ (32)
σ
1 1 μ
p1 ≤ þ ≡p
2 2σ 2
in period 1.
61


g 2 ½0; 1 ðpm þ μÞ.

62
2.5 Conclusion


(33)
2σp1 ≤ x1 þ VE

63
3
Two Illustrative Cases
How do institutions constrain leaders and under what conditions do

autocrats establish institutionalized regimes? The model in Chapter 2
produced three important mechanisms and findings that address these
questions. This chapter will present illustrative case studies of
Cameroon (a highly institutionalized regime) and Côte d’Ivoire (a weakly
institutionalized regime), two countries in sub-Saharan Africa that have
experienced decades of authoritarian rule. The goal of this chapter is to
use these case studies to highlight key insights from the model.
I choose to focus on Cameroon and Côte d’Ivoire due to the compar-
ability of these countries. They share similar histories, economies, geogra-
phy, and populations. Both cases are former French colonies1 located in
West Africa, and they were granted independence in the same year. Both
countries have similar population sizes and degrees of ethnic heterogeneity
within the population. They both have large port cities along the Atlantic
coast and generally stable economies. In fact, as Figure 3.1 illustrates, the
gross domestic product (GDP) of both Cameroon and Côte d’Ivoire has
grown along very similar trajectories from 1960 to 2010. These similarities
allow me to hold background conditions constant while exploring impor-
tant differences in the independent and dependent variables.
I first present the case of Cameroon, which is an example of a regime
with high levels of institutionalization under the founding president,
Ahmadou Ahidjo. I show how Ahidjo, who was an initially weak leader,
used institutional bargains in order to maintain support from other elites –
therefore establishing a rule-based system. Unlike other founding leaders,
1
As will be discussed later in this chapter, a small sliver of Cameroon was controlled
by the British, but the country was mostly under French colonial rule. British rule in
Cameroon was very limited, as the French took the vast majority of the land (four-
fifths of the total area of Cameroon) – including the capital city of Yaounde.
Francophile elites and politicians have always dominated the executive in Cameroon.
64
3 Two Illustrative Cases
2000
1500
GDP per capita
1000
500
0
1960 1970 1980 1990 2000 2010 2020
Year
Côte d’Ivoire Cameroon
Figure 3.1 Yearly gross domestic product (GDP) of Cameroon and Côte d’Ivoire
such as Léopold Sédar Senghor of Senegal or Félix Houphouët-Boigny of

Côte d’Ivoire, Ahidjo was not a renowned, charismatic, popular indepen-
dence leader. He was initially encouraged to run for office by the French
colonial authorities and was perceived to be a part of the colonial
machine. When independence was granted, Ahidjo ascended to the pre-
sidency as a highly unpopular leader. To compensate for this initial lack of
support, Ahidjo distributed important cabinet positions to other elites,
including appointing Paul Biya to the position of prime minister – the
designated constitutional successor to the president. In 1982, power was
transferred peacefully from Ahidjo to Biya, and the same regime remains
in power today.
I juxtapose the case of Cameroon against Côte d’Ivoire under the rule
of the founding president, Félix Houphouët-Boigny, which is an example
of a regime with low levels of institutionalization. Houphouët was
a renowned independence leader, who lobbied for the right to self-
governance throughout French West Africa. Upon taking power,
Houphouët was extremely influential and faced very few credible chal-
lenges to his authority. Throughout his tenure, Houphouët centralized
power within his cabinet, leaving key ministerial positions, such as the
vice presidency, vacant. Houphouët died while in office in 1993, and
power was transferred to Henri Konan Bédié. However, Bédié’s claim to
the presidency was tenuous, as Houphouët had intentionally left
65
succession plans vague while he was in power. The military launched

a coup d’état in 1999 and removed Bédié from office.
Before I present the cases of Cameroon and Côte d’Ivoire, I will first
summarize the main claims developed in Chapter 2. This summary will
highlight the connective tissues that link the analysis from the formal
model with substantive details of the cases.
3.1 t he formal model summarized

The model presented in Chapter 2 produced a number of insights and
findings. Here, I summarize three key aspects of the model that will be
highlighted in the case studies. First, I argued that institutions constrain
leaders only when they empower specific elites. The model formalized this
mechanism as a shift in the underlying distribution of power away from
the leader. Executive constraints have the power to bind the leader’s hands
when they provide elites with direct access to the state. When an elite is
given an important cabinet position, such as the vice presidency or min-
ister of defense, they gain control of the entire ministry and its resources.
Within the context of sub-Saharan Africa, this is an especially valuable
source of power because the government was and continues to be a main
source of funding for development projects, employment, and the distri-
bution of resources to localities. Having authority over a piece of the
central government provides elites with a great deal of power.
The case of Cameroon will demonstrate how elites who were given
important cabinet positions used this access to power to build up their
own coalitions. Cabinet ministers frequently fill their ministries with
coethnics and bureaucrats from their home region. They have discretion
over funding and the distribution of resources to localities. State coffers
provide elites with a rich access to sources of patronage in order to reward
their supporters.
Moreover, when an elite is appointed to an important ministerial
position, they become a focal point for other elites. This is especially
true if the constitution specifies a succession order. If, for example, the
vice president is listed as the designated successor in the constitution, then
the elite in the position of vice president is formally and publicly declared
as the second highest ranked member in the government. Executive con-
straints that empower particular ministers help solve coordination pro-
blems amongst elites by providing a clear signal of the most influential
elites in the regime. This eliminates conflict within elites about who would
be the most likely challenger to the leader. Similar to the “crown prince
problem” famously identified by Herz (1957), when the leader identifies
66
3.1 The Formal Model Summarized
a successor, that person automatically becomes a potential challenger

simply by the designation of an alternative center of power.
This mechanism is well illustrated by a comparison of the first leader-
ship transitions in Cameroon and Côte d’Ivoire. In Cameroon, Ahidjo
appointed Biya as his prime minister in 1975 – the constitutionally
designated position to succeed the president. Seven years later, Ahidjo
voluntarily retired and passed power to Biya, citing health reasons. In
part, due to the legitimacy of the constitutional succession procedure, Biya
maintained control of the regime – even in light of an attempted coup
a year after his ascension to the presidency. By contrast, in Côte d’Ivoire
Houphouët largely avoided designating a successor, keeping the position
of vice president vacant throughout most of his rule. Toward the end of his
rule, he reluctantly designated Bédié as his successor, but factional battles
over succession persisted, and Bédié was not able to consolidate power
upon Houphouët’s death. Six years later, Bédié was deposed in the coun-
try’s first coup d’état since independence.
Second, the model showed that leaders who initially enter power weak
must make the counterintuitive decision to give power away to elites in
order to maintain support of the regime. Leaders who enter power already
strong face weak elites who do not pose a credible threat of launching
a successful rebellion against the regime. Strong leaders can therefore
always pay off weak elites without making long-term institutional com-
mitments. Leaders who enter power weak, on the other hand, face
a commitment problem in the initial period. In this case, if elites were to
try to depose the leader, they would likely succeed. In order to survive in
power past this initial period, weak leaders must make institutional guar-
antees of rent distribution to elites. This is achieved by creating executive
constraints that empower elites, solidifying their future ability to hold the
leader accountable by transferring state resources and influence to them.
The comparison of Cameroon and Côte d’Ivoire highlights this finding.
Ahidjo entered power in Cameroon extremely weak due to the perception
that he was an ally of the French colonial administration and therefore
had simply inherited his position of power from the outgoing authorities.
As a result, Ahidjo could not rule alone; he used important cabinet posi-
tions to make institutional commitments to elites and solidified limitations
on his personal executive authority through constitutional rules. On the
other hand, Houphouët entered office after independence as the single
most powerful politician in Côte d’Ivoire. He had long built a reputation
as a freedom fighter not only within his own country but also for all
Africans in the region. Upon ascending to the presidency, he systemati-
cally shut out elites from his own party and coalition from positions of
power, preferring to keep French technocrats in the bureaucracy. This
67
strategy of personal rule allowed Houphouët to remain the single most

important figure in Ivorian politics throughout his tenure.
Initial weakness has long-term consequences. Leaders who enter power
especially weak must institutionalize the regime in order to avoid being
deposed in the first period. This is true regardless of whether the leader
would have been stronger in future periods; without institutionalizing
after coming into power, the leader would not have made it to a future
period! The period right after a leader comes into power is therefore
a critical juncture: the initial distribution of power between the leader
and elites determines the institutional makeup of the regime in the long
run. Since institutionalization shifts power in favor of elites, it is self-
reinforcing. The leader’s initial decisions to create executive constraints
shape the long-term trajectory of the regime.
The institutional arrangements established under Ahidjo largely
remain in Cameroon today. Biya has been in power since 1982, and
there has never been a coup d’état or civil war in Cameroon.2 The regime
in Côte d’Ivoire fell six years after Houphouët’s death, as his successor
was deposed in a coup in 1999. There have also been two civil wars – the
first from 2002 to 2007 and the second from 2010 to 2011 – since the first
presidential transition.
Third, the model illustrated that initially weak leaders who insti-
tutionalize the regime and strong leaders who do not institutionalize
can both stay in power for long periods of time. Indeed, in the model,
conflict never occurs because initially weak leaders prefer to institu-
tionalize in order to maintain support of elites. Likewise, conflict also
never occurs in regimes with strong leaders because elites do not pose
a credible threat. The empirical implication of this finding is that we
generally should not observe a systematic relationship between leader
longevity and executive constraints because institutionalization is an
endogenous process.
Ahidjo was the president of Cameroon for twenty-two years, from
1960 until his voluntary retirement in 1982. Houphouët was the president
of Côte d’Ivoire for thirty-three years, from 1960 until his death in 1993.
It is also difficult to know how long Ahidjo would have remained in the
presidency if he had not voluntarily retired due to ill health. Regardless,
both leaders remained in power for decades despite two very different
models of rule.
2
Though in recent years an anglophone separatist movement has emerged in
Cameroon. The country has also experienced insurgent violence from Boko
Haram. Cameroon experienced an unsuccessful coup attempt in 1984, as this chap-
ter will later discuss.
68
3.2 Cameroon Under Ahidjo
3.2 cameroon under ahidjo: high levels o f

regime institutionalization
The case of Cameroon under the presidency of Amadou Ahidjo is an
example of a regime with high levels of institutionalization. Cameroon
has consistently had formal succession policies written into the constitu-
tion since the country gained independence. Article 9 of the constitution
states: “In the event of the Presidency falling vacant for any reason
whatsoever, the powers of the President of the Federal Republic shall
automatically devolve upon the Vice-President until such time as a new
President is elected.” In 1975, the title for the office of vice president was
changed to prime minister, and the succession clause in the constitution
was amended to reflect the change in title for this office. This stands in
contrast with the case of Côte d’Ivoire, in which the constitution under
Houphouët lacked detailed succession rules almost the entire time he was
in power.
Key cabinet ministerial positions were consistently filled and remained
quite stable under Ahidjo. Importantly, the office of vice president/prime
minister was almost always filled.3 Not only is this position the “second in
command” within the cabinet, the person in this office is the designated
constitutional successor. In addition, cabinet rotation rates under Ahidjo
were relatively low. Throughout his tenure, Ahidjo only had three different
vice presidents. Ahidjo also appointed a minister of defense the entire time
he was in power. This pattern of cabinet stability contrasts starkly with the
Houphouët government, which did not have a vice president or
prime minister over 90 percent of the time Houphouët was in power.
Houphouët also kept the minister of defense position vacant a third of the
time he was in power.
The fact that Cameroon, especially in the first decades after indepen-
dence was granted, had high levels of regime institutionalization may
come as a surprise. The existing scholarship on colonial legacies generally
claims that former British colonies had stronger institutions that kept
rulers in check as well as more robust legal traditions (Hayek 1960; La
Porta et al. 1998; Landes 1998). Cameroon was largely a French colony
prior to independence, yet institutional checks on executive power were
established, and most of these constraints were created early on during
Ahidjo’s rule.
3
For a short period during Ahidjo’s rule, from 1973 to 1974, the office of the vice
president was eliminated from the cabinet. In 1975 the title of the office of vice
president was changed to prime minister and was added back to the cabinet. From
1975 through until the end of Ahidjo’s rule, Paul Biya was appointed as prime
minister.
69
For the past three years, Cameroon’s English speaking regions (the
areas formerly ruled by the British) have started to engage in separatist
conflict against the French-dominated Cameroon government. However,
for most of Cameroon’s history – and especially during the time period
that is highlighted in the case study for this book (1960–2010) – this
linguistic cleavage was not particularly salient. Moreover, I argue that
Cameroon is an example of a regime with high levels of institutionaliza-
tion, making it durable against threats. The fact that the regime in
Cameroon has persisted, despite these separatist conflicts, provides addi-
tional support for the argument that the regime is highly institutionalized.
3.2.1 Background and History

Cameroon is a country located in Central Africa though it borders the
Atlantic Ocean on its western side. It currently has a population of
23.44 million and about 55 percent of the population lives in urban
areas (World Bank 2016). Cameroon shares a border with Nigeria,
Chad, the Central African Republic, Equatorial Guinea, Gabon, and the
Republic of the Congo. The capital and second largest city of Cameroon is
Yaoundé, located in the central region of the country. Douala, located on
the Atlantic coastline, is the largest city in Cameroon, and is the economic
center of Cameroon and largest seaport in Central Africa.
Cameroon has an extremely heterogeneous population, with over 250
different ethnic groups. It is considered one the most linguistically diverse
countries in the world. Though groups such as the Bantoid, Bamileke,
Bantu, Kirdi, and Fulani comprise of large numbers, no particular ethnic
group dominates the population of Cameroon. Roughly two-thirds of the
population is Christian, and less than a quarter of the population, notably
the Fulani tribe (of which Ahidjo was a member), is Muslim.
The economy of Cameroon is roughly one of the ten largest economies
within sub-Saharan Africa. For the first two decades after independence,
agriculture made up the bulk of Cameroon’s economy, and important
cash crops included cocoa, coffee, cotton, and bananas. In 1977,
Cameroon began producing and distributing oil. It is now the sixth largest
oil producer in sub-Saharan Africa, though agriculture remains a large
part of the Cameroonian economy. Its current GDP is US$34.8 billion
(World Bank 2016).
Cameroon gained independence in 1960 after decades of colonial rule
by Germany, England, and France. Prior to colonization, the territory was
organized into chiefdoms and kingdoms, including the Bornu Empire.
Colonial rule in Cameroon began when Germany established the territory
as a German protectorate in 1884. Germany’s loss in the First World
70
War resulted in a partition of the territory between the French and the
British in 1916. This agreement divided the country into two territories –
the French took the vast majority of the land in the country (four-fifths of
the total area of Cameroon), and the British took the remaining land.
Under this agreement, the two Cameroons became League of Nations
mandated colonies under France and the United Kingdom (Le Vine
1964, 4–8).
3.2.2 Ahidjo’s Rise to Power: Initial Leader Weakness

Amadou Ahidjo, the first president of Cameroon, entered office weak and
vulnerable to being deposed. Unlike other founding presidents in newly
independent African countries, Ahidjo was not a national independence
hero. On the contrary, he largely inherited his position of power from the
colonial government and remained a close ally of France throughout his
tenure. As a result, Ahidjo was largely perceived to be a collaborator of the
French colonial authorities and was initially very unpopular. As Joseph
(1978) notes, “In spite of the efforts of the propagandists of the regime,
President Ahidjo cannot be considered a heroic figure from the nationalist
past . . . it is a matter of historical fact he was initially “fabricated” and
brought to power by the colonial regime” (76).4
Ahidjo was born in northern Cameroon to a Muslim Fulani family in
1924. His father was the local Fulani village chief. In 1939, Ahidjo began
attending Ecole Primaire Supérieur, a secondary school in Yaoundé, with
the goal of beginning a career in civil service. In 1942, Ahidjo joined the
state bureaucracy by becoming a post-office radio operator. Over the next
few years, his appointment in the civil service would take him to several
posts across Cameroon (Glickman 1992; Ngoh 1987).
Meanwhile, in Cameroon, emerging parties and political elites began to
lobby for independence. One of the earliest major parties to call for
independence and the reunification of British and French Cameroon was
the Union des Populations du Cameroun (UPC). The UPC, which was
formed in 1948, grew out of the French communist-oriented trade unions
that had formed near the end of the Second World War. The UPC was
strongly in favor of self-government, and two of its major platforms
involved independence from France and the reunification of British and
French Cameroon (Ngoh 1987).
4
Joseph also notes that “in view of his uncertain and controversial origins, [Ahidjo] is
not strongly identified with any particular ethnic group,” making it even more
difficult for him to have a mass popular following (77).
71
In 1951, Ahidjo joined the Bloc Democratique Camerouanis (BDC) as

one of the founding members of the party. The BDC was much less radical
than the UPC, and it was initially opposed to lobbying for immediate
independence from France. In 1958, Ahidjo broke from the BDC and
formed his own party, the Union Camerounaise (UC).
It is important to stress that a cohesive and unified independence move-
ment never developed in Cameroon. This was partly because Cameroon
was divided into British and French territories, with each side having
regional factions with difference goals and approaches toward indepen-
dence (DeLancey 1989; Rubin 1971). More importantly, the colonial
administration actively stymied the independence movement by encoura-
ging the development of more moderate political parties. Since these other
political parties, such as the BDC, tended to favor the status quo rather
than call for immediate independence, the French supported the creation
of these parties to serve as a counterweight against the popular UPC. In
1955, the colonial administration banned the UPC, driving the party
underground (Ngoh 1987).
In addition to encouraging the development of politically moderate par-
ties that did not call for immediate independence, the French also sought out
African elites who would be reliable allies for the colonial administration.
They were eager to find young Cameroonian civil servants who were edu-
cated and trained by the French, and could act as a counterbalance against
the nascent independence movement inspired by the UPC (Joseph 1978).
In 1956, in response to growing agitations for independence from
French West African colonies, the French National Assembly passed the
“Loi Cadre,” which gave overseas territories increased self-governance
including the ability to create national assemblies and internal govern-
ments. Elections were held in 1956, and Ahidjo’s party, the UC, won
a large share of seats in the territorial assembly, since the more radical
(and popular) UPC party remained banned from politics.
Though Cameroon had not yet gained independence, the first
Cameroon government was formed out of these 1956 elections. The first
prime minister, Andre-Marie Mbida resigned in response to a ministerial
crisis and unpopular policies less than a year into the position (Le Vine
1964). This provided Ahidjo, who had been Mbida’s vice premier and
minister of the interior, with the opportunity to rise to the post of prime
minister in 1958 (Ngoh 1987, 150). As prime minister of the French
Cameroon territory, Ahidjo remained a close ally of the French authorities
and even conducted military attacks against the underground UPC inde-
pendence movement (DeLancey 1989, 40).
Cameroon achieved full independence from France on January 1, 1960,
and Ahidjo was elected as the first president of the newly independent
72
country three months later. A year later, British Cameroon voted to

reunify with the newly independent territory, and on October 1, 1961,
Cameroon became a federal republic. However, Ahidjo began his tenure
in a position of weakness, relative to other political and economic elites in
the country. There were three main reasons for his initial weakness.
First, Ahidjo was never perceived as a nationalist fighter, and in fact, was
accurately known to be a close collaborator of the French, allowing him to
inherit his position of power. The French authorities actually encouraged
Ahidjo to run for office in the first place because they viewed him as
a politician who would act in the interest of the colonial state. The colonial
administration even referred to the “Ahidjo option”: given that indepen-
dence seemed to be increasingly inevitable, the French authorities preferred
to have Ahidjo as head of state (Joseph 1978, 46). As DeLancey (1989)
notes, “Ahidjo owed his rise to the presidency of Cameroon entirely to
French manipulations” (281). Ahidjo was perceived largely as a “French
tool” and this image made him deeply unpopular (Le Vine 1964, 173). In
response, France made a number of attempts to convey the image that
Cameroon was a self-run government. In 1959, France initiated the Statue
of the Cameroun, which gave Cameroon full internal autonomy.
Yet the image of Ahidjo as a “French tool” was not a false perception.
In fact, Ahidjo maintained close relations with France throughout his
tenure in order to encourage continued French investment in the
Cameroon economy. Even after the Statue of the Cameroun was passed,
the statute still preserved French jurisdiction over money and exchange
rates, foreign policy, frontier security, and the right to intervene in the
event of armed insurrection or war. In fact, DeLancey (1989) argues that
Ahidjo’s ability to maintain support from the French by protecting French
investments was one of the key aspects of his strategy for rule (41, 58).
French business, commercial, and industrial investors dominated the
nonagricultural sectors of the economy in Cameroon, and major devel-
opment projects were often funded by private French investment. This
stands in stark contrast to the approach of many other African indepen-
dence leaders, such as Sekou Toure of Guinea, who voted for immediate
and complete severance of ties (including trade) with France in 1958.
A second major reason why Ahidjo entered power weak relative to
other elites is because his ethnic group, the Fulani, did not produce any of
the major exports. Though the Fulani dominated cattle herding and trade
in the northern region, they did not produce any of the major cash crops
and were not the leading economic elites in Cameroon. By contrast, ethnic
groups in the south, such as the Bamileke and Bassa, were the leading
producers of cocoa and coffee – the major cash crops in Cameroon. In the
years leading up to independence, one southern group in particular, the
73
Bamileke, became some of the richest farmers in Cameroon, and began to

expand into other areas of commerce. By the end of the Second World
War, the Bamileke began to compete for land and labor with French
plantation owners (Arriola 2013).
In the 1950s, the Bamileke became one of the colonial administra-
tion’s (and by proxy, Ahidjo’s) primary opposition forces. As they
became more economically powerful, the Bamileke began to seek poli-
tical influence. They threw their support behind the UPC – the radical
independence party that was banned by the French in 1955. By the time
Ahidjo became president in 1960, the Bamileke were a formidable
opposition – not only due to their economic strength but also their ability
to mobilize southern interests (Konings 2007; van de Walle 1993).
A third source of Ahidjo’s initial weakness was that Cameroon, having
been split into two colonial territories (British Cameroon and French
Cameroon), was not a unified state when Ahidjo first came to power.
Recall that French Cameroon gained independence in 1960 and British
Cameroon voted to reunify with the independent territory in 1961. Yet
from 1960 through 1963, Eastern Cameroon (which was formerly British
Cameroon) maintained its own separate cabinet in the federal govern-
ment. Even though Ahidjo was the president of the Cameroon, the regio-
nal government of East Cameroon was largely out of his political control.
It wasn’t until 1972 that Ahidjo was successful in abolishing the federal
system of government and centralized power in Yaoundé (DeLancey
1989, 54).
3.2.3 Ahidjo in Power: Creating Institutional Constraints

Upon taking power, Ahidjo needed to create official structures that would
allow him to buy the support of key elites. He did this by creating
constitutional limits on his own authority that empowered other elites
(in particular, by naming a prime minister who would serve as his succes-
sor) as well as the distribution of cabinet positions in exchange for support
from elites. As Takougang (2004) argues, “with the tremendous resources
of the state at his disposal, [Ahidjo] could use them to maintain the loyalty
of various groups and regions to himself” (75).
Ahidjo systematically used cabinet positions in order to secure support
from other elites. His ruling strategy is perhaps best summarized by
DeLancey (1989): “The essence of [Ahidjo’s] party was the cohesion of
a few important people, each of whom brought in his/her loyalists to the
party” (52). Importantly, the “important people” who were brought into
the government comprised of would-be challengers to Ahidjo.
74
[Ahidjo] treated his opponents firmly, sometimes harshly, but made sure that even
his bitterest enemies had both the chance of joining his side and of actively sharing
in the perquisites of rule. That he was never vindictive is to his credit: Mbida was
repeatedly offered various portfolios, Okala came out of prison to become an
ambassador, and several former UPC leaders have taken high and well-paying jobs
in government. (Le Vine 1964, 181)
Charles Assale, for example, the leader of the Mouvement d’Action

Nationale, a regional party in Eastern Cameroon, was offered the position
of prime minister of East Cameroon. “Such positions not only honored
the individuals concerned but also gave them access to sources of patron-
age with which to reward their supporters” (DeLancey 1989, 52). Elites
who opposed Ahidjo’s policies or refused to support his regime were shut
out of the cabinet and state bureaucracy.
Ministerial appointments provided “a majority opportunity for Ahidjo
to reward influential people in society – or even to build influence for
individuals – and to tie them to him” (DeLancey 1989, 59). Once
appointed, why did elites not defect? The positions in the presidential
cabinet gave these elites the ability to provide his followers with jobs and
resources he controlled through the ministry. This chain dependency is
illustrated by the following example: From 1970 to 1981 the Minister of
Posts and Telecommunications was Emmanuel Egbe Tabi. Tabi was from
the Anglophone region of Mamfe. During that time, Tabi filled his min-
istry with other Anglophones, who were otherwise underrepresented in
the predominately Francophone bureaucracy. Anglophones were even
appointed to post-office positions in French speaking parts of the country.
Moreover, many of the higher positions within the ministry went to
bureaucrats who were from Tabi’s home region – the Mamfe area
(DeLancey 1989, 59).
To sum up, on one hand, ministerial appointments empowered elites by
providing them the influence and resources to reward their supporters and
consolidate their authority. Yet on the other hand, these appointments
tied these ministers to the Ahidjo regime and provided them an incentive
to continue supporting the incumbent in order to maintain their own
positions within the cabinet. As Takougang (2004) argues,
through this system of appointments to cabinet and other important government
and party offices that was based on regional and/or ethnic representation, Ahidjo
was able to maintain the loyalty of most groups in the country. That is because
these high-ranking appointees in turn were able to tie their respective regions and
groups to the president by building their own network of support through the
provision of administrative and party jobs. (76)
Ahidjo also took great pains to make sure that elites from various
regions, ethnicities, and religions were represented in his cabinet. “The
75
cabinet as a group was an arena for Ahidjo to play ‘ethnic arithmetic’ or

‘ethnic balancing’” (DeLancey 1989, 60). He also ensured regional repre-
sentation: East (French) Cameroon and West (British) Cameroon were
both represented in the cabinet. Ahidjo’s presidential cabinet in 1975, for
example, included six ministers from the north, eight from the central-
south, seven from the west, and two from the east (Kofele-Kale 1986, 73).
Power-sharing, however, came at a cost. As this chapter will later show,
Ahidjo’s own ethnic and regional elites did not control the majority of the
cabinet seats. Power-sharing under the Ahidjo regime required that the
leader relinquish (some) control over the state.
Finally, Ahidjo also struck bargains with traditional chiefs. At inde-
pendence the lamidos (Fulani chiefs) still dominated the northern region
of the country. Even though Ahidjo was a Fulani from the north, he was
unpopular with the lamidos, who believed that Ahidjo wanted to mod-
ernize society, therefore weakening their authority (Ngoh 1987, 152).
To gain their support, Ahidjo created a constitutional stipulation that
stated that the status of traditional chiefs would be protected. He also
filled key cabinet appointments, such as the minister of the interior, with
elites that the lamidos favored. However, Ahidjo made it clear that in
exchange for these benefits, he expected the lamidos to support his rule
(Joseph 1978, 53).
It is important to acknowledge that, in addition to institutional strate-
gies of rule, Ahidjo also relied on repression and force to maintain his grip
on power on the wider population. Where certain parties or elites refused
to be coopted, Ahidjo used police and military repression. Though
opposition parties were technically legal, individuals who attempted to
establish other political parties were arrested or tortured (DeLancey 1989,
63). However, these repressive tactics were generally reserved for the
broader citizenry and potential opposition. For his most inner circle of
elites, Ahidjo relied on institutional forms of cooptation.
3.2.4 Ahidjo Constrained, Biya Empowered: The Presidential

Transition
Our discussion of Cameroon concludes with an example of how institu-
tions constrained the leader. In 1982, Ahidjo voluntarily retired from the
presidency due to health reasons, transferring power peacefully to his
prime minister and constitutionally designated successor, Paul Biya.
Though he stepped down, Ahidjo had hoped to retain informal control
of the presidency. These attempts culminated in a failed coup attempt,
which was successfully thwarted by Biya. Ahidjo had empowered Biya by
giving him access to the state as well as constitutional legitimacy as his
76
successor, therefore Ahidjo could not simply control Biya as a puppet

president. The first presidential transition in Cameroon provides a useful
test to examine whether institutionalization truly does shift power from
leaders to elites in a meaningful way. It also demonstrates how Ahidjo
became constrained by the very institutions he created while he was
in power.
Since independence, the Cameroon constitution has always included
stipulations regarding the process of leadership succession. Article 9 of the
constitution states: “In the event of vacancy of the Presidency for
whatever cause the powers of the President of the Federal Republic shall
without more devolve upon the Vice-President until election of a new
President.” In 1977, the title for the office of vice president was changed to
prime minister, and the succession clause in the constitution was amended
to reflect the change in title for this office (Article 7).5
When Ahidjo appointed Paul Biya as his prime minister in 1975, Biya
became the designated successor to the presidency. Biya served as Ahidjo’s
prime minister for seven years, leading up to the presidential transition.
Notably, Biya was a Christian from Southern Cameroon, in contrast with
Ahidjo, who was an ethnic Fulani Muslim from the north. Biya began his
political career in Ahidjo’s government soon after independence was
granted. He served in various ministerial positions, such as the director
of the Ministry of National Education and minister of state, and worked
his way up the cabinet (Glickman 1992).
On November 4, 1982, after being in power for twenty-two years,
Ahidjo – citing health concerns – voluntarily retired from the presidency
and passed power down to his appointed successor, Biya. In announcing
his retirement, Ahidjo expressed his support for Biya, stating in a radio
broadcast:
Fellow Cameroonians, my dear countrymen, I have decided to resign my duties as
Head of State of the United Republic of Cameroon . . . I request all of you
Cameroonians to give your full confidence and whole-hearted support to my
constitutional heir, Mr. Paul Biya. He merits the confidence of all, both at home
and abroad. (Ngoh 1987, 300)
Although Ahidjo stepped down from the presidency, he clearly expected

to retain influence and control over the Biya presidency. Notably, after
stepping down as president, Ahidjo continued to serve as the chairman of
the ruling Cameroonian National Union (CNU) party. In fact, in an inter-
view two months after the leadership transition, Ahidjo, whose only
5
From 1977 to 1979, the succession order was changed to the president of the
National Assembly and then prime minister. In 1979, the constitution was amended
so that the prime minister was the direct successor.
77
remaining official title was party chairman of the CNU, asserted that the
party takes precedence over the government (Konde 2012).
As Ahidjo’s prime minister Biya had maintained an image of
a technocrat and made it clear that he had no ambitions to compete
with Ahidjo (Mbaku and Takougang 2004). After acceding to the pre-
sidency, however, Biya made a number of administrative and policy
changes that diminished the influence of Ahidjo’s allies in favor of his
own power base. This included multiple cabinet reshuffles in which most
of Ahidjo’s appointees were replaced. Within a year Biya replaced over
half the cabinet he had inherited from Ahidjo. Biya also undertook admin-
istrative reforms, most notably dividing the northern region – Ahidjo’s
home district – into three provinces (CIA 1983; 1984).
In response to Biya’s consolidation of power, Ahidjo tried to reclaim his
influence. He reportedly attempted to introduce a constitutional amend-
ment that would increase his own authority as chairman of the party while
reducing the powers of the president. He asked his remaining ministers
who had not been replaced in the cabinet reshuffle to resign in protest –
none of whom did (Ngoh 1987, 309–310).
On April 6, 1984, members of the palace guard who were loyal to Ahidjo
attempted to launch a coup against Biya. It is believed, though not verified,
that Ahidjo himself help to orchestrate the coup. This coup failed, in part
because of tactical mistakes, and also because the coup plotters failed to
gain support of key members of the armed forces. Though the constitu-
tional succession provisions was clearly not the only reason Biya was able to
retain power in light of the failed coup attempt, it conferred legality and
legitimacy to his claim to the presidency (Hughes and May 1988).
Chapter 2 opened with the following observation by Montesquieu: “At the
birth of new polities, leaders mold institutions, whereas afterwards institu-
tions mold leaders” (qtd in Putnam 1993, 26). Much like Montesquieu’s
remark, after relinquishing power to his successor, Paul Biya, Ahidjo was shut
out of returning to power by the very institutions that he put in place himself.
Moreover, the constitutionally designated succession procedures ensured that
Ahidjo could not renege on promises to transfer power to Biya. As Ngoh
(1989) argues, Ahidjo “realized that after twenty-five years as the undisputed
strongman in Cameroon, he had to play second fiddle to Paul Biya – some-
thing he was not prepared to do” (319).
Paul Biya remains in power in Cameroon today – even surviving the
introduction of multiparty elections in 1992. How did he remain in power
despite demands for political liberalization in the 1990s? In large part,
Biya benefited from the fact that the transfer of power took place long
before pressures to introduce multiparty elections arose by the end of the
Cold War. By 1991, Biya had already been in power for almost a decade.
78
3.3 Côte D’Ivoire Under Houphouët
He therefore had already secured his tenure through a combination of

cooptation through institutionalized mechanisms of power sharing as
well as coercion by the time he was forced to introduce multiparty
competition.
By contrast, as the case of Côte d’Ivoire will demonstrate, Houphouët’s
successor, Bédié, came into office right after the introduction of multiparty
elections. In addition to his already weak claims of legitimacy due to the
absence of institutionalized succession mechanisms, he did not have the
benefit of time in the way that Biya had.
3 . 3 côte d’ivoire under houphouët : l o w l e ve l s

of regime institutionalization
Unlike Cameroon, Côte d’Ivoire is an example of a regime with low levels
of institutionalization. For most of Houphouët’s rule, specific constitu-
tional rules designating the successor order were not established. In 1975,
a constitutional succession order was created. Article 11 of the constitu-
tion stated: “In case of the death or incapacitation of the President his
functions are carried out by the President of the National Assembly.” Yet
Houphouët continuously amended this constitutional rule, creating
uncertainty about who the designated successor would be. In 1980, the
constitutional rule was changed so that the designated successor would be
the vice president. In practice, however, Houphouët kept the vice presi-
dent position vacant during this entire period, intentionally avoiding the
designation of a second in command. In 1987, the succession order was
changed to the president of the National Assembly and then the vice
president. In 1990, the constitution was amended once more, so that the
constitutional successor was the president of the National Assembly.
In addition, key cabinet ministerial positions were routinely left vacant
under the Houphouët administration. From 1960 to 1981, Houphouët’s
cabinet did not have a clear “second in command” position. During these
two decades, there was not a vice president or prime minister in the
presidential cabinet. In 1982, the position of vice president was created
but it was left vacant every year from 1982 to 1985. In 1986, the position of
vice president was eliminated from the cabinet, which once again lacked
a “second in command” position from 1986 to 1990. In short, for the vast
majority of Houphouët’s rule he never designated an informal successor,
nor did he appoint a prime minister or vice president. From 1960 to 1990,
this position was either eliminated from the cabinet or it sat empty. It wasn’t
until 1991 – two years before his death, that Houphouët created a prime
minister position. Moreover, the minister of defense position was also kept
vacant a third of the time Houphouët was in office.
79
This case study will demonstrate how Houphouët entered power in

a position of extreme strength – he had already completely consolidated
political authority within Côte d’Ivoire even before entering office. After
coming into power, Houphouët systematically shut out Ivorian elites
from positions of power, preferring to rely on French technocrats or by
appointing less influential members of other ethnic groups to cabinet
positions. In the language of the formal model from Chapter 2,
Houphouët entered office with an extremely favorable draw of the dis-
tribution of power – no other elite possessed the ability to depose him. As
a result, Houphouët did not need to institutionalize power-sharing
arrangements to remain in power. He remained in power for over three
decades as a personalist leader. The collapse of the regime after the death
of Houphouët and the two civil wars that followed in the 2000s is
perhaps surprising, given that Côte d’Ivoire was considered the economic
miracle of West Africa in the 1980s. Yet without institutionalized
mechanisms for power-sharing, personalist rule could not sustain the
regime in the long run.

Côte d’Ivoire, also known as the Ivory Coast, is a country located in West
Africa, bordering the Atlantic Ocean on its southern coast. It currently has
a population of 23.7 million and about 55 percent of the population lives in
urban areas (World Bank 2017). Côte d’Ivoire shares a border with Guinea,
Liberia, Burkina Faso, Mali, and Ghana. The capital of the country is
Yamoussoukro, though it is only the sixth largest city. The port city of
Abidjan, located on the Atlantic coastline, is the economic center of Côte
d’Ivoire. It is the largest city in Côte d’Ivoire, with a population size of
almost five million, and is the fourth largest city in Africa.
Though it is not as ethnically heterogeneous as Cameroon, Côte
d’Ivoire is home to at least sixty different ethnic groups and dialects.
The Akan is the single largest ethnic group in the country, comprising of
roughly 37 percent of the population. Within the Akan the Baoulé is the
largest subgroup, and this is the tribe that Houphouët belonged to. There
is not one dominant religion in Côte d’Ivoire, and Muslims and Christians
each make up about a third of the population.
Côte d’Ivoire has the second largest economy in West Africa and is
currently the fourth largest exporter of goods within sub-Saharan Africa.
The economy is heavily dependent on agricultural exports, and almost
70 percent of the population works in agriculture. Since independence, its
main cash crops have been coffee, cocoa beans, and palm oil. Côte d’Ivoire
experienced extraordinary growth and prosperity in the first few decades
80
of independence. Its GDP grew by as much as 360 percent in the 1970s

and it was considered the “economic miracle” of West Africa (Daddieh
2001). The current GDP in Côte d’Ivoire is US$40.39 billion, and it is the
world’s largest exporter of cocoa beans (World Bank 2017).
Côte d’Ivoire gained independence in 1960 after decades of French
colonial rule. Prior to colonization, the territory was organized into several
states and kingdoms. European contact was established in the late 1800s.
Côte d’Ivoire was designated a French protectorate in 1843, becoming
a French colony in 1893. In response to growing agitations for indepen-
dence, France began to grant increased measures of self-governance for its
colonies, including representation in the French Constituent Assembly, of
which Houphouët was a representative. He would serve as one of the most
influential African representatives for fourteen years leading up to indepen-
dence. Côte d’Ivoire achieved full independence on August 7, 1960, with
Houphouët as the country’s first president.
3.3.2 Houphouët’s Rise to Power: Initial Leader Strength

Political rule under Houphouët is perhaps best summarized by the open-
ing sentence of Aristide Zolberg’s (1969) definitive volume on politics in
Côte d’Ivoire: “Political life in the Ivory Coast has been centered in recent
years around the activities of one man: Félix Houphouët-Boigny . . . By
1957, three years before the country became independent, [he] had gained
a virtual monopoly over access to public office at all levels” (3). As
Jackson and Rosberg (1982, 145), observed, the government of Côte
d’Ivoire is “the government of virtually one man” and no other politicians
achieved similar political stature under Houphouët’s rule.
Houphouët was a major player in Ivorian politics from the very begin-
ning. He was born into a family of hereditary chiefs of the Baoulé group
and became chief of the Akoué tribe through hereditary succession at
a very young age. Beyond his family’s high status, Houphouët further
elevated his prominence upon his first marriage. His first wife was of royal
Agni lineage through her mother and was Senegalese through her father.
Through this marriage, Houphouët acquired kinship ties and alliances
with other ethnic groups in Côte d’Ivoire as well as Senegalese elites
(McGovern 2011, 151; Zolberg 1969, 73).
Houphouët attended the Ecole William Ponty, an elite college in
Senegal that was attended by many young Africans who would eventually
be important independence politicians in their countries (other students
included Modibo Keita of Mali, Hamani Diori of Niger, Mamadou Dia of
Senegal, and Maurice Yaméogo of Burkina Faso). He then attended the
Ecole de medicine de I’AOF (French West Africa School of Medicine) in
81
Senegal, becoming a medical aide by 1925 – one of the first Baoulé to

achieve this status. Houphouët served in colonial medical service in var-
ious parts of Côte d’Ivoire until 1940. During this time, upon seeing their
mistreatment by the colonial administration, he became interested in
helping to organize cocoa and coffee farmers (McGovern 2011).
When his uncle died in 1940, Houphouët inherited large amounts of
land in his village and was appointed the chief of his Akoué tribe. From
then on, Houphouët would go on to begin his political career as one of the
richest African farmers in the entire country. His economic status served
as an important symbol of a growing “Ivory Coast bourgeoisie” and it
also provided him with the ability to finance his political campaigns
(Zolberg 1969, 73–74).
In 1944, Houphouët along with other African elites, established the
Syndicat Agricole Africain (SAA), an African agricultural union within
Côte d’Ivoire. Importantly, the SAA was the first quasi-political party that
agitated for better conditions and some measures of self-governance for
Africans within Côte d’Ivoire. Its goals were to secure premiums for
African producers, to obtain a quota of imported cloth and agricultural
commodities, etc. The eight elites who founded the SAA, including
Houphouët, were all traditional leaders who were leading planters and
literate in French. By the end of 1944, membership to the SAA had grown
to 8,548 and nearly half of the members were Baoulé. Out of the founders,
Houphouët was the only Baoulé, and he was elected president of the
organization when the SAA held its first congress in Abidjan in
September 1944 (Zolberg 1969, 67).
There are two main reasons why Houphouët was able to consolidate
power early and begin his presidency in an extremely strong position. In the
years between 1944 and independence in 1960, two major developments
cemented Houphouët’s stature as a leading politician and national hero in
Côte d’Ivoire. The first was his election to the French Constituent Assembly
and subsequent creation of his party, the Parti Democratique de la Côte
d’Ivoire (PDCI), as well as a pan-African coalition that would become the
basis of the independence movement within French West Africa. Though in
reality, Houphouët maintained close ties with Paris even after
independence,6 his leadership in these independence organizations solidi-
fied his image as a freedom fighter and founding father to an independent
6
There is a famous story that on the eve of independence, Houphouët met with
Kwame Nkrumah, the founding president of Ghana. Nkrumah favored a policy of
completely breaking ties with the British, while Houphouët favored a closer relation-
ship with the French. The two men agreed to meet again a decade later to see which
approach worked better. Nkrumah was deposed in a military coup six years after
independence (Woronoff 1972).
82
Côte d’Ivoire. The second major development was that of a bill in the
French Assembly to abolish forced labor in West Africa. Houphouët played
a major role in shepherding this bill through the legislative process. This
achievement won Houphouët the national spotlight as the Ivorian politician
who stood up to the French. From that point on, his moral and political
authority within Côte d’Ivoire would be unparalleled. I will discuss both of
these major developments in detail below.
In 1946, a law was passed that granted French citizenship to all people
living in French colonial territories. Citizenship came with the ability to
vote and participate in French elections.7 The ability to elect representa-
tives for the French Assembly provided Côte d’Ivoire limited measures
of political representation and a seat at the table. It also provided
Houphouët, who would serve for fourteen years in the French National
Assembly, with the opportunity to consolidate power as the top politician
in Côte d’Ivoire.
That same year an election was held to select a politician who would
represent Côte d’Ivoire in Paris. This was the first real opportunity that
most Africans had to express their political concerns and preferences
within the French Constituent Assembly. For these elections, European
and African politicians formed candidate lists that were composed of both
African and French representations. By contrast, Houphouët formed a list
that consisted of a coalition of Côte d’Ivoire tribes and foreign Africans,
one of the few proposals to exclude whites altogether. As a result, many
other African candidates withdrew their lists in support of Houphouët’s
slate. Houphouët’s list easily won both rounds of voting, and he was
elected to the French National Assembly (Zolberg 1969, 69).
A year later, building on the foundations of the SAA, Houphouët
formed his own political party, the PDCI. During the 1946 French
Constituent Assembly elections, Houphouët ran under his new party
label, and swept the elections, winning 98 percent of the vote in Côte
d’Ivoire districts (Zolberg 1969, 76). When Houphouët won, the PDCI
became the first successful independent African political party. Within
a year, the PDCI had an estimated 80,000 members, and from then on, the
PDCI gained so much influence that no African candidate could be elected
without the endorsement of the party, and by extension, Houphouët. The
French colonial administration often tried to undermine the PDCI. Prior
to elections, they promoted anti-PDCI candidates and tried to organize
ethnic and regional associations alliances that could challenge the PDCI’s
7
Initially, however, suffrage was far from universal. Many restrictions were placed,
for instance, French literacy was a requirement for voting. In 1956, the Loi Cadre bill
was passed, guaranteeing universal suffrage for all French citizens.
83
monopoly over political power. All efforts failed. Houphouët continued

to dominate elections, even as other PDCI candidates lost.
Houphouët’s continued presence in the French Constituent Assembly
had an important consequence. Representatives of the French Assembly
enjoyed a special privilege available only to its members: they were
immune from prosecution and could not be put in jail. After failing to
fully discredit the PDCI, the French administration began to use coercive
strategies to eliminate the influence of African elites by prosecuting and
jailing politicians once they lost an election (therefore losing their parlia-
mentary immunity privilege). Yet Houphouët continued to win every
election and therefore was immune from persecution. On the eve of
independence, he stood alone at the top of Ivorian politics (Gyimah-
Boadi and Daddieh 1999).
The second most important development in Houphouët’s ascension to
power was his successful proposal to outlaw forced labor within the colonies
in French West Africa. Forced labor was established by the French in Côte
d’Ivoire in the 1920s. Côte d’Ivoire was one of the few countries in West
Africa with a significant settler population. French settlers owned about
a third of cocoa and coffee plantations in the country. In reaction to increased
migration of African farmers from Côte d’Ivoire to Ghana, the colonial
government established the practice of forced labor in Côte d’Ivoire.
Although those who were “recruited” in this fashion “were paid by their
employers, wages were extremely low, and the government strictly super-
vised the performance of the contract they were forced to accept” (Zolberg
1969, 56). This practice was especially detested by Ivorians, who viewed the
system as one that worked to the sole advantage of the Europeans.
On March 1, 1946, a bill was proposed by Houphouët calling for the
abolition of all forms of forced labor in overseas France. The French
viewed this as a minor item – they did not even hold a floor debate to
discuss the bill, and it was passed. To Africans, however, this was one of
the most significant achievements of Africa’s emancipation from France
(Woronoff 1972, 36–37). To people within Côte d’Ivoire, the bill not only
removed a hated symbol of colonial rule throughout all of West French
Africa. It was also the achievement of an Ivorian – therefore elevating both
Houphouët and his independent movement across the region.
“Overnight, Félix Houphouët-Boigny became a mythical hero who had
imposed his will upon the French” (Zolberg 1969, 74).
For many years, people throughout the country – often encouraged by
the PDCI – firmly believed that unless Houphouët remained in office, forced
labor would be reinstated. The gratitude he earned from his countrymen
“has remained a foremost element in his political power and it has prevailed
over the hesitations of many followers who questioned his later policies”
84
(Zolberg 1969, 74–75). To provide a sense of what this bill meant to

Africans, a member of the Legislative Assembly explained:
It was very simple. The peasants of my region suffered a great deal from forced labor.
In my own family, there were twenty-four men and boys, but most of the time there
were few around to work for the family. The others were out slaving for the comman-
dant. When politics began, President Houphouët said he would do everything to
eliminate this barbarous practice, I decided that this was for me, and I followed him.
We didn’t know whether he could do it, but he kept his word. Since then, I have
followed him blindly and I shall continue to do so as long as he remains faithful to this
principle. (Interview with El Hadj Moussa Kone, 1959, qtd in Zolberg 1969, 75)
After years of incremental representation in the French National

Assembly, Côte d’Ivoire finally gained independence on October 31,
1960. The country’s first constitution was drafted by Houphouët and
a small circle of elites, and it was approved by the National Assembly
unanimously without debate (Zolberg 1969, 252). Presidential and legis-
lative elections were held, and Houphouët was the only presidential
candidate on the ballot. On November 27, 1960, he became the first
president of Côte d’Ivoire.
3.3.3 Houphouët In Power: Ruling Without Constraints

Upon taking power, Houphouët ruled Côte d’Ivoire as a one-man
dictator. Zolberg (1969), for instance, notes that the independence con-
stitution was “framed by a government and an assembly dominated by
a single party, itself controlled by one man” (25). Houphouët’s strategies
of personalist rule encompassed several important factors. For one, he was
charismatic. Handloff (1991) discusses the “myth of Houphouëtism,”
where the leader portrayed himself as the paternalistic founding father
of the country. Houphouët transformed himself into a “transcendent
symbol of unity to the disparate groups in Côte d’Ivoire, and his charis-
matic authority supplanted the traditional authority of the local chiefs”
(143). In fact, as McGovern (2011) notes, “the cult of personality that
grew around Houphouët was so great that he was able to move the capital
of Côte d’Ivoire to his home village of Yamoussoukro in 1983” (16).
Houphouët also benefited from strong economic growth in the first two
decades following independence. From 1960 to 1980 Côte d’Ivoire
experienced one of the highest rates of economic growth among the nonoil
exporting African countries. Houphouët resisted imposing high taxes on
cash crops to fund urban development projects, in part because his own
group, the Baoulé, dominated cocoa and coffee farming (Arriola 2013).
Moreover, he maintained financial ties with France, relying on French
investment, technology, and funding to develop the country’s economy
85
and administrative infrastructure (Handloff 1991, 144). The “Ivorian

Miracle” resulted in the development of impressive roadway and highway
systems and Abidjan, the main economic hub, became known as the
“Paris of West Africa” (Rabinowitz 2018, 74).
Most importantly, Houphouët did not empower other elites through
institutionalized power-sharing mechanisms, and instead, retained his sta-
tus as the single most powerful politician in the country. The first part of
this strategy was the use of French bureaucrats to run the state, rather than
appointing Africans. There were many educational requirements to work
within the bureaucracy, and Houphouët maintained these restrictions.
Since Africans had limited opportunities to obtain a university degree,
they were shut out of top positions. The highest ranks of the state remained
in the hands of European technocrats (Zolberg 1969, 100–101). As
Jackson and Rosberg (1982) note, “the French managerial elite in particu-
lar has provided loyal and capable agents who have contributed greatly to
the overall effectiveness of both the state and the private sectors of the
economy and enabled Houphouët to monopolize political power” (146).
Furthermore, Houphouët shut out the most influential members of
his own ruling coalition from the cabinet. In making government
appointments, Houphouët maintained a relative balance of ethnic
representation.8 He used an informal ethnic quota system to fill important
party, government, and administrative positions (Zolberg 1969, 141). Yet
within his presidential cabinet, “Houphouët has been careful to insure
that all major ethnic groups are represented – but not by their most
prominent politicians” (Jackson and Rosberg 1982, 148 emphasis
added). In this way, Houphouët was able to maintain support of various
groups through ethnic balancing, while at the same time, preventing the
strongest elites from accessing state power. “The cabinet is less a collegial
body of powerful and independent incumbents and more a technical
advisory body to the ruler. In recent years there has been an increase of
young technocrats both in the cabinet and in the bureaucracy. Of the
thirty-six ministers in 1974, only six were Houphouët’s old political party
colleagues” (Jackson and Rosberg 1982, 147–148).
In addition, the Ivorian constitution was extremely weak under
Houphouët. Presidential power was expanded by combining it with the
functions of prime minister while subordinating the role of the National
Assembly (Handloff 1991, 147–148). Côte d’Ivoire was also the first of
the former French colonies to abandon the parliamentary system and
adopt the presidential system. The chief executive, who is both head of
8
Though he still favored his own Baoulé group, especially for top positions.
86
state and head of government, is given the power to appoint a cabinet that
is not accountable to the National Assembly (Zolberg 1969, 108).
The ruling party and legislature under Houphouët was similarly weak.
The PDCI, which had been a political machine during the independence
era, began to atrophy once Houphouët came into power. Though the PDCI
routinely won all seats in the National Assembly, this was purely because
opposition parties were shut out. By the mid-1960s, party structures fell
prey to factionalism and were “inadequate instruments of control over the
administration and over the population” and “did not provide an effective
network of political communication” (Zolberg 1971, 21). Similarly, under
Houphouët the National Assembly served only to rubber stamp executive
decrees and instructions (Jackson and Rosberg 1982, 147–150).
3.3.4 Weak Institutions, Regime Collapse: The Aftermath

of Houphouët’s Death
Beginning in the 1980s, due to a sharp decrease in world cocoa prices,
Côte d’Ivoire experienced a severe economic crisis. To make matters
worse, Houphouët refused to export Ivorian cocoa from 1987 to 1989,
in hopes of driving world prices up, but this strategy failed, severely
weakening the economy of Côte d’Ivoire (McGovern 2011, 145). The
country’s debt increased, and crime rose dramatically in Abidjan due to
unemployment. In 1990, hundreds of civil servants went on strike, forcing
Houphouët to allow multiparty elections (Gyimah-Boadi and Daddieh
1999). On October 28, 1990, Côte d’Ivoire held its first multiparty elec-
tions. The main opposition candidate was Laurent Gbagbo, the leader of
an opposition party, the Front Populaire Ivoirien (FPI). Houphouët won
the election, receiving 82 percent of the vote, and remained in office.
However, the question of presidential succession loomed large.
Houphouët, now 85 years old, had always avoided designating a clear
successor. Constitutional rules regarding the succession order had chan-
ged several times throughout the three decades of Houphouët’s rule. The
designated successor vacillated between the vice president (though this
office was always left vacant) and president of the National Assembly.
Furthermore, the nature of “one-man politics [raised] the question of
succession” (Jackson and Rosberg 1982, 1152). Due to the fact that
Houphouët ruled Côte d’Ivoire as a personalist dictatorship where
power was centralized around him, there was not an obvious politician
of similar stature who would clearly serve as the presidential successor.
After public speculation about the presidential succession for over
a decade, Houphouët was finally pressured into naming a successor,
designating Henri Konan Bédié for the job (Rabinowitz 2018, 201–202).
87
Henri Konan Bédié had served in the Ivorian government since inde-
pendence. He was Côte d’Ivoire’s ambassador to the United States and
Canada, then the minister of economy and finance. In 1980, Bédié was
elected to the National Assembly, where he was then subsequently elected
as the president of the National Assembly, the position he served from
1980 through to Houphouët’s death in 1993.
In 1987 the constitution was amended to specify that the president of
the National Assembly would succeed the president in the event of the
incumbent’s death or incapacitation. Despite the creation of this constitu-
tional rule, Houphouët did not publicly support Bédié as his designated
successor. The lack of clear and early support from Houphouët of his
designated heir resulted in Bédié’s lack of legitimacy as the presidential
successor.
In 1991, two years before Houphouët’s death, a power struggle over
the impending presidential succession emerged between Bédié and
Houphouët’s newly designated prime minister, Alassane Ouattara. The
position of prime minister was created in 1991, and it was the first time
a second in command position had been created and filled in the presi-
dential cabinet since independence in 1960. In response to the financial
crises that had developed in the 1980s, the International Monetary Fund
(IMF) pressured Houphouët to appoint Ouattara, a former IMF econo-
mist who had been educated in the United States, as prime minister. Two
factions, one supporting Bédié and the other supporting Ouattara
emerged during this period (Konate 2004; McGovern 2011).
Houphouët died on December 7, 1993, and immediately following his
death, Bédié announced on state television that he had assumed the pre-
sidency. However, having failed to consolidate power prior to Houphouët’s
death, Bédié entered power weak. On one hand, there is some evidence that
Bédié tried to institutionalize his regime. Important cabinet positions, such
as prime minister and defense minister, were always filled and were rela-
tively stable. However, Bédié shut out one key challenger: Ouattara. He
purged the cabinet of anyone who supported Ouattara and introduced
a new electoral law that was clearly aimed at preventing Ouattara from
running in future elections (McCauley 2017; Rabinowitz 2018).
On December 23, 1999, Bédié was ousted in a coup that was staged by
members of the military who were linked to Ouattara. While Bédié was
out of the capital celebrating Christmas with his family, a group of
soldiers took over the airport, government television station, and radio
station. Many of the officers who led the coup had personal connections
with Ouattara – one officer had been a member of Ouattara’s guard. The
armed forces did not step in to defend Bédié, and the regime ended due to
the coup (Rabinowitz 2018).
88
3.4 Power Sharing in Presidential Cabinets
The regime in Côte d’Ivoire fell due to the absence of institutionalized

power-sharing mechanisms. Since Houphouët entered office extremely
powerful and influential, he did not need to establish institutional
constraints in order to share power with other elite opponents: viable
challengers simply did not exist. Although Houphouët was able to
remain in power without institutionalizing, the regime did not persist
long after his death. Due to the lack of clear succession planning, Bédié
entered office weak and lacking legitimacy as Houphouët’s heir.
Although there is some evidence that Bédié tried to use ministerial
appointments to coopt elites, he shut out his prime opponent,
Ouattara, which later led to a coup.
3 . 4 e m p i r i ca l a n al y s i s o f p o w e r sh a r i n g
in pres idential cabinets
In this final section, I analyze data on ethnicities of cabinet ministers to
demonstrate that power sharing was more prevalent in the Ahidjo regime
compared with the Houphouët regime. While ethnic favoritism prevailed
under Houphouët, Ahidjo routinely appointed non coethnics to his cabinet,
even appointing leaders from other ethnic groups to top cabinet positions,
such as prime minister, minister of finance, and minister of defense.
I use data from Francois, Rainer, and Trebbi (2015), which includes
comprehensive data on the ethnicities of cabinet ministers in fourteen
African countries from 1960 to 2010, including Cameroon and Côte
d’Ivoire. The descriptive statistics paint a clear picture of ethnic favoritism
in Côte d’Ivoire in contrast to elite power sharing in Cameroon.
Ahmadou Ahidjo, the first president of Cameroon, was an ethnic
Fulani, but he appointed many non coethnics to his cabinet. On average,
Ahidjo had twenty-three ministers in his presidential cabinet. The smallest
cabinets, from 1963 to 1964, had thirteen ministers, and the largest
cabinet, in 1982, had thirty-three ministers. There are 496 total minis-
ter–year observations (excluding the president) during Ahidjo’s rule from
1960 through 1982. Summary statistics are presented in Appendix Table
3.1.
Ahidjo did not favor ethnic Fulanis in his cabinet, only assigning
12 percent of all cabinet ministerial positions to his coethnics. The same
pattern emerges, even if we restrict the sample to top ministerial posts:
only 24 percent of top ministerial posts were assigned to ethnic Fulanis.9
9
Top ministerial posts, as defined by Francois, Rainer, and Trebbi (2015), include vice
president, prime minister, defense, budget, commerce, finance, treasury, economy,
agriculture, justice, and state/foreign affairs.
89
Moreover, Ahidjo offered cabinet positions to elites from opposing

groups. The Bamileke and Bassa groups were the leaders and supporters
of the UPC party, which was the main challenger to Ahidjo’s administra-
tion. Yet, 33 percent of total cabinet portfolios and 24 percent of top
ministerial positions were assigned to Bamileke or Bassa elites.
By contrast, Félix Houphouët-Boigny, the first president of Côte
d’Ivoire, systematically favored his coethnic Baoulés in his cabinet appoint-
ments. On average, Houphouët had twenty-five ministers in his presidential
cabinet. The smallest cabinet, in 1964, had ten ministers, and the largest
cabinet, in 1987, had forty ministers. There are 862 total minister–year
observations (excluding the president) during Houphouët’s rule from 1960
through 1993.
In contrast to Ahidjo, Houphouët did favor ethnic Baoulés in his
cabinet, assigning 26 percent of all cabinet ministerial positions to his
coethnics. Tellingly, 50 percent of all top ministerial posts were assigned
to Houphouët’s coethnics.
Figure 3.2 graphs the percentage of coethnics each leader appointed to
top ministerial posts throughout their tenure. The graph shows that
Ahidjo often shared power with important regional elites by appointing
leaders of other ethnic groups to top cabinet positions. As a result,
the share of top cabinet positions that went to elites from his own group
were often quite low. By comparison, Houphouët did not use cabinet
appointments as a power-sharing mechanism, and appointed elites from
his own ethnic group at a much higher rate.
A possible counterargument to this analysis is that leaders in Africa
simply assign cabinet portfolios in accordance with the population share
of various ethnic groups (Francois, Rainer, and Trebbi 2015). Since Fulanis
(Adhijo’s group) consists of 9 percent of the population in Cameroon, we
should expect Ahidjo to assign a smaller share of cabinet positions to Fulani
elites. On the other hand, the Baoulé (Houphouët’s group) consists of
21 percent of the population in Côte d’Ivoire. One can make the counter-
argument that Houphouët distributed many cabinet ministries to elites of
his own group simply because he was just making cabinet appointments
that were consistent with the population size of the group.
Yet this argument does not hold when we examine changes to cabinet
appointments over time. The population sizes of ethnic groups in Cameroon
and Côte d’Ivoire generally do not change over time (Vogt et al. 2015). Yet,
we do see differences in the distribution of cabinet positions to coethnic elites.
In the 1980s, Côte d’Ivoire experienced a severe economic crisis.
Unemployment and crime rose dramatically, and Houphouët’s government
was blamed for these conditions. From Figure 3.1 we can see that there was
90
3.5 Conclusion
Percent coethnics appointed to top ministerial post
.8
.6
.4
.2
0
1960 1965 1970 1975 1980 1985 1990 1995
year
Cameroon Côte d’Ivoire
Figure 3.2 Ethnic favoritism in president cabinets

Note: The graphs portray the percent of top cabinet positions that were assigned to elites
from the same ethnic group as the president.
a sharp drop in coethnic appointments in Houphouët’s cabinet, going into

the 1980s, the period where Houphouët was significantly weaker.
The proportion of coethnic cabinet appointments under Ahidjo also
fluctuated with time. Figure 3.1 shows that when Ahidjo initially came
into office he made virtually no coethnic cabinet appointments. As this
chapter discusses, this was because Ahidjo was especially weak when he
first became president upon independence. Over time, as Ahidjo consoli-
dated his authority, he began to make more coethnic appointments. Yet, it
is important to note that despite these changes over time for each leader,
ethnic favoritism was always higher in Côte d’Ivoire under Houphouët
compared with in Cameroon under Ahidjo.
3.5 conclusion
This chapter presented two case studies that illustrated the mechanisms and
findings articulated by the formal model in Chapter 2. First, I discussed the
case of Cameroon, a highly institutionalized regime under the presidency of
Ahmadou Ahidjo. Ahidjo entered power weak relative to other elites,
largely because he inherited his position of power from the outgoing
French colonial authorities. I argued that in order to retain support from
other elites, Ahidjo established checks on his own authority and delegated
91
power to other elites through the appointment of key cabinet positions.

Importantly, he designated a prime minister, Paul Biya, who was the
constitutional successor to the president. The leadership transition from
Ahidjo to Biya occurred peacefully and Biya remains in power today.
I then presented the case of Côte d’Ivoire, a weakly institutionalized
regime under the presidency of Félix Houphouët-Boigny. Houphouët,
who had been a renewed independence leader in the decades prior to his
presidency, entered power extremely strong relative to other elites. As
a result, he did not have to make institutional concessions to other elites in
order to remain in power simply because there were no other elite chal-
lengers of his stature. During most of Houphouët’s tenure, key cabinet
positions were left vacant and a constitutional successor was not estab-
lished. Upon his death, Houphouët’s successor (who had essentially been
designated at the last moment) took office but was soon deposed by the
military in a coup. Though Houphouët was able to remain in power as
a strongman leader, the regime fell soon after his death.
The next chapter will present data on regime institutionalization in all
African dictatorships from 1960 to 2010. While this chapter presented
more fine-grained evidence of the theoretical argument and mechanisms,
we will now take a step back and examine systematic evidence of the
relationship between initial leader strength and regime institutionalization.
Appendix
Appendix Table 3.1 Summary statistics
Ahidjo Houphouët
(Cameroon) (Côte d’Ivoire)
Mean number of cabinet ministers 23 25

Total number of minister–year observations 496 862
Number of minister–year observations with 59 220
coethnic appointee
Percentage of cabinet appointments with coethnic 12% 26%
appointee
Total number of minister–year observations for 154 295
top posts
Number of minister-year observations with 37 118
coethnic appointee for top posts
appointee for top posts
92
3
Two Illustrative Cases
How do institutions constrain leaders and under what conditions do

autocrats establish institutionalized regimes? The model in Chapter 2
produced three important mechanisms and findings that address these
questions. This chapter will present illustrative case studies of
Cameroon (a highly institutionalized regime) and Côte d’Ivoire (a weakly
institutionalized regime), two countries in sub-Saharan Africa that have
experienced decades of authoritarian rule. The goal of this chapter is to
use these case studies to highlight key insights from the model.
I choose to focus on Cameroon and Côte d’Ivoire due to the compar-
ability of these countries. They share similar histories, economies, geogra-
phy, and populations. Both cases are former French colonies1 located in
West Africa, and they were granted independence in the same year. Both
countries have similar population sizes and degrees of ethnic heterogeneity
within the population. They both have large port cities along the Atlantic
coast and generally stable economies. In fact, as Figure 3.1 illustrates, the
gross domestic product (GDP) of both Cameroon and Côte d’Ivoire has
grown along very similar trajectories from 1960 to 2010. These similarities
allow me to hold background conditions constant while exploring impor-
tant differences in the independent and dependent variables.
I first present the case of Cameroon, which is an example of a regime
with high levels of institutionalization under the founding president,
Ahmadou Ahidjo. I show how Ahidjo, who was an initially weak leader,
used institutional bargains in order to maintain support from other elites –
therefore establishing a rule-based system. Unlike other founding leaders,
1
As will be discussed later in this chapter, a small sliver of Cameroon was controlled
by the British, but the country was mostly under French colonial rule. British rule in
Cameroon was very limited, as the French took the vast majority of the land (four-
fifths of the total area of Cameroon) – including the capital city of Yaounde.
Francophile elites and politicians have always dominated the executive in Cameroon.
64
2000
1500
GDP per capita
1000
500
0
1960 1970 1980 1990 2000 2010 2020
Year
Côte d’Ivoire Cameroon
Figure 3.1 Yearly gross domestic product (GDP) of Cameroon and Côte d’Ivoire
such as Léopold Sédar Senghor of Senegal or Félix Houphouët-Boigny of

Côte d’Ivoire, Ahidjo was not a renowned, charismatic, popular indepen-
dence leader. He was initially encouraged to run for office by the French
colonial authorities and was perceived to be a part of the colonial
machine. When independence was granted, Ahidjo ascended to the pre-
sidency as a highly unpopular leader. To compensate for this initial lack of
support, Ahidjo distributed important cabinet positions to other elites,
including appointing Paul Biya to the position of prime minister – the
designated constitutional successor to the president. In 1982, power was
transferred peacefully from Ahidjo to Biya, and the same regime remains
in power today.
I juxtapose the case of Cameroon against Côte d’Ivoire under the rule
of the founding president, Félix Houphouët-Boigny, which is an example
of a regime with low levels of institutionalization. Houphouët was
a renowned independence leader, who lobbied for the right to self-
governance throughout French West Africa. Upon taking power,
Houphouët was extremely influential and faced very few credible chal-
lenges to his authority. Throughout his tenure, Houphouët centralized
power within his cabinet, leaving key ministerial positions, such as the
vice presidency, vacant. Houphouët died while in office in 1993, and
power was transferred to Henri Konan Bédié. However, Bédié’s claim to
the presidency was tenuous, as Houphouët had intentionally left
65
succession plans vague while he was in power. The military launched

a coup d’état in 1999 and removed Bédié from office.
Before I present the cases of Cameroon and Côte d’Ivoire, I will first
summarize the main claims developed in Chapter 2. This summary will
highlight the connective tissues that link the analysis from the formal
model with substantive details of the cases.
3.1 t he formal model summarized

The model presented in Chapter 2 produced a number of insights and
findings. Here, I summarize three key aspects of the model that will be
highlighted in the case studies. First, I argued that institutions constrain
leaders only when they empower specific elites. The model formalized this
mechanism as a shift in the underlying distribution of power away from
the leader. Executive constraints have the power to bind the leader’s hands
when they provide elites with direct access to the state. When an elite is
given an important cabinet position, such as the vice presidency or min-
ister of defense, they gain control of the entire ministry and its resources.
Within the context of sub-Saharan Africa, this is an especially valuable
source of power because the government was and continues to be a main
source of funding for development projects, employment, and the distri-
bution of resources to localities. Having authority over a piece of the
central government provides elites with a great deal of power.
The case of Cameroon will demonstrate how elites who were given
important cabinet positions used this access to power to build up their
own coalitions. Cabinet ministers frequently fill their ministries with
coethnics and bureaucrats from their home region. They have discretion
over funding and the distribution of resources to localities. State coffers
provide elites with a rich access to sources of patronage in order to reward
their supporters.
Moreover, when an elite is appointed to an important ministerial
position, they become a focal point for other elites. This is especially
true if the constitution specifies a succession order. If, for example, the
vice president is listed as the designated successor in the constitution, then
the elite in the position of vice president is formally and publicly declared
as the second highest ranked member in the government. Executive con-
straints that empower particular ministers help solve coordination pro-
blems amongst elites by providing a clear signal of the most influential
elites in the regime. This eliminates conflict within elites about who would
be the most likely challenger to the leader. Similar to the “crown prince
problem” famously identified by Herz (1957), when the leader identifies
66
3.1 The Formal Model Summarized
a successor, that person automatically becomes a potential challenger

simply by the designation of an alternative center of power.
This mechanism is well illustrated by a comparison of the first leader-
ship transitions in Cameroon and Côte d’Ivoire. In Cameroon, Ahidjo
appointed Biya as his prime minister in 1975 – the constitutionally
designated position to succeed the president. Seven years later, Ahidjo
voluntarily retired and passed power to Biya, citing health reasons. In
part, due to the legitimacy of the constitutional succession procedure, Biya
maintained control of the regime – even in light of an attempted coup
a year after his ascension to the presidency. By contrast, in Côte d’Ivoire
Houphouët largely avoided designating a successor, keeping the position
of vice president vacant throughout most of his rule. Toward the end of his
rule, he reluctantly designated Bédié as his successor, but factional battles
over succession persisted, and Bédié was not able to consolidate power
upon Houphouët’s death. Six years later, Bédié was deposed in the coun-
try’s first coup d’état since independence.
Second, the model showed that leaders who initially enter power weak
must make the counterintuitive decision to give power away to elites in
order to maintain support of the regime. Leaders who enter power already
strong face weak elites who do not pose a credible threat of launching
a successful rebellion against the regime. Strong leaders can therefore
always pay off weak elites without making long-term institutional com-
mitments. Leaders who enter power weak, on the other hand, face
a commitment problem in the initial period. In this case, if elites were to
try to depose the leader, they would likely succeed. In order to survive in
power past this initial period, weak leaders must make institutional guar-
antees of rent distribution to elites. This is achieved by creating executive
constraints that empower elites, solidifying their future ability to hold the
leader accountable by transferring state resources and influence to them.
The comparison of Cameroon and Côte d’Ivoire highlights this finding.
Ahidjo entered power in Cameroon extremely weak due to the perception
that he was an ally of the French colonial administration and therefore
had simply inherited his position of power from the outgoing authorities.
As a result, Ahidjo could not rule alone; he used important cabinet posi-
tions to make institutional commitments to elites and solidified limitations
on his personal executive authority through constitutional rules. On the
other hand, Houphouët entered office after independence as the single
most powerful politician in Côte d’Ivoire. He had long built a reputation
as a freedom fighter not only within his own country but also for all
Africans in the region. Upon ascending to the presidency, he systemati-
cally shut out elites from his own party and coalition from positions of
power, preferring to keep French technocrats in the bureaucracy. This
67
strategy of personal rule allowed Houphouët to remain the single most

important figure in Ivorian politics throughout his tenure.
Initial weakness has long-term consequences. Leaders who enter power
especially weak must institutionalize the regime in order to avoid being
deposed in the first period. This is true regardless of whether the leader
would have been stronger in future periods; without institutionalizing
after coming into power, the leader would not have made it to a future
period! The period right after a leader comes into power is therefore
a critical juncture: the initial distribution of power between the leader
and elites determines the institutional makeup of the regime in the long
run. Since institutionalization shifts power in favor of elites, it is self-
reinforcing. The leader’s initial decisions to create executive constraints
shape the long-term trajectory of the regime.
The institutional arrangements established under Ahidjo largely
remain in Cameroon today. Biya has been in power since 1982, and
there has never been a coup d’état or civil war in Cameroon.2 The regime
in Côte d’Ivoire fell six years after Houphouët’s death, as his successor
was deposed in a coup in 1999. There have also been two civil wars – the
first from 2002 to 2007 and the second from 2010 to 2011 – since the first
presidential transition.
Third, the model illustrated that initially weak leaders who insti-
tutionalize the regime and strong leaders who do not institutionalize
can both stay in power for long periods of time. Indeed, in the model,
conflict never occurs because initially weak leaders prefer to institu-
tionalize in order to maintain support of elites. Likewise, conflict also
never occurs in regimes with strong leaders because elites do not pose
a credible threat. The empirical implication of this finding is that we
generally should not observe a systematic relationship between leader
longevity and executive constraints because institutionalization is an
endogenous process.
Ahidjo was the president of Cameroon for twenty-two years, from
1960 until his voluntary retirement in 1982. Houphouët was the president
of Côte d’Ivoire for thirty-three years, from 1960 until his death in 1993.
It is also difficult to know how long Ahidjo would have remained in the
presidency if he had not voluntarily retired due to ill health. Regardless,
both leaders remained in power for decades despite two very different
models of rule.
2
Though in recent years an anglophone separatist movement has emerged in
Cameroon. The country has also experienced insurgent violence from Boko
Haram. Cameroon experienced an unsuccessful coup attempt in 1984, as this chap-
ter will later discuss.
68
3.2 cameroon under ahidjo: high levels o f

regime institutionalization
The case of Cameroon under the presidency of Amadou Ahidjo is an
example of a regime with high levels of institutionalization. Cameroon
has consistently had formal succession policies written into the constitu-
tion since the country gained independence. Article 9 of the constitution
states: “In the event of the Presidency falling vacant for any reason
whatsoever, the powers of the President of the Federal Republic shall
automatically devolve upon the Vice-President until such time as a new
President is elected.” In 1975, the title for the office of vice president was
changed to prime minister, and the succession clause in the constitution
was amended to reflect the change in title for this office. This stands in
contrast with the case of Côte d’Ivoire, in which the constitution under
Houphouët lacked detailed succession rules almost the entire time he was
in power.
Key cabinet ministerial positions were consistently filled and remained
quite stable under Ahidjo. Importantly, the office of vice president/prime
minister was almost always filled.3 Not only is this position the “second in
command” within the cabinet, the person in this office is the designated
constitutional successor. In addition, cabinet rotation rates under Ahidjo
were relatively low. Throughout his tenure, Ahidjo only had three different
vice presidents. Ahidjo also appointed a minister of defense the entire time
he was in power. This pattern of cabinet stability contrasts starkly with the
Houphouët government, which did not have a vice president or
prime minister over 90 percent of the time Houphouët was in power.
Houphouët also kept the minister of defense position vacant a third of the
time he was in power.
The fact that Cameroon, especially in the first decades after indepen-
dence was granted, had high levels of regime institutionalization may
come as a surprise. The existing scholarship on colonial legacies generally
claims that former British colonies had stronger institutions that kept
rulers in check as well as more robust legal traditions (Hayek 1960; La
Porta et al. 1998; Landes 1998). Cameroon was largely a French colony
prior to independence, yet institutional checks on executive power were
established, and most of these constraints were created early on during
Ahidjo’s rule.
3
For a short period during Ahidjo’s rule, from 1973 to 1974, the office of the vice
president was eliminated from the cabinet. In 1975 the title of the office of vice
president was changed to prime minister and was added back to the cabinet. From
1975 through until the end of Ahidjo’s rule, Paul Biya was appointed as prime
minister.
69
For the past three years, Cameroon’s English speaking regions (the
areas formerly ruled by the British) have started to engage in separatist
conflict against the French-dominated Cameroon government. However,
for most of Cameroon’s history – and especially during the time period
that is highlighted in the case study for this book (1960–2010) – this
linguistic cleavage was not particularly salient. Moreover, I argue that
Cameroon is an example of a regime with high levels of institutionaliza-
tion, making it durable against threats. The fact that the regime in
Cameroon has persisted, despite these separatist conflicts, provides addi-
tional support for the argument that the regime is highly institutionalized.

Cameroon is a country located in Central Africa though it borders the
Atlantic Ocean on its western side. It currently has a population of
23.44 million and about 55 percent of the population lives in urban
areas (World Bank 2016). Cameroon shares a border with Nigeria,
Chad, the Central African Republic, Equatorial Guinea, Gabon, and the
Republic of the Congo. The capital and second largest city of Cameroon is
Yaoundé, located in the central region of the country. Douala, located on
the Atlantic coastline, is the largest city in Cameroon, and is the economic
center of Cameroon and largest seaport in Central Africa.
Cameroon has an extremely heterogeneous population, with over 250
different ethnic groups. It is considered one the most linguistically diverse
countries in the world. Though groups such as the Bantoid, Bamileke,
Bantu, Kirdi, and Fulani comprise of large numbers, no particular ethnic
group dominates the population of Cameroon. Roughly two-thirds of the
population is Christian, and less than a quarter of the population, notably
the Fulani tribe (of which Ahidjo was a member), is Muslim.
The economy of Cameroon is roughly one of the ten largest economies
within sub-Saharan Africa. For the first two decades after independence,
agriculture made up the bulk of Cameroon’s economy, and important
cash crops included cocoa, coffee, cotton, and bananas. In 1977,
Cameroon began producing and distributing oil. It is now the sixth largest
oil producer in sub-Saharan Africa, though agriculture remains a large
part of the Cameroonian economy. Its current GDP is US$34.8 billion
(World Bank 2016).
Cameroon gained independence in 1960 after decades of colonial rule
by Germany, England, and France. Prior to colonization, the territory was
organized into chiefdoms and kingdoms, including the Bornu Empire.
Colonial rule in Cameroon began when Germany established the territory
as a German protectorate in 1884. Germany’s loss in the First World
70
War resulted in a partition of the territory between the French and the
British in 1916. This agreement divided the country into two territories –
the French took the vast majority of the land in the country (four-fifths of
the total area of Cameroon), and the British took the remaining land.
Under this agreement, the two Cameroons became League of Nations
mandated colonies under France and the United Kingdom (Le Vine
1964, 4–8).
3.2.2 Ahidjo’s Rise to Power: Initial Leader Weakness

Amadou Ahidjo, the first president of Cameroon, entered office weak and
vulnerable to being deposed. Unlike other founding presidents in newly
independent African countries, Ahidjo was not a national independence
hero. On the contrary, he largely inherited his position of power from the
colonial government and remained a close ally of France throughout his
tenure. As a result, Ahidjo was largely perceived to be a collaborator of the
French colonial authorities and was initially very unpopular. As Joseph
(1978) notes, “In spite of the efforts of the propagandists of the regime,
President Ahidjo cannot be considered a heroic figure from the nationalist
past . . . it is a matter of historical fact he was initially “fabricated” and
brought to power by the colonial regime” (76).4
Ahidjo was born in northern Cameroon to a Muslim Fulani family in
1924. His father was the local Fulani village chief. In 1939, Ahidjo began
attending Ecole Primaire Supérieur, a secondary school in Yaoundé, with
the goal of beginning a career in civil service. In 1942, Ahidjo joined the
state bureaucracy by becoming a post-office radio operator. Over the next
few years, his appointment in the civil service would take him to several
posts across Cameroon (Glickman 1992; Ngoh 1987).
Meanwhile, in Cameroon, emerging parties and political elites began to
lobby for independence. One of the earliest major parties to call for
independence and the reunification of British and French Cameroon was
the Union des Populations du Cameroun (UPC). The UPC, which was
formed in 1948, grew out of the French communist-oriented trade unions
that had formed near the end of the Second World War. The UPC was
strongly in favor of self-government, and two of its major platforms
involved independence from France and the reunification of British and
French Cameroon (Ngoh 1987).
4
71
In 1951, Ahidjo joined the Bloc Democratique Camerouanis (BDC) as

one of the founding members of the party. The BDC was much less radical
than the UPC, and it was initially opposed to lobbying for immediate
independence from France. In 1958, Ahidjo broke from the BDC and
formed his own party, the Union Camerounaise (UC).
It is important to stress that a cohesive and unified independence move-
ment never developed in Cameroon. This was partly because Cameroon
was divided into British and French territories, with each side having
regional factions with difference goals and approaches toward indepen-
dence (DeLancey 1989; Rubin 1971). More importantly, the colonial
administration actively stymied the independence movement by encoura-
ging the development of more moderate political parties. Since these other
political parties, such as the BDC, tended to favor the status quo rather
than call for immediate independence, the French supported the creation
of these parties to serve as a counterweight against the popular UPC. In
1955, the colonial administration banned the UPC, driving the party
underground (Ngoh 1987).
In addition to encouraging the development of politically moderate par-
ties that did not call for immediate independence, the French also sought out
African elites who would be reliable allies for the colonial administration.
They were eager to find young Cameroonian civil servants who were edu-
cated and trained by the French, and could act as a counterbalance against
the nascent independence movement inspired by the UPC (Joseph 1978).
In 1956, in response to growing agitations for independence from
French West African colonies, the French National Assembly passed the
“Loi Cadre,” which gave overseas territories increased self-governance
including the ability to create national assemblies and internal govern-
ments. Elections were held in 1956, and Ahidjo’s party, the UC, won
a large share of seats in the territorial assembly, since the more radical
(and popular) UPC party remained banned from politics.
Though Cameroon had not yet gained independence, the first
Cameroon government was formed out of these 1956 elections. The first
prime minister, Andre-Marie Mbida resigned in response to a ministerial
crisis and unpopular policies less than a year into the position (Le Vine
1964). This provided Ahidjo, who had been Mbida’s vice premier and
minister of the interior, with the opportunity to rise to the post of prime
minister in 1958 (Ngoh 1987, 150). As prime minister of the French
Cameroon territory, Ahidjo remained a close ally of the French authorities
and even conducted military attacks against the underground UPC inde-
pendence movement (DeLancey 1989, 40).
Cameroon achieved full independence from France on January 1, 1960,
and Ahidjo was elected as the first president of the newly independent
72
country three months later. A year later, British Cameroon voted to

reunify with the newly independent territory, and on October 1, 1961,
Cameroon became a federal republic. However, Ahidjo began his tenure
in a position of weakness, relative to other political and economic elites in
the country. There were three main reasons for his initial weakness.
First, Ahidjo was never perceived as a nationalist fighter, and in fact, was
accurately known to be a close collaborator of the French, allowing him to
inherit his position of power. The French authorities actually encouraged
Ahidjo to run for office in the first place because they viewed him as
a politician who would act in the interest of the colonial state. The colonial
administration even referred to the “Ahidjo option”: given that indepen-
dence seemed to be increasingly inevitable, the French authorities preferred
to have Ahidjo as head of state (Joseph 1978, 46). As DeLancey (1989)
notes, “Ahidjo owed his rise to the presidency of Cameroon entirely to
French manipulations” (281). Ahidjo was perceived largely as a “French
tool” and this image made him deeply unpopular (Le Vine 1964, 173). In
response, France made a number of attempts to convey the image that
Cameroon was a self-run government. In 1959, France initiated the Statue
of the Cameroun, which gave Cameroon full internal autonomy.
Yet the image of Ahidjo as a “French tool” was not a false perception.
In fact, Ahidjo maintained close relations with France throughout his
tenure in order to encourage continued French investment in the
Cameroon economy. Even after the Statue of the Cameroun was passed,
the statute still preserved French jurisdiction over money and exchange
rates, foreign policy, frontier security, and the right to intervene in the
event of armed insurrection or war. In fact, DeLancey (1989) argues that
Ahidjo’s ability to maintain support from the French by protecting French
investments was one of the key aspects of his strategy for rule (41, 58).
French business, commercial, and industrial investors dominated the
nonagricultural sectors of the economy in Cameroon, and major devel-
opment projects were often funded by private French investment. This
stands in stark contrast to the approach of many other African indepen-
dence leaders, such as Sekou Toure of Guinea, who voted for immediate
and complete severance of ties (including trade) with France in 1958.
A second major reason why Ahidjo entered power weak relative to
other elites is because his ethnic group, the Fulani, did not produce any of
the major exports. Though the Fulani dominated cattle herding and trade
in the northern region, they did not produce any of the major cash crops
and were not the leading economic elites in Cameroon. By contrast, ethnic
groups in the south, such as the Bamileke and Bassa, were the leading
producers of cocoa and coffee – the major cash crops in Cameroon. In the
years leading up to independence, one southern group in particular, the
73
Bamileke, became some of the richest farmers in Cameroon, and began to

expand into other areas of commerce. By the end of the Second World
War, the Bamileke began to compete for land and labor with French
plantation owners (Arriola 2013).
In the 1950s, the Bamileke became one of the colonial administra-
tion’s (and by proxy, Ahidjo’s) primary opposition forces. As they
became more economically powerful, the Bamileke began to seek poli-
tical influence. They threw their support behind the UPC – the radical
independence party that was banned by the French in 1955. By the time
Ahidjo became president in 1960, the Bamileke were a formidable
opposition – not only due to their economic strength but also their ability
to mobilize southern interests (Konings 2007; van de Walle 1993).
A third source of Ahidjo’s initial weakness was that Cameroon, having
been split into two colonial territories (British Cameroon and French
Cameroon), was not a unified state when Ahidjo first came to power.
Recall that French Cameroon gained independence in 1960 and British
Cameroon voted to reunify with the independent territory in 1961. Yet
from 1960 through 1963, Eastern Cameroon (which was formerly British
Cameroon) maintained its own separate cabinet in the federal govern-
ment. Even though Ahidjo was the president of the Cameroon, the regio-
nal government of East Cameroon was largely out of his political control.
It wasn’t until 1972 that Ahidjo was successful in abolishing the federal
system of government and centralized power in Yaoundé (DeLancey
1989, 54).
3.2.3 Ahidjo in Power: Creating Institutional Constraints

Upon taking power, Ahidjo needed to create official structures that would
allow him to buy the support of key elites. He did this by creating
constitutional limits on his own authority that empowered other elites
(in particular, by naming a prime minister who would serve as his succes-
sor) as well as the distribution of cabinet positions in exchange for support
from elites. As Takougang (2004) argues, “with the tremendous resources
of the state at his disposal, [Ahidjo] could use them to maintain the loyalty
of various groups and regions to himself” (75).
Ahidjo systematically used cabinet positions in order to secure support
from other elites. His ruling strategy is perhaps best summarized by
DeLancey (1989): “The essence of [Ahidjo’s] party was the cohesion of
a few important people, each of whom brought in his/her loyalists to the
party” (52). Importantly, the “important people” who were brought into
the government comprised of would-be challengers to Ahidjo.
74
[Ahidjo] treated his opponents firmly, sometimes harshly, but made sure that even
his bitterest enemies had both the chance of joining his side and of actively sharing
in the perquisites of rule. That he was never vindictive is to his credit: Mbida was
repeatedly offered various portfolios, Okala came out of prison to become an
ambassador, and several former UPC leaders have taken high and well-paying jobs
in government. (Le Vine 1964, 181)
Charles Assale, for example, the leader of the Mouvement d’Action

Nationale, a regional party in Eastern Cameroon, was offered the position
of prime minister of East Cameroon. “Such positions not only honored
the individuals concerned but also gave them access to sources of patron-
age with which to reward their supporters” (DeLancey 1989, 52). Elites
who opposed Ahidjo’s policies or refused to support his regime were shut
out of the cabinet and state bureaucracy.
Ministerial appointments provided “a majority opportunity for Ahidjo
to reward influential people in society – or even to build influence for
individuals – and to tie them to him” (DeLancey 1989, 59). Once
appointed, why did elites not defect? The positions in the presidential
cabinet gave these elites the ability to provide his followers with jobs and
resources he controlled through the ministry. This chain dependency is
illustrated by the following example: From 1970 to 1981 the Minister of
Posts and Telecommunications was Emmanuel Egbe Tabi. Tabi was from
the Anglophone region of Mamfe. During that time, Tabi filled his min-
istry with other Anglophones, who were otherwise underrepresented in
the predominately Francophone bureaucracy. Anglophones were even
appointed to post-office positions in French speaking parts of the country.
Moreover, many of the higher positions within the ministry went to
bureaucrats who were from Tabi’s home region – the Mamfe area
(DeLancey 1989, 59).
To sum up, on one hand, ministerial appointments empowered elites by
providing them the influence and resources to reward their supporters and
consolidate their authority. Yet on the other hand, these appointments
tied these ministers to the Ahidjo regime and provided them an incentive
to continue supporting the incumbent in order to maintain their own
positions within the cabinet. As Takougang (2004) argues,
through this system of appointments to cabinet and other important government
and party offices that was based on regional and/or ethnic representation, Ahidjo
was able to maintain the loyalty of most groups in the country. That is because
these high-ranking appointees in turn were able to tie their respective regions and
groups to the president by building their own network of support through the
provision of administrative and party jobs. (76)
Ahidjo also took great pains to make sure that elites from various
regions, ethnicities, and religions were represented in his cabinet. “The
75
cabinet as a group was an arena for Ahidjo to play ‘ethnic arithmetic’ or

‘ethnic balancing’” (DeLancey 1989, 60). He also ensured regional repre-
sentation: East (French) Cameroon and West (British) Cameroon were
both represented in the cabinet. Ahidjo’s presidential cabinet in 1975, for
example, included six ministers from the north, eight from the central-
south, seven from the west, and two from the east (Kofele-Kale 1986, 73).
Power-sharing, however, came at a cost. As this chapter will later show,
Ahidjo’s own ethnic and regional elites did not control the majority of the
cabinet seats. Power-sharing under the Ahidjo regime required that the
leader relinquish (some) control over the state.
Finally, Ahidjo also struck bargains with traditional chiefs. At inde-
pendence the lamidos (Fulani chiefs) still dominated the northern region
of the country. Even though Ahidjo was a Fulani from the north, he was
unpopular with the lamidos, who believed that Ahidjo wanted to mod-
ernize society, therefore weakening their authority (Ngoh 1987, 152).
To gain their support, Ahidjo created a constitutional stipulation that
stated that the status of traditional chiefs would be protected. He also
filled key cabinet appointments, such as the minister of the interior, with
elites that the lamidos favored. However, Ahidjo made it clear that in
exchange for these benefits, he expected the lamidos to support his rule
(Joseph 1978, 53).
It is important to acknowledge that, in addition to institutional strate-
gies of rule, Ahidjo also relied on repression and force to maintain his grip
on power on the wider population. Where certain parties or elites refused
to be coopted, Ahidjo used police and military repression. Though
opposition parties were technically legal, individuals who attempted to
establish other political parties were arrested or tortured (DeLancey 1989,
63). However, these repressive tactics were generally reserved for the
broader citizenry and potential opposition. For his most inner circle of
elites, Ahidjo relied on institutional forms of cooptation.
3.2.4 Ahidjo Constrained, Biya Empowered: The Presidential

Transition
Our discussion of Cameroon concludes with an example of how institu-
tions constrained the leader. In 1982, Ahidjo voluntarily retired from the
presidency due to health reasons, transferring power peacefully to his
prime minister and constitutionally designated successor, Paul Biya.
Though he stepped down, Ahidjo had hoped to retain informal control
of the presidency. These attempts culminated in a failed coup attempt,
which was successfully thwarted by Biya. Ahidjo had empowered Biya by
giving him access to the state as well as constitutional legitimacy as his
76
successor, therefore Ahidjo could not simply control Biya as a puppet

president. The first presidential transition in Cameroon provides a useful
test to examine whether institutionalization truly does shift power from
leaders to elites in a meaningful way. It also demonstrates how Ahidjo
became constrained by the very institutions he created while he was
in power.
Since independence, the Cameroon constitution has always included
stipulations regarding the process of leadership succession. Article 9 of the
constitution states: “In the event of vacancy of the Presidency for
whatever cause the powers of the President of the Federal Republic shall
without more devolve upon the Vice-President until election of a new
President.” In 1977, the title for the office of vice president was changed to
prime minister, and the succession clause in the constitution was amended
to reflect the change in title for this office (Article 7).5
When Ahidjo appointed Paul Biya as his prime minister in 1975, Biya
became the designated successor to the presidency. Biya served as Ahidjo’s
prime minister for seven years, leading up to the presidential transition.
Notably, Biya was a Christian from Southern Cameroon, in contrast with
Ahidjo, who was an ethnic Fulani Muslim from the north. Biya began his
political career in Ahidjo’s government soon after independence was
granted. He served in various ministerial positions, such as the director
of the Ministry of National Education and minister of state, and worked
his way up the cabinet (Glickman 1992).
On November 4, 1982, after being in power for twenty-two years,
Ahidjo – citing health concerns – voluntarily retired from the presidency
and passed power down to his appointed successor, Biya. In announcing
his retirement, Ahidjo expressed his support for Biya, stating in a radio
broadcast:
Fellow Cameroonians, my dear countrymen, I have decided to resign my duties as
Head of State of the United Republic of Cameroon . . . I request all of you
Cameroonians to give your full confidence and whole-hearted support to my
constitutional heir, Mr. Paul Biya. He merits the confidence of all, both at home
and abroad. (Ngoh 1987, 300)
Although Ahidjo stepped down from the presidency, he clearly expected

to retain influence and control over the Biya presidency. Notably, after
stepping down as president, Ahidjo continued to serve as the chairman of
the ruling Cameroonian National Union (CNU) party. In fact, in an inter-
view two months after the leadership transition, Ahidjo, whose only
5
From 1977 to 1979, the succession order was changed to the president of the
National Assembly and then prime minister. In 1979, the constitution was amended
so that the prime minister was the direct successor.
77
remaining official title was party chairman of the CNU, asserted that the
party takes precedence over the government (Konde 2012).
As Ahidjo’s prime minister Biya had maintained an image of
a technocrat and made it clear that he had no ambitions to compete
with Ahidjo (Mbaku and Takougang 2004). After acceding to the pre-
sidency, however, Biya made a number of administrative and policy
changes that diminished the influence of Ahidjo’s allies in favor of his
own power base. This included multiple cabinet reshuffles in which most
of Ahidjo’s appointees were replaced. Within a year Biya replaced over
half the cabinet he had inherited from Ahidjo. Biya also undertook admin-
istrative reforms, most notably dividing the northern region – Ahidjo’s
home district – into three provinces (CIA 1983; 1984).
In response to Biya’s consolidation of power, Ahidjo tried to reclaim his
influence. He reportedly attempted to introduce a constitutional amend-
ment that would increase his own authority as chairman of the party while
reducing the powers of the president. He asked his remaining ministers
who had not been replaced in the cabinet reshuffle to resign in protest –
none of whom did (Ngoh 1987, 309–310).
On April 6, 1984, members of the palace guard who were loyal to Ahidjo
attempted to launch a coup against Biya. It is believed, though not verified,
that Ahidjo himself help to orchestrate the coup. This coup failed, in part
because of tactical mistakes, and also because the coup plotters failed to
gain support of key members of the armed forces. Though the constitu-
tional succession provisions was clearly not the only reason Biya was able to
retain power in light of the failed coup attempt, it conferred legality and
legitimacy to his claim to the presidency (Hughes and May 1988).
Chapter 2 opened with the following observation by Montesquieu: “At the
birth of new polities, leaders mold institutions, whereas afterwards institu-
tions mold leaders” (qtd in Putnam 1993, 26). Much like Montesquieu’s
remark, after relinquishing power to his successor, Paul Biya, Ahidjo was shut
out of returning to power by the very institutions that he put in place himself.
Moreover, the constitutionally designated succession procedures ensured that
Ahidjo could not renege on promises to transfer power to Biya. As Ngoh
(1989) argues, Ahidjo “realized that after twenty-five years as the undisputed
strongman in Cameroon, he had to play second fiddle to Paul Biya – some-
thing he was not prepared to do” (319).
Paul Biya remains in power in Cameroon today – even surviving the
introduction of multiparty elections in 1992. How did he remain in power
despite demands for political liberalization in the 1990s? In large part,
Biya benefited from the fact that the transfer of power took place long
before pressures to introduce multiparty elections arose by the end of the
Cold War. By 1991, Biya had already been in power for almost a decade.
78
He therefore had already secured his tenure through a combination of

cooptation through institutionalized mechanisms of power sharing as
well as coercion by the time he was forced to introduce multiparty
competition.
By contrast, as the case of Côte d’Ivoire will demonstrate, Houphouët’s
successor, Bédié, came into office right after the introduction of multiparty
elections. In addition to his already weak claims of legitimacy due to the
absence of institutionalized succession mechanisms, he did not have the
benefit of time in the way that Biya had.
3 . 3 côte d’ivoire under houphouët : l o w l e ve l s

of regime institutionalization
Unlike Cameroon, Côte d’Ivoire is an example of a regime with low levels
of institutionalization. For most of Houphouët’s rule, specific constitu-
tional rules designating the successor order were not established. In 1975,
a constitutional succession order was created. Article 11 of the constitu-
tion stated: “In case of the death or incapacitation of the President his
functions are carried out by the President of the National Assembly.” Yet
Houphouët continuously amended this constitutional rule, creating
uncertainty about who the designated successor would be. In 1980, the
constitutional rule was changed so that the designated successor would be
the vice president. In practice, however, Houphouët kept the vice presi-
dent position vacant during this entire period, intentionally avoiding the
designation of a second in command. In 1987, the succession order was
changed to the president of the National Assembly and then the vice
president. In 1990, the constitution was amended once more, so that the
constitutional successor was the president of the National Assembly.
In addition, key cabinet ministerial positions were routinely left vacant
under the Houphouët administration. From 1960 to 1981, Houphouët’s
cabinet did not have a clear “second in command” position. During these
two decades, there was not a vice president or prime minister in the
presidential cabinet. In 1982, the position of vice president was created
but it was left vacant every year from 1982 to 1985. In 1986, the position of
vice president was eliminated from the cabinet, which once again lacked
a “second in command” position from 1986 to 1990. In short, for the vast
majority of Houphouët’s rule he never designated an informal successor,
nor did he appoint a prime minister or vice president. From 1960 to 1990,
this position was either eliminated from the cabinet or it sat empty. It wasn’t
until 1991 – two years before his death, that Houphouët created a prime
minister position. Moreover, the minister of defense position was also kept
vacant a third of the time Houphouët was in office.
79
This case study will demonstrate how Houphouët entered power in

a position of extreme strength – he had already completely consolidated
political authority within Côte d’Ivoire even before entering office. After
coming into power, Houphouët systematically shut out Ivorian elites
from positions of power, preferring to rely on French technocrats or by
appointing less influential members of other ethnic groups to cabinet
positions. In the language of the formal model from Chapter 2,
Houphouët entered office with an extremely favorable draw of the dis-
tribution of power – no other elite possessed the ability to depose him. As
a result, Houphouët did not need to institutionalize power-sharing
arrangements to remain in power. He remained in power for over three
decades as a personalist leader. The collapse of the regime after the death
of Houphouët and the two civil wars that followed in the 2000s is
perhaps surprising, given that Côte d’Ivoire was considered the economic
miracle of West Africa in the 1980s. Yet without institutionalized
mechanisms for power-sharing, personalist rule could not sustain the
regime in the long run.

Côte d’Ivoire, also known as the Ivory Coast, is a country located in West
Africa, bordering the Atlantic Ocean on its southern coast. It currently has
a population of 23.7 million and about 55 percent of the population lives in
urban areas (World Bank 2017). Côte d’Ivoire shares a border with Guinea,
Liberia, Burkina Faso, Mali, and Ghana. The capital of the country is
Yamoussoukro, though it is only the sixth largest city. The port city of
Abidjan, located on the Atlantic coastline, is the economic center of Côte
d’Ivoire. It is the largest city in Côte d’Ivoire, with a population size of
almost five million, and is the fourth largest city in Africa.
Though it is not as ethnically heterogeneous as Cameroon, Côte
d’Ivoire is home to at least sixty different ethnic groups and dialects.
The Akan is the single largest ethnic group in the country, comprising of
roughly 37 percent of the population. Within the Akan the Baoulé is the
largest subgroup, and this is the tribe that Houphouët belonged to. There
is not one dominant religion in Côte d’Ivoire, and Muslims and Christians
each make up about a third of the population.
Côte d’Ivoire has the second largest economy in West Africa and is
currently the fourth largest exporter of goods within sub-Saharan Africa.
The economy is heavily dependent on agricultural exports, and almost
70 percent of the population works in agriculture. Since independence, its
main cash crops have been coffee, cocoa beans, and palm oil. Côte d’Ivoire
experienced extraordinary growth and prosperity in the first few decades
80
of independence. Its GDP grew by as much as 360 percent in the 1970s

and it was considered the “economic miracle” of West Africa (Daddieh
2001). The current GDP in Côte d’Ivoire is US$40.39 billion, and it is the
world’s largest exporter of cocoa beans (World Bank 2017).
Côte d’Ivoire gained independence in 1960 after decades of French
colonial rule. Prior to colonization, the territory was organized into several
states and kingdoms. European contact was established in the late 1800s.
Côte d’Ivoire was designated a French protectorate in 1843, becoming
a French colony in 1893. In response to growing agitations for indepen-
dence, France began to grant increased measures of self-governance for its
colonies, including representation in the French Constituent Assembly, of
which Houphouët was a representative. He would serve as one of the most
influential African representatives for fourteen years leading up to indepen-
dence. Côte d’Ivoire achieved full independence on August 7, 1960, with
Houphouët as the country’s first president.
3.3.2 Houphouët’s Rise to Power: Initial Leader Strength

Political rule under Houphouët is perhaps best summarized by the open-
ing sentence of Aristide Zolberg’s (1969) definitive volume on politics in
Côte d’Ivoire: “Political life in the Ivory Coast has been centered in recent
years around the activities of one man: Félix Houphouët-Boigny . . . By
1957, three years before the country became independent, [he] had gained
a virtual monopoly over access to public office at all levels” (3). As
Jackson and Rosberg (1982, 145), observed, the government of Côte
d’Ivoire is “the government of virtually one man” and no other politicians
achieved similar political stature under Houphouët’s rule.
Houphouët was a major player in Ivorian politics from the very begin-
ning. He was born into a family of hereditary chiefs of the Baoulé group
and became chief of the Akoué tribe through hereditary succession at
a very young age. Beyond his family’s high status, Houphouët further
elevated his prominence upon his first marriage. His first wife was of royal
Agni lineage through her mother and was Senegalese through her father.
Through this marriage, Houphouët acquired kinship ties and alliances
with other ethnic groups in Côte d’Ivoire as well as Senegalese elites
(McGovern 2011, 151; Zolberg 1969, 73).
Houphouët attended the Ecole William Ponty, an elite college in
Senegal that was attended by many young Africans who would eventually
be important independence politicians in their countries (other students
included Modibo Keita of Mali, Hamani Diori of Niger, Mamadou Dia of
Senegal, and Maurice Yaméogo of Burkina Faso). He then attended the
Ecole de medicine de I’AOF (French West Africa School of Medicine) in
81
Senegal, becoming a medical aide by 1925 – one of the first Baoulé to

achieve this status. Houphouët served in colonial medical service in var-
ious parts of Côte d’Ivoire until 1940. During this time, upon seeing their
mistreatment by the colonial administration, he became interested in
helping to organize cocoa and coffee farmers (McGovern 2011).
When his uncle died in 1940, Houphouët inherited large amounts of
land in his village and was appointed the chief of his Akoué tribe. From
then on, Houphouët would go on to begin his political career as one of the
richest African farmers in the entire country. His economic status served
as an important symbol of a growing “Ivory Coast bourgeoisie” and it
also provided him with the ability to finance his political campaigns
(Zolberg 1969, 73–74).
In 1944, Houphouët along with other African elites, established the
Syndicat Agricole Africain (SAA), an African agricultural union within
Côte d’Ivoire. Importantly, the SAA was the first quasi-political party that
agitated for better conditions and some measures of self-governance for
Africans within Côte d’Ivoire. Its goals were to secure premiums for
African producers, to obtain a quota of imported cloth and agricultural
commodities, etc. The eight elites who founded the SAA, including
Houphouët, were all traditional leaders who were leading planters and
literate in French. By the end of 1944, membership to the SAA had grown
to 8,548 and nearly half of the members were Baoulé. Out of the founders,
Houphouët was the only Baoulé, and he was elected president of the
organization when the SAA held its first congress in Abidjan in
September 1944 (Zolberg 1969, 67).
There are two main reasons why Houphouët was able to consolidate
power early and begin his presidency in an extremely strong position. In the
years between 1944 and independence in 1960, two major developments
cemented Houphouët’s stature as a leading politician and national hero in
Côte d’Ivoire. The first was his election to the French Constituent Assembly
and subsequent creation of his party, the Parti Democratique de la Côte
d’Ivoire (PDCI), as well as a pan-African coalition that would become the
basis of the independence movement within French West Africa. Though in
reality, Houphouët maintained close ties with Paris even after
independence,6 his leadership in these independence organizations solidi-
fied his image as a freedom fighter and founding father to an independent
6
There is a famous story that on the eve of independence, Houphouët met with
Kwame Nkrumah, the founding president of Ghana. Nkrumah favored a policy of
completely breaking ties with the British, while Houphouët favored a closer relation-
ship with the French. The two men agreed to meet again a decade later to see which
approach worked better. Nkrumah was deposed in a military coup six years after
independence (Woronoff 1972).
82
Côte d’Ivoire. The second major development was that of a bill in the
French Assembly to abolish forced labor in West Africa. Houphouët played
a major role in shepherding this bill through the legislative process. This
achievement won Houphouët the national spotlight as the Ivorian politician
who stood up to the French. From that point on, his moral and political
authority within Côte d’Ivoire would be unparalleled. I will discuss both of
these major developments in detail below.
In 1946, a law was passed that granted French citizenship to all people
living in French colonial territories. Citizenship came with the ability to
vote and participate in French elections.7 The ability to elect representa-
tives for the French Assembly provided Côte d’Ivoire limited measures
of political representation and a seat at the table. It also provided
Houphouët, who would serve for fourteen years in the French National
Assembly, with the opportunity to consolidate power as the top politician
in Côte d’Ivoire.
That same year an election was held to select a politician who would
represent Côte d’Ivoire in Paris. This was the first real opportunity that
most Africans had to express their political concerns and preferences
within the French Constituent Assembly. For these elections, European
and African politicians formed candidate lists that were composed of both
African and French representations. By contrast, Houphouët formed a list
that consisted of a coalition of Côte d’Ivoire tribes and foreign Africans,
one of the few proposals to exclude whites altogether. As a result, many
other African candidates withdrew their lists in support of Houphouët’s
slate. Houphouët’s list easily won both rounds of voting, and he was
elected to the French National Assembly (Zolberg 1969, 69).
A year later, building on the foundations of the SAA, Houphouët
formed his own political party, the PDCI. During the 1946 French
Constituent Assembly elections, Houphouët ran under his new party
label, and swept the elections, winning 98 percent of the vote in Côte
d’Ivoire districts (Zolberg 1969, 76). When Houphouët won, the PDCI
became the first successful independent African political party. Within
a year, the PDCI had an estimated 80,000 members, and from then on, the
PDCI gained so much influence that no African candidate could be elected
without the endorsement of the party, and by extension, Houphouët. The
French colonial administration often tried to undermine the PDCI. Prior
to elections, they promoted anti-PDCI candidates and tried to organize
ethnic and regional associations alliances that could challenge the PDCI’s
7
Initially, however, suffrage was far from universal. Many restrictions were placed,
for instance, French literacy was a requirement for voting. In 1956, the Loi Cadre bill
was passed, guaranteeing universal suffrage for all French citizens.
83
monopoly over political power. All efforts failed. Houphouët continued

to dominate elections, even as other PDCI candidates lost.
Houphouët’s continued presence in the French Constituent Assembly
had an important consequence. Representatives of the French Assembly
enjoyed a special privilege available only to its members: they were
immune from prosecution and could not be put in jail. After failing to
fully discredit the PDCI, the French administration began to use coercive
strategies to eliminate the influence of African elites by prosecuting and
jailing politicians once they lost an election (therefore losing their parlia-
mentary immunity privilege). Yet Houphouët continued to win every
election and therefore was immune from persecution. On the eve of
independence, he stood alone at the top of Ivorian politics (Gyimah-
Boadi and Daddieh 1999).
The second most important development in Houphouët’s ascension to
power was his successful proposal to outlaw forced labor within the colonies
in French West Africa. Forced labor was established by the French in Côte
d’Ivoire in the 1920s. Côte d’Ivoire was one of the few countries in West
Africa with a significant settler population. French settlers owned about
a third of cocoa and coffee plantations in the country. In reaction to increased
migration of African farmers from Côte d’Ivoire to Ghana, the colonial
government established the practice of forced labor in Côte d’Ivoire.
Although those who were “recruited” in this fashion “were paid by their
employers, wages were extremely low, and the government strictly super-
vised the performance of the contract they were forced to accept” (Zolberg
1969, 56). This practice was especially detested by Ivorians, who viewed the
system as one that worked to the sole advantage of the Europeans.
On March 1, 1946, a bill was proposed by Houphouët calling for the
abolition of all forms of forced labor in overseas France. The French
viewed this as a minor item – they did not even hold a floor debate to
discuss the bill, and it was passed. To Africans, however, this was one of
the most significant achievements of Africa’s emancipation from France
(Woronoff 1972, 36–37). To people within Côte d’Ivoire, the bill not only
removed a hated symbol of colonial rule throughout all of West French
Africa. It was also the achievement of an Ivorian – therefore elevating both
Houphouët and his independent movement across the region.
“Overnight, Félix Houphouët-Boigny became a mythical hero who had
imposed his will upon the French” (Zolberg 1969, 74).
For many years, people throughout the country – often encouraged by
the PDCI – firmly believed that unless Houphouët remained in office, forced
labor would be reinstated. The gratitude he earned from his countrymen
“has remained a foremost element in his political power and it has prevailed
over the hesitations of many followers who questioned his later policies”
84
(Zolberg 1969, 74–75). To provide a sense of what this bill meant to

Africans, a member of the Legislative Assembly explained:
It was very simple. The peasants of my region suffered a great deal from forced labor.
In my own family, there were twenty-four men and boys, but most of the time there
were few around to work for the family. The others were out slaving for the comman-
dant. When politics began, President Houphouët said he would do everything to
eliminate this barbarous practice, I decided that this was for me, and I followed him.
We didn’t know whether he could do it, but he kept his word. Since then, I have
followed him blindly and I shall continue to do so as long as he remains faithful to this
principle. (Interview with El Hadj Moussa Kone, 1959, qtd in Zolberg 1969, 75)
After years of incremental representation in the French National

Assembly, Côte d’Ivoire finally gained independence on October 31,
1960. The country’s first constitution was drafted by Houphouët and
a small circle of elites, and it was approved by the National Assembly
unanimously without debate (Zolberg 1969, 252). Presidential and legis-
lative elections were held, and Houphouët was the only presidential
candidate on the ballot. On November 27, 1960, he became the first
president of Côte d’Ivoire.
3.3.3 Houphouët In Power: Ruling Without Constraints

Upon taking power, Houphouët ruled Côte d’Ivoire as a one-man
dictator. Zolberg (1969), for instance, notes that the independence con-
stitution was “framed by a government and an assembly dominated by
a single party, itself controlled by one man” (25). Houphouët’s strategies
of personalist rule encompassed several important factors. For one, he was
charismatic. Handloff (1991) discusses the “myth of Houphouëtism,”
where the leader portrayed himself as the paternalistic founding father
of the country. Houphouët transformed himself into a “transcendent
symbol of unity to the disparate groups in Côte d’Ivoire, and his charis-
matic authority supplanted the traditional authority of the local chiefs”
(143). In fact, as McGovern (2011) notes, “the cult of personality that
grew around Houphouët was so great that he was able to move the capital
of Côte d’Ivoire to his home village of Yamoussoukro in 1983” (16).
Houphouët also benefited from strong economic growth in the first two
decades following independence. From 1960 to 1980 Côte d’Ivoire
experienced one of the highest rates of economic growth among the nonoil
exporting African countries. Houphouët resisted imposing high taxes on
cash crops to fund urban development projects, in part because his own
group, the Baoulé, dominated cocoa and coffee farming (Arriola 2013).
Moreover, he maintained financial ties with France, relying on French
investment, technology, and funding to develop the country’s economy
85
and administrative infrastructure (Handloff 1991, 144). The “Ivorian

Miracle” resulted in the development of impressive roadway and highway
systems and Abidjan, the main economic hub, became known as the
“Paris of West Africa” (Rabinowitz 2018, 74).
Most importantly, Houphouët did not empower other elites through
institutionalized power-sharing mechanisms, and instead, retained his sta-
tus as the single most powerful politician in the country. The first part of
this strategy was the use of French bureaucrats to run the state, rather than
appointing Africans. There were many educational requirements to work
within the bureaucracy, and Houphouët maintained these restrictions.
Since Africans had limited opportunities to obtain a university degree,
they were shut out of top positions. The highest ranks of the state remained
in the hands of European technocrats (Zolberg 1969, 100–101). As
Jackson and Rosberg (1982) note, “the French managerial elite in particu-
lar has provided loyal and capable agents who have contributed greatly to
the overall effectiveness of both the state and the private sectors of the
economy and enabled Houphouët to monopolize political power” (146).
Furthermore, Houphouët shut out the most influential members of
his own ruling coalition from the cabinet. In making government
appointments, Houphouët maintained a relative balance of ethnic
representation.8 He used an informal ethnic quota system to fill important
party, government, and administrative positions (Zolberg 1969, 141). Yet
within his presidential cabinet, “Houphouët has been careful to insure
that all major ethnic groups are represented – but not by their most
prominent politicians” (Jackson and Rosberg 1982, 148 emphasis
added). In this way, Houphouët was able to maintain support of various
groups through ethnic balancing, while at the same time, preventing the
strongest elites from accessing state power. “The cabinet is less a collegial
body of powerful and independent incumbents and more a technical
advisory body to the ruler. In recent years there has been an increase of
young technocrats both in the cabinet and in the bureaucracy. Of the
thirty-six ministers in 1974, only six were Houphouët’s old political party
colleagues” (Jackson and Rosberg 1982, 147–148).
In addition, the Ivorian constitution was extremely weak under
Houphouët. Presidential power was expanded by combining it with the
functions of prime minister while subordinating the role of the National
Assembly (Handloff 1991, 147–148). Côte d’Ivoire was also the first of
the former French colonies to abandon the parliamentary system and
adopt the presidential system. The chief executive, who is both head of
8
Though he still favored his own Baoulé group, especially for top positions.
86
state and head of government, is given the power to appoint a cabinet that
is not accountable to the National Assembly (Zolberg 1969, 108).
The ruling party and legislature under Houphouët was similarly weak.
The PDCI, which had been a political machine during the independence
era, began to atrophy once Houphouët came into power. Though the PDCI
routinely won all seats in the National Assembly, this was purely because
opposition parties were shut out. By the mid-1960s, party structures fell
prey to factionalism and were “inadequate instruments of control over the
administration and over the population” and “did not provide an effective
network of political communication” (Zolberg 1971, 21). Similarly, under
Houphouët the National Assembly served only to rubber stamp executive
decrees and instructions (Jackson and Rosberg 1982, 147–150).
3.3.4 Weak Institutions, Regime Collapse: The Aftermath

of Houphouët’s Death
Beginning in the 1980s, due to a sharp decrease in world cocoa prices,
Côte d’Ivoire experienced a severe economic crisis. To make matters
worse, Houphouët refused to export Ivorian cocoa from 1987 to 1989,
in hopes of driving world prices up, but this strategy failed, severely
weakening the economy of Côte d’Ivoire (McGovern 2011, 145). The
country’s debt increased, and crime rose dramatically in Abidjan due to
unemployment. In 1990, hundreds of civil servants went on strike, forcing
Houphouët to allow multiparty elections (Gyimah-Boadi and Daddieh
1999). On October 28, 1990, Côte d’Ivoire held its first multiparty elec-
tions. The main opposition candidate was Laurent Gbagbo, the leader of
an opposition party, the Front Populaire Ivoirien (FPI). Houphouët won
the election, receiving 82 percent of the vote, and remained in office.
However, the question of presidential succession loomed large.
Houphouët, now 85 years old, had always avoided designating a clear
successor. Constitutional rules regarding the succession order had chan-
ged several times throughout the three decades of Houphouët’s rule. The
designated successor vacillated between the vice president (though this
office was always left vacant) and president of the National Assembly.
Furthermore, the nature of “one-man politics [raised] the question of
succession” (Jackson and Rosberg 1982, 1152). Due to the fact that
Houphouët ruled Côte d’Ivoire as a personalist dictatorship where
power was centralized around him, there was not an obvious politician
of similar stature who would clearly serve as the presidential successor.
After public speculation about the presidential succession for over
a decade, Houphouët was finally pressured into naming a successor,
designating Henri Konan Bédié for the job (Rabinowitz 2018, 201–202).
87
Henri Konan Bédié had served in the Ivorian government since inde-
pendence. He was Côte d’Ivoire’s ambassador to the United States and
Canada, then the minister of economy and finance. In 1980, Bédié was
elected to the National Assembly, where he was then subsequently elected
as the president of the National Assembly, the position he served from
1980 through to Houphouët’s death in 1993.
In 1987 the constitution was amended to specify that the president of
the National Assembly would succeed the president in the event of the
incumbent’s death or incapacitation. Despite the creation of this constitu-
tional rule, Houphouët did not publicly support Bédié as his designated
successor. The lack of clear and early support from Houphouët of his
designated heir resulted in Bédié’s lack of legitimacy as the presidential
successor.
In 1991, two years before Houphouët’s death, a power struggle over
the impending presidential succession emerged between Bédié and
Houphouët’s newly designated prime minister, Alassane Ouattara. The
position of prime minister was created in 1991, and it was the first time
a second in command position had been created and filled in the presi-
dential cabinet since independence in 1960. In response to the financial
crises that had developed in the 1980s, the International Monetary Fund
(IMF) pressured Houphouët to appoint Ouattara, a former IMF econo-
mist who had been educated in the United States, as prime minister. Two
factions, one supporting Bédié and the other supporting Ouattara
emerged during this period (Konate 2004; McGovern 2011).
Houphouët died on December 7, 1993, and immediately following his
death, Bédié announced on state television that he had assumed the pre-
sidency. However, having failed to consolidate power prior to Houphouët’s
death, Bédié entered power weak. On one hand, there is some evidence that
Bédié tried to institutionalize his regime. Important cabinet positions, such
as prime minister and defense minister, were always filled and were rela-
tively stable. However, Bédié shut out one key challenger: Ouattara. He
purged the cabinet of anyone who supported Ouattara and introduced
a new electoral law that was clearly aimed at preventing Ouattara from
running in future elections (McCauley 2017; Rabinowitz 2018).
On December 23, 1999, Bédié was ousted in a coup that was staged by
members of the military who were linked to Ouattara. While Bédié was
out of the capital celebrating Christmas with his family, a group of
soldiers took over the airport, government television station, and radio
station. Many of the officers who led the coup had personal connections
with Ouattara – one officer had been a member of Ouattara’s guard. The
armed forces did not step in to defend Bédié, and the regime ended due to
the coup (Rabinowitz 2018).
88
3.4 Power Sharing in Presidential Cabinets
The regime in Côte d’Ivoire fell due to the absence of institutionalized

power-sharing mechanisms. Since Houphouët entered office extremely
powerful and influential, he did not need to establish institutional
constraints in order to share power with other elite opponents: viable
challengers simply did not exist. Although Houphouët was able to
remain in power without institutionalizing, the regime did not persist
long after his death. Due to the lack of clear succession planning, Bédié
entered office weak and lacking legitimacy as Houphouët’s heir.
Although there is some evidence that Bédié tried to use ministerial
appointments to coopt elites, he shut out his prime opponent,
Ouattara, which later led to a coup.
3 . 4 e m p i r i ca l a n al y s i s o f p o w e r sh a r i n g
in pres idential cabinets
In this final section, I analyze data on ethnicities of cabinet ministers to
demonstrate that power sharing was more prevalent in the Ahidjo regime
compared with the Houphouët regime. While ethnic favoritism prevailed
under Houphouët, Ahidjo routinely appointed non coethnics to his cabinet,
even appointing leaders from other ethnic groups to top cabinet positions,
such as prime minister, minister of finance, and minister of defense.
I use data from Francois, Rainer, and Trebbi (2015), which includes
comprehensive data on the ethnicities of cabinet ministers in fourteen
African countries from 1960 to 2010, including Cameroon and Côte
d’Ivoire. The descriptive statistics paint a clear picture of ethnic favoritism
in Côte d’Ivoire in contrast to elite power sharing in Cameroon.
Ahmadou Ahidjo, the first president of Cameroon, was an ethnic
Fulani, but he appointed many non coethnics to his cabinet. On average,
Ahidjo had twenty-three ministers in his presidential cabinet. The smallest
cabinets, from 1963 to 1964, had thirteen ministers, and the largest
cabinet, in 1982, had thirty-three ministers. There are 496 total minis-
ter–year observations (excluding the president) during Ahidjo’s rule from
1960 through 1982. Summary statistics are presented in Appendix Table
3.1.
Ahidjo did not favor ethnic Fulanis in his cabinet, only assigning
12 percent of all cabinet ministerial positions to his coethnics. The same
pattern emerges, even if we restrict the sample to top ministerial posts:
only 24 percent of top ministerial posts were assigned to ethnic Fulanis.9
9
Top ministerial posts, as defined by Francois, Rainer, and Trebbi (2015), include vice
president, prime minister, defense, budget, commerce, finance, treasury, economy,
agriculture, justice, and state/foreign affairs.
89
Moreover, Ahidjo offered cabinet positions to elites from opposing

groups. The Bamileke and Bassa groups were the leaders and supporters
of the UPC party, which was the main challenger to Ahidjo’s administra-
tion. Yet, 33 percent of total cabinet portfolios and 24 percent of top
ministerial positions were assigned to Bamileke or Bassa elites.
By contrast, Félix Houphouët-Boigny, the first president of Côte
d’Ivoire, systematically favored his coethnic Baoulés in his cabinet appoint-
ments. On average, Houphouët had twenty-five ministers in his presidential
cabinet. The smallest cabinet, in 1964, had ten ministers, and the largest
cabinet, in 1987, had forty ministers. There are 862 total minister–year
observations (excluding the president) during Houphouët’s rule from 1960
through 1993.
In contrast to Ahidjo, Houphouët did favor ethnic Baoulés in his
cabinet, assigning 26 percent of all cabinet ministerial positions to his
coethnics. Tellingly, 50 percent of all top ministerial posts were assigned
to Houphouët’s coethnics.
Figure 3.2 graphs the percentage of coethnics each leader appointed to
top ministerial posts throughout their tenure. The graph shows that
Ahidjo often shared power with important regional elites by appointing
leaders of other ethnic groups to top cabinet positions. As a result,
the share of top cabinet positions that went to elites from his own group
were often quite low. By comparison, Houphouët did not use cabinet
appointments as a power-sharing mechanism, and appointed elites from
his own ethnic group at a much higher rate.
A possible counterargument to this analysis is that leaders in Africa
simply assign cabinet portfolios in accordance with the population share
of various ethnic groups (Francois, Rainer, and Trebbi 2015). Since Fulanis
(Adhijo’s group) consists of 9 percent of the population in Cameroon, we
should expect Ahidjo to assign a smaller share of cabinet positions to Fulani
elites. On the other hand, the Baoulé (Houphouët’s group) consists of
21 percent of the population in Côte d’Ivoire. One can make the counter-
argument that Houphouët distributed many cabinet ministries to elites of
his own group simply because he was just making cabinet appointments
that were consistent with the population size of the group.
Yet this argument does not hold when we examine changes to cabinet
appointments over time. The population sizes of ethnic groups in Cameroon
and Côte d’Ivoire generally do not change over time (Vogt et al. 2015). Yet,
we do see differences in the distribution of cabinet positions to coethnic elites.
In the 1980s, Côte d’Ivoire experienced a severe economic crisis.
Unemployment and crime rose dramatically, and Houphouët’s government
was blamed for these conditions. From Figure 3.1 we can see that there was
90
3.5 Conclusion
Percent coethnics appointed to top ministerial post
.8
.6
.4
.2
0
1960 1965 1970 1975 1980 1985 1990 1995
year
Cameroon Côte d’Ivoire
Figure 3.2 Ethnic favoritism in president cabinets

Note: The graphs portray the percent of top cabinet positions that were assigned to elites
from the same ethnic group as the president.
a sharp drop in coethnic appointments in Houphouët’s cabinet, going into

the 1980s, the period where Houphouët was significantly weaker.
The proportion of coethnic cabinet appointments under Ahidjo also
fluctuated with time. Figure 3.1 shows that when Ahidjo initially came
into office he made virtually no coethnic cabinet appointments. As this
chapter discusses, this was because Ahidjo was especially weak when he
first became president upon independence. Over time, as Ahidjo consoli-
dated his authority, he began to make more coethnic appointments. Yet, it
is important to note that despite these changes over time for each leader,
ethnic favoritism was always higher in Côte d’Ivoire under Houphouët
compared with in Cameroon under Ahidjo.
3.5 conclusion
This chapter presented two case studies that illustrated the mechanisms and
findings articulated by the formal model in Chapter 2. First, I discussed the
case of Cameroon, a highly institutionalized regime under the presidency of
Ahmadou Ahidjo. Ahidjo entered power weak relative to other elites,
largely because he inherited his position of power from the outgoing
French colonial authorities. I argued that in order to retain support from
other elites, Ahidjo established checks on his own authority and delegated
91
power to other elites through the appointment of key cabinet positions.

Importantly, he designated a prime minister, Paul Biya, who was the
constitutional successor to the president. The leadership transition from
Ahidjo to Biya occurred peacefully and Biya remains in power today.
I then presented the case of Côte d’Ivoire, a weakly institutionalized
regime under the presidency of Félix Houphouët-Boigny. Houphouët,
who had been a renewed independence leader in the decades prior to his
presidency, entered power extremely strong relative to other elites. As
a result, he did not have to make institutional concessions to other elites in
order to remain in power simply because there were no other elite chal-
lengers of his stature. During most of Houphouët’s tenure, key cabinet
positions were left vacant and a constitutional successor was not estab-
lished. Upon his death, Houphouët’s successor (who had essentially been
designated at the last moment) took office but was soon deposed by the
military in a coup. Though Houphouët was able to remain in power as
a strongman leader, the regime fell soon after his death.
The next chapter will present data on regime institutionalization in all
African dictatorships from 1960 to 2010. While this chapter presented
more fine-grained evidence of the theoretical argument and mechanisms,
we will now take a step back and examine systematic evidence of the
relationship between initial leader strength and regime institutionalization.
Appendix
Ahidjo Houphouët
(Cameroon) (Côte d’Ivoire)
Mean number of cabinet ministers 23 25

Total number of minister–year observations 496 862
Number of minister–year observations with 59 220
coethnic appointee
appointee
Total number of minister–year observations for 154 295
top posts
Number of minister-year observations with 37 118
coethnic appointee for top posts
appointee for top posts
92
4
How Should Institutionalization Be Measured?
One of the core arguments of this book is that autocratic regimes vary
drastically in the extent to which they are institutionalized. This chapter
lays the empirical groundwork for this claim by presenting detailed data on
regime institutionalization. In doing so, I also provide a discussion of how
we should define and measure autocratic regime institutionalization – a key
concept often used, but not well operationalized – in the literature.
I define regime institutionalization as the creation of rules and proce-
dures that structure the distribution of power and resources within the
ruling coalition. In an autocratic setting, regime institutionalization serves
to tie the leader’s hands. This chapter presents an original time-series
cross-sectional dataset measuring regime institutionalization in forty-six
countries in sub-Saharan Africa from 1960 to 2010. I discuss my coding
strategy in detail and present examples of each type of executive
constraint. I also summarize general trends and patterns in the data to
provide a broad perspective of what institutionalization looks like within
sub-Saharan Africa and how it differs across countries.
This chapter also compares this dataset against other commonly used
datasets of authoritarian institutions, such as regime typologies (Geddes,
Wright, and Frantz 2014). I show that regime typologies often do not reflect
the extent to which executive constraints are in place, and many countries
coded as party-based regimes are not very institutionalized at all. In addi-
tion, I also compare my measures of institutionalization against other
commonly used datasets of institutional strength – Polity (Marshall,
Gurr, and Jaggers 2016) and V-Dem (Coppedge et al. 2016).
Another important implication that emerges from this chapter is that
countries in sub-Saharan Africa vary widely in their level of institutiona-
lization, even during the decades of authoritarian rule following indepen-
dence. A small number of recent studies document that African countries
have become more institutionalized over time, especially after the end of
the Cold War (Posner and Young 2007). Others show that African party
93
4 How Should Institutionalization Be Measured?
systems in the multiparty era are not as uniformly un-institutionalized as

previous scholars have argued (Riedl 2014). Yet scholarly consensus
about the quality and presence of executive constraints in African auto-
cracies, especially during the decades immediately following indepen-
dence, has not changed. Scholars routinely write off institutions from
that period, and politics prior to the 1990s is believed to have been driven
primarily by “big man rule” and personalist dictatorships (Bienen 1978;
Decalo 1976; Jackson and Rosberg 1982; Zolberg 1966). By contrast, the
data presented here tells a different story. African politics was not uni-
formly un-institutionalized during the Cold War era – some incumbents
did indeed regulate power through formal or informal institutions. These
mechanisms help to explain why some dictatorships have lasted for
decades, even surviving the introduction of multiparty elections.
4 . 1 co n c e p t u a li z i n g au t h o r i ta r i a n r e g i m e
i n s t it u t i o n a l iz a t i o n
Before I define regime institutionalization it is first helpful to note what I
mean by “regime.” I consider a regime as the set of rules, norms, and
institutions that govern the way in which the government is run. Most
relevant for this study, these rules determine the way in which leaders
enter and exit office. A regime can have multiple leaders, as long as these
leaders enter and exit office according to established rules and norms. I
consider an irregular leadership turnover – often due to the forced exit of a
leader and/or the entry of a leader via a coup – as signaling a break in the
regime (Goemans, Gleditsch, and Chiozza 2009).1
For instance, the Gambia experienced a regime change in 1994 when
the armed forces staged a coup and deposed Gawda Jawara – the leader
who had been in power since 1962. Ghana also experienced a regime
change in 1993 when it became classified as a democracy (more on how I
define democracy later) – though the leader did not change. Flight lieute-
nant Jerry Rawlings took power via a coup in Ghana in 1979. In the early
1990s, in response to increasing pressures to democratize, Rawlings lega-
lized political parties and allowed for multiparty elections to occur in
1992. Although Rawlings won the elections, and remained the president
of Ghana, these elections were considered free and fair, and importantly -
there has been party and leadership turnover since then. Ghana is there-
fore coded as having a regime break in 1993. Finally, Tanzania has
remained the same regime since independence, despite having multiple
1
This definition of an autocratic regime is similar to Svolik’s (2012) definition of a
“ruling-coalition spell.”
94
4.1 Conceptualizing Regime Institutionalization
leaders. Since independence was granted in 1961, Tanzania has had five
different presidents, but the same ruling coalition has remained in power,
and the ways in which the incumbent is selected has not drastically
changed.2
I define autocratic regime institutionalization as the creation of rules,
procedures, and hierarchies that structure the ways in which power and
resources are distributed within the ruling coalition. Examples of this
include establishing leadership succession procedures or creating rules
that govern elite promotion. When a leader institutionalizes the regime,
she takes actions that constrain her ability to take arbitrary actions in the
future. Autocratic regime institutionalization occurs when executive
constraints are created that tie the leader’s hands.
My conceptualization of institutionalization closely follows Huntington’s
(1968) definition of political institutionalization as “the process by
which organizations and procedures acquire value and stability” (12).
In particular, Huntington identifies several key features of institutiona-
lization that are most relevant for the survival of authoritarian regimes.
First, adaptability and the ability to outlive the founder is a key char-
acteristic of institutionalized organizations (13). Second, autonomy and
the ability of an organization to exist independently of particular actors
is a key feature of institutionalization. According to Huntington, an
organization that is merely an instrument of a leader lacks autonomy
and institutionalization (20).
This discussion is also related to Levitsky and Murillo’s (2009) discus-
sion of institutional strength.3 They argue that institutional strength
should be judged across two dimensions: enforcement and stability.
Strong institutions are those that political elites comply with in practice.
This conceptualization of institutional strength ties directly with my dis-
cussion of regime institutionalization. Regimes with institutionalized
2
Note that this conceptual discussion of “regime” does not affect the coding decisions
for this project because the measures of institutionalization I construct operate at the
country–year level, not at the regime level.
3
Much of the discussions of institutionalization has centered around the institutiona-
lization of political parties, rather than regimes. Huntington (1965), for instance,
defines party institutionalization as the process by which parties become established
and acquire value and stability. Levitsky (1998) likens institutionalization to the
entrenchment of the “rules of the game” that shape the expectations, payoffs, and
behavior of actors in his discussion of the Peronist party in Argentina (p. 80).
Panebianco (1988) argues that institutionalization transfers authority from the
leader to the party, and very few charismatic parties survive this transfer. Research
on party system institutionalization also describes strong party organizations as
having independent status and value, not subordinated to the interests of ambitious
leaders (Mainwaring and Scully 1995).
95
practices – those governed by stable rules that do not shift with changing
circumstances – are durable regimes that are likely to survive leadership
transitions precisely because they have rules that can outlast the circum-
stances in which they were initially created.4
In the context of an authoritarian regime, institutionalization has
several important implications. First, the creation of rules and procedures
depersonalizes the ways in which the regime operates by constraining the
autocratic leader’s ability to make arbitrary decisions in the future. This is
especially relevant in dictatorships, where power is often concentrated in
the hands of a single leader. When a leader establishes term limits, for
instance, she loses the ability to decide arbitrarily when and whether she
will step down from office. This kind of depersonalization can also hap-
pen through the creation of merit-based promotion within the executive.
Regimes with set rules, procedures, and hierarchies that regulate elite
promotion are more institutionalized compared with systems in which
a single incumbent can appoint or purge elites at any time without
constraint.
Second, institutionalization can empower other elites by establishing
hierarchical positions and rules regarding the appointment of elites to
important positions. A regime with a presidential cabinet in which elites
are appointed to important ministerial portfolios, such as defense or
finance, represents a regime in which power and resources are not entirely
concentrated within the hands of the leader. Such a regime is more
institutionalized compared with a presidential cabinet in which the incum-
bent holds all the important portfolios herself, and there are no estab-
lished rules or norms for the appointment of these positions.
However, it is important to note that institutionalization does not
always result in the empowerment of other elites. Some types of executive
constraints only limit the incumbent. Term limits on their own create
restrictions on the length of time that the incumbent can remain in office,
but do not specify who will replace the incumbent or how that replace-
ment process will occur.
Third, institutionalization separates the ruler from the regime.
Importantly, an institutionalized regime is one that can perpetuate its
own existence beyond the influence of an individual leader. A regime
that is not institutionalized is a government of one: the ruler. The ability
4
My conceptualization differs from Slater’s (2003) view that highly institutionalized
regimes do not necessarily have to contain institutions that constrain leaders, as long
as the institutions have “infrastructural power” – the ability to implement and
execute the leader’s demands. By contrast, I argue that the ability to keep the
executive in check is a necessary and central component of institutionalization.
96
to survive beyond a particular leader is of particular importance in

authoritarian regimes. Elections in democracies serve as a reliable vehi-
cle in which future incumbents are selected without interrupting the
regime. The absence of such procedures in autocracy is often the greatest
weakness of dictatorships; how a regime can perpetuate itself beyond a
particular leader is not an automatically clear process. This challenge is
often exacerbated by the fact that many leaders give themselves the title
of “president for life” or claim to have godlike qualities, implying that
they are irreplaceable. Unsurprisingly, many leadership transitions
within authoritarian regimes are violent and irregular, and the majority
of regimes are unable to survive past the death or departure of the first
leader.
Institutionalized regimes resemble autonomous organizations, capable
of functioning regardless of which leader is in power. They tend to have
mechanisms in place to regulate leadership succession. Regimes with high
levels of institutionalization also have rules and autonomous organiza-
tional structures that ensure that the regime can function after the current
incumbent departs office. When we think about the quality of autocratic
regimes, the extent to which there are structures and procedures in place
to guard against personalist rule and perpetuate the survival of the ruling
coalition are of critical importance.
4.1.1 How Do Institutions Differ from Institutionalization?

While institutions are rules, institutionalization is the entrenchment of
those rules. Existing scholarship generally defines formal institutions as
rules that shape and constrain the behavior of political actors. In the first
sentence of his book, North (1990) defines institutions as “the rules of
the game in a society, or more formally, the humanly devised constraints
that shape human interaction” (3). According to North, “institutions
reduce uncertainty by providing a structure to everyday life” (3).
Following North’s definition, Lauth (2015) further notes that “[i]nstitu-
tions are a set of rules … The set of rules also creates and shapes social
order in such a way that the behavior of all actors involved in that social
order is predictable. Institutions affect performance by voluntarily
following the rules or being motivated by the threat of sanctions”
(Lauth 2015, 57).
However, as Huntington (1968) stresses, all institutions vary in their
degree of institutionalization. In their review of the literature on institu-
tional strength in comparative politics, Levitsky and Murillo (2009) stress
that institutional strength should be treated as a variable, rather than an
assumption, especially in the context of developing countries or
97
authoritarian states, where rule-making environments are less established.

They argue that “institutions are stable to the degree that they survive not
only the passage of time but also changes in the conditions – i.e., under-
lying power and preference distributions – under which they were initially
created and reproduced” (117). In terms of my conceptualization, institu-
tions are stable when they are institutionalized.
It is important to note here that while institutions are most commonly
defined as rules that constrain and shape behavior, another approach is to
think about institutions as equilibrium behavior. Knight (1992), for
example, treats self-enforcement as the defining criterion of institutions.
In other words, according to this approach, institutions are simply equili-
bria in which actors do not have any incentives to deviate – regardless of
whether the equilibrium is beneficial to the parties involved and regardless
of whether they are formally codified. I do not conceptualize institutiona-
lization in line with this school of thought. In fact, a regime that is, in
equilibrium, highly predatory but stable, is an example of an un-institu-
tionalized regime.5 This book focuses on institutions as rules and con-
ceptualizes regime institutionalization as the establishment of rules that
constrain the incumbent.
4.1.2 Can Informal Institutions Constrain?

Much of the literature on authoritarian institutions has focused either on
“parchment institutions” (Carey 2000) – formal rules that are written
down such as laws, regulations, constitutions – or on nominally demo-
cratic institutions such as parties, legislatures, or elections. But does
institutionalization have to occur on paper? Can informal rules and
norms constrain the behavior of executives? Helmke and Levitsky
(2004, 725) rightly emphasize that informal institutions have “remained
at the margins of the institutionalist turn in comparative politics,” even
though informal norms structure political behavior and are highly influ-
ential for political outcomes.
I follow Helmke and Levitsky’s definition of informal institutions as
“socially shared rules, usually unwritten, that are created, communicated,
and enforced outside of officially sanctioned channels” (727). This defini-
tion builds on North’s (1990) argument that norms can be considered
institutions in that they create rules that incentivize or constrain behavior.
They contrast this concept against formal institutions, which are defined
5
As a parallel example, a country might be stuck in an equilibrium where there are
high levels of corruption that operate in defiance of formal regulatory institutions,
but I would not consider corruption practices to be a set of institutions.
98
as “rules and procedures that are created communicated, and enforced

through channels widely accepted as official” (727).6
The rules of the game that structure political outcomes are often
informal – even in democracies – but this is especially the case in author-
itarian regimes where politics is often weakly institutionalized. A com-
monly cited example of a highly influential informal institution is the way
in which incumbents chose future presidents in twentieth-century Mexico.
Under the rule of the PRI, sitting presidents had the right to choose their
successors, and this practice was known as the dedazo (“big finger”).
From 1934 through 2000, this mechanism facilitated eleven regular lea-
dership turnovers (Langston 2006). Other common examples of informal
institutions include corruption and clientelism (see Hicken 2011 and
Treisman 2007 for surveys on these topics), although informal institutions
do not always produce negative outcomes. Tsai (2007), for instance,
illustrates how rural villages in China provide public goods where village
temple associations reinforce norms of social obligation.
When discussing informal institutions in developing and undemocratic
countries, we often think of clientelistic and patronage-based practices
that expand executive power. Presidents in Africa – especially in the
period immediately following independence – have frequently been
labeled patrimonial (or neo-patrimonial) leaders (Zolberg 1966).
Scholars agree that African politics is often driven by clientelistic practices
in which the president has complete authority over distributive decisions.
Leaders often use cabinet portfolios as patronage to buy elite support
(Arriola 2009), and they can even be used to coopt opposition politicians
(Arriola, DeVaro, and Meng 2018; Buckles and Gandhi 2017). Presidents
in many Latin American countries also have informal authority over state
institutions that lead to relatively unconstrained executives (O’Donnell
1994; Hartlyn 1994).
What often goes overlooked, however, is that informal institutions
can also be used to constrain executives. In fact, autocratic leaders are
often informally constrained in the extent to which they can monopolize
rents by their commitments to share resources with elites. This type of
informal constraint is especially effective when backed by credible
threats of deposition. I argue that regime institutionalization can indeed
be carried out via informal institutions. The primary function of institu-
tionalization in autocratic regimes is to constrain the executive, and the
mechanism of tying the leader’s hands can happen formally or
informally.
6
Formal rules include state institutions (parties, legislatures, and courts) as well as
state-enforced rules (constitutions and legislation).
99
In present day China, for instance, the appointment of the politburo

standing committee – the highest echelon of power within the ruling
Chinese Communist Party – has consistently followed three unwritten
norms. First, officials within the politburo and the politburo standing
committee retire by a set age limit. Officials are expected to retire by the
time they are sixty-eight, and generally do not receive appointments after
they turn sixty-seven. Minister-level officials are expected to retire at the
age of sixty-five, and lower-level officials are expected to retire at the age
of sixty. Second, officials who are appointed to the politburo standing
committee are routinely drawn from existing officials within the polit-
buro. Appointments follow a specific path to promotion. Third, among
eligible politburo members, appointments to the standing committee are
generally made on the basis of age (Miller 2017). Since appointment
are made according to these norms, the president faces limitations on
who he can appoint to the standing committee.
At the same time, as Thelen (1999a) cautions, norms and informal
institutions should “exert some independent power over individual beha-
vior” beyond simply reinforcing formal institutions (377). In Chapter 6,
I show that the appointment of stable vice presidents, prime ministers, and
defense ministers secures autocratic rule by lengthening incumbent
tenures and alleviating coup threats. Chapter 7 demonstrates that the
appointment of a de facto successor through the office of the vice
president/prime minister promotes peaceful leadership succession beyond
the existence of constitutional succession rules.
4.1.3 When Does Institutionalization Take Hold?

The reader may also be wondering when institutionalization actually
takes hold: when institutions are created or when they are tested? For
instance, if a leader implements leadership succession policies is the regime
institutionalized along this dimension when the constitutional rule is first
created? Or does the regime become institutionalized at the moment of the
leadership transition when the succession procedure follows the constitu-
tional rule?
I argue that institutionalization takes hold soon after the institutions
are created, since rules that credibly constrain leaders shift the distribution
of power in favor of elites.7 The longer these rules are kept in place, the
more time elites have to consolidate their authority by using the resources
7
Recall that in the formal model presented in Chapter 2, positive levels of institutio-
nalization increase the likelihood that the autocrat would be deposed if fighting were
to occur in the second period.
100
4.2 Measuring Regime Institutionalization
and influence afforded by positions of power. Moments where constitu-

tional rules are tested allow researchers to observe whether institutiona-
lization has actually taken place. As Chapter 6 will show, most instances
of leadership transitions do actually follow the procedures laid out in the
constitution (if such a rule existed). Chapters 6 and 7 will also discuss how
leaders are more likely to violate term limits, making such rules a weaker
form of regime institutionalization.
4.2 measuring regime institutionalization

Now that we’ve conceptualized regime institutionalization, how should
this variable be measured? Developing high-quality indicators of author-
itarian institutions poses some real challenges. Dictatorships are
frequently closed off, therefore restricting or completely eliminating
access to reliable and accurate information. Moreover, conventional
measures of institutional strength in democracies cannot simply be
imported to autocracies due to the lack of free and fair political competi-
tion. For example, while electoral results from presidential or legislative
elections can serve as a credible measure of incumbent or party strength
in democracies, the same approach cannot be imported to autocracies
because election results are often either falsified or do not reflect the true
preferences of citizens.
This section introduces my original dataset on regime institutionaliza-
tion in all sub-Saharan African countries, from 1960 to 2010. I present
measures of institutionalization that reflect the extent to which the regime
is autonomous from the leader. Since institutionalized regimes are those
that can outlive individual leaders, several of my measures focus on the
extent to which formal leadership succession policies exist. A regime with
clear and specific succession procedures represents a more institutiona-
lized organization because it has internal mechanisms to perpetuate itself
beyond the lifespan of a single leader. Another important dimension of
institutionalization is the extent to which the regime is able to function
independently of the leader. To gauge this dimension of institutionaliza-
tion, I present additional measures reflecting the extent to which other
elites fill key cabinet positions. Such cases where decision-making author-
ity is not concentrated entirely within the hands of the leader also repre-
sent regimes with higher levels of institutionalization. I collect historical
data on state constitutions and presidential cabinets, for which I have
comprehensive records for every country and year in my dataset. From
these records, I document the creation of executive constraints, hierarch-
ical positions, and implementation of rules and procedures that structure
the distribution of power.
101
4.2.1 Defining the Sample: What Qualifies as an

Authoritarian Regime?
Before I discuss my measurement strategy, it is useful to first define the
sample of countries that fall within the scope of this study. What qualifies
as an authoritarian regime? Many scholars have recently reevaluated this
question.8 There are two general camps. On one hand, the minimalist
definition of democracy requires political actors to acquire positions of
power through “competitive struggle for the people’s vote” (Shumpeter
1947, 269). Democracies are therefore defined through competitive
elections and regimes that fall short of this requirement are considered
autocracies. On the other hand, one can also consider a broader notion of
“democracy” by focusing not only on free and fair elections but also on
citizen access to impartial information, and the extent to which freedoms
(such as the freedom of speech or assembly) are upheld (Dahl 1972).
For my analysis in this book, I define an authoritarian regime as a country
in which executives do not come to power through competitive national
elections. For the coding of regimes, I follow Cheibub, Gandhi, and
Vreeland’s (2010) criteria. They define an autocracy as a country in which
any of the following criteria are violated: (1) the executive is selected either by
a popular vote or the election of a committee for the purposes of executive
selection (such as the Electoral College in the United States); (2) the legislative
is elected either directly or indirectly by popular election; (3) multiple poli-
tical parties are legally allowed to exist in the regime and the legislature; and
(4) there has been alternation in the party that is currently in power.
The last criterion, referred to as the “Botswana Rule,” is particularly
relevant when studying regimes within sub-Saharan Africa. As Przeworski et
al. (2000) and Cheibub, Gandhi, and Vreeland (2010) discuss, some regimes
may allow for multiparty elections only because they are confident that the
opposition will lose or because the opposition will not be allowed to take
office if they win. If a ruling party has never lost an election or has not
stepped down after losing an election, then it is difficult to ascertain whether
the regime is a true democracy or whether it is simply an autocracy that
allows for multiparty elections without the real intention of ceding power.
We can therefore only be certain that a regime will indeed allow for party
alternation in cases where the opposition is allowed to take office after
winning an election.
Within sub-Saharan Africa a number of countries led by ruling parties
have never experienced party alternation since the ruling party took power
after independence. Angola, for instance, has always been ruled by the
8
See Diamond 2002; Gandhi 2008; Geddes 2003; Pzeworski 2000; Svolik 2012.
102
Moviemento Popular de Libertacao de Angola (MPLA), despite holding

multiparty elections in 1992, 2008, and 2012. Although multiparty elections
were introduced in 1992, Tanzania has never had party alternation, and the
ruling Chama Cha Mapinduzi (CCM)9 has always won presidential elections
and legislative majorities. The Polity scores for Angola and Tanzania during
the period of multiparty elections remain low – ranging between −6 and −1.
In sum, simply having the appearance of multiparty elections is not sufficient
in guaranteeing that a country has democratized, especially when the incum-
bent and ruling party have never been tested by an electoral loss. Party
alternation is therefore a key criterion in defining regime type.
In addition to the criteria delineated by Cheibub, Gandhi and Vreeland
(2010) I further require that a country must be coded as a democracy for
over ten years in order to be considered a truly consolidated democracy.
There are a number of countries that have brief periods in which they are
coded as democracies by Cheibub, Gandhi, and Vreeland (2010), but then
revert back to authoritarianism. Ghana, for instance, is coded as a democ-
racy from 1969–1971 and 1979–1980, but was considered an autocracy
for long periods before and after those intervals. A truly consolidated
democracy should not be easily dismantled and requiring that countries
must be coded as democracies for over ten years eliminates countries that
only appear democratic for short periods of time.10
In sum, to derive my sample of authoritarian regimes I start with all
countries in sub-Saharan Africa. Countries enter the dataset either at the
year of independence or in 1960,11 and remain in the dataset either until
they democratize (following the criteria discussed above) or until 2010.
The final sample includes forty-six countries from 1960 to 2010.
Appendix Table 4.1 lists all the countries, leaders, and time periods that
qualify as authoritarian regimes and therefore are included in this study.
An alternative way of creating my sample of cases would have been to
utilize Polity scores (Marshall, Jaggers, and Gurr 2011) and drop cases that
fall below a certain numerical threshold. This procedure more closely reflects
the Dahl conceptualization of democracy in which the quality of political
9
The CCM was created in 1977 after the former ruling party, the Tanganyika
African National Union (TANU), merged with the ruling party of Zanzibar, follow-
ing the merger of the two countries.
10
Using this coding scheme, the following countries and time periods are considered
democratic and therefore dropped from the sample: Benin after 1991, Cape Verde
after 1990, Ghana after 1993, Kenya after 1998, Madagascar after 1993, Malawi
after 1994, Mali after 1992, Mauritius for the whole period, Nigeria after 1999, Sao
Tome and Principe after 1991, Senegal after 2000, and Sierra Leone after 1998.
11
Only Ghana and Guinea gained independence a few years prior to 1960. Liberia and
Ethiopia were never colonized. Each of these three countries enter the dataset in 1960.
103
competition and institutions are taken into account. However, I do not base
my coding of authoritarian regimes on Polity scores for three main reasons.
First, the vast majority of countries in Africa do not meet the minimalist
definition of contestation of power through competitive elections, at least
not until the 1990s. The “less stringent” criterion is sufficient to identify the
authoritarian sample of cases in this region. Second, Polity uses criteria
about the quality of executive constraints and competitiveness to create an
aggregated score.12 Since I am examining constraints on executive power, I
do not want to eliminate cases by using a coding scheme that takes into
account the quality of institutions and constraints on the executive. Doing
so would risk eliminating cases of highly institutionalized authoritarian
regimes in my final sample. Third, Polity scores do not reflect party alter-
nation; therefore, a case such as Botswana can score exceptionally high
despite not having had party alternation in the executive.
4.2.2 The Data

I define institutionalization as the creation of hierarchies, rules, and proce-
dures that structure the distribution of power and resources within the ruling
coalition. An important implication of institutionalization within authoritar-
ian regimes is that it functions to constrain the leader’s ability to make
arbitrary decisions. In order to operationalize this concept, I create measures
that focus on the extent to which the leader’s hands are tied. Such constraints
can take the form of formal constitutional rules that explicitly limit the leader’s
authority. In particular, I focus on the creation of constitutional leadership
succession procedures that dictate how a peaceful transition of power should
occur, thereby prolonging the tenure of the regime beyond a single leader.
I also focus on the establishment of clear hierarchical positions by
examining the extent to which the president is willing to distribute impor-
tant cabinet positions (such as vice president, prime minister, and defense
minister) to other elites. The creation of stable cabinet appointments limits
the incumbent’s power by placing other elites in a position of power or
influence within the presidential cabinet. It also reflects the process of
institutionalization by creating an independent hierarchy and decision-
making apparatus separate from the leader. In sum, these measures pro-
mote the creation of an independent functioning organization and also
serve to limit the authority of a single leader.
12
One of the component variables of Polity scores is “XCONST,” which reflects the
extent to which there are institutionalized constraints on the decision-making
powers of the executive. A comparison of my measures of executive constraints
and XCONST is discussed later in this chapter.
104
4.2.3 Formal Institutions: Autocratic Constitutions

To construct my measures of regime institutionalization, I utilize the
Europa World Year Book (1960–2010), which has yearly records of all
executive posts, ministerial positions, and constitutions for all countries in
sub-Saharan Africa. First, I examine the creation of leadership succession
rules within constitutions. Scholars have identified constitutions as useful
mechanisms that can establish formal rules and procedures in a public
way that holds autocrats accountable (Albertus and Menaldo 2012;
Ginsburg and Simpser 2013). In fact, autocratic constitutions are also a
type of “focal point” for elites, creating a universally observed set of rules
and enforcement behavior surrounding the incumbent.
For every country–year observation, I document whether the constitu-
tion had an amendment outlining procedures governing presidential
succession. The Kenyan constitution, which has been in effect since inde-
pendence in 1963, includes specific instructions for succession. Chapter II,
Part I, Section 6 reads:
(1) If the office of President becomes vacant by reason of the death or

resignation of the President … an election of a President shall be held
within the period of ninety days immediately following the occur-
rence of the vacancy.
(2) While the office of the President is vacant, the functions of that office
shall be exercised –
a. By the Vice President
(Peaslee and Peaslee Xydis 1974, 341)
The transfer of power from Jomo Kenyatta to Daniel arap Moi during
the 1978 presidential succession in Kenya illustrates the importance of
succession rules. Near the end of Kenyatta’s rule, a faction within the
ruling KANU party tried to contest the authority of then-vice president
Moi on the grounds that he was not a member of the dominant ethnic
group. Moi and his supporters were able to effectively dispute their claims
by utilizing the policy governing presidential succession outlined in the
constitution (Tamarkin, 1979, 21–26).
Beyond this particular case, constitutional succession rules play a signifi-
cant role in promoting successful leadership transitions. In the chapter 7,
I show that regimes with constitutions that outline specific succession pro-
cedures are significantly more likely to undergo peaceful leadership succes-
sion. Furthermore, once in place, succession rules are not easily overturned.
In my sample, conditional on creating a constitutional amendment governing
the rules of succession, 74 percent of these procedures have been left intact.
105
If succession rules do exist, I also distinguish whether the procedures

specify who would succeed the president, rather than simply providing
vague rules about nominating a successor. Unlike the Kenyan constitu-
tion, which specifies that the vice president should become the interim
president, the constitution of Angola provides guidelines that are much
vaguer. Article 33 of the constitution states: “In case of the death, resig-
nation, or permanent impediment of the President of the Republic, the
Council of the Revolution will designate from among its members
the person who will exercise temporarily the charge of President of the
Republic” (Blaustein and Flanz 1976, 6).
In addition to succession rules, I examine the creation of constitutional
term limits. Scholars have increasingly pointed to term limits as a key type
of executive constraint in authoritarian regimes, particularly in African
states after the end of the Cold War (Posner and Young 2007). According
to Svolik (2012), “a term limit on a leader’s tenure therefore amounts to a
line in the sand: Compliance is easily and publicly observable. Term limits
thus both embody a compromise about the limited authority of any single
leader and provide an unambiguous signal of commitment to such a
constraint” (198). Scholars and practitioners within sub-Saharan Africa
pay particular attention to the creation of term limits, arguing that “term
limits are a necessary bulwark against abuse of power, especially when
electoral systems are weak” (Louw-Vaudran 2016).
For every country–year observation, I record whether the constitution
includes term limits. The constitution of the Democratic Republic of
Congo, for instance, from 1978–1996 included a clause stating that the
President would be elected for a seven-year term, renewable only once.
Article 66 of the constitution of Ghana has included the following provi-
sion since 1993: “(1) A person elected as President shall … hold office for a
term of four years … (2) A person shall not be elected to hold office as
President of Ghana for more than two terms” (Peaslee and Peaslee Xydis
1974). Executives with formal term limits written into the constitution
represent leaders who are operating within more highly institutionalized
regimes.
4.2.4 Informal Institutions: Presidential Cabinets

In order to codify informal constraints on presidential power, I turn to the
appointment of key positions within the presidential cabinet. For every
country–year observation, I document the name of the president, the name
of the vice president, prime minister, and minister of defense if one had
been designated.
106
The appointment of a vice president (VP) or prime minister (PM)13 –

so-called “second-in-command” positions – represents the creation of a
hierarchical structure within the executive as well as a distribution of
authority to other elites. Designating a second-in-command, a very visible
national position, casts another elite as a potential focus of power and
possible successor to the executive. In fact, we can even think of the act of
naming a second-in-command as the autocrat solving the collective action
problem for other elites as an alternative leader to rally around.14 The
creation of these positions, therefore, reduces reliance on a single autocrat
and promotes the survival of the regime past the tenure of the first leader.
Moreover, within African presidential cabinets, most presidential
successors were former vice presidents or prime ministers. Under the
Cameroon Union (UC) in Cameroon, for instance, Paul Biya had been
the prime minister in 1975 under the presidency of Ahmadou Ahidjo
before becoming president himself in 1982. To verify that this was indeed
the larger pattern in my data, I code the previous positions of presidential
successors. The data reveals that the position of vice president or prime
minister is often a stepping stone for the presidency. Table 4.1 provides a
full summary of previous positions held by presidents in my sample,
conditional on a peaceful transfer of power. The data show that 23
percent of all former presidents held the position of vice president or
prime minister. Minister of defense, the state/interior, and finance are
also commonly held portfolios by elites who later became the incumbent.
For successions that were peaceful, the vast majority of successors were
chosen within the president’s cabinet.
I also analyze past positions of incumbents who came into power via
irregular means. Table 4.2 provides a full summary of previous positions
held by presidents in my sample, conditional on a nonpeaceful transfer of
power. Most of the leaders in this sample came into power via a coup.
Unsurprisingly, these leaders commonly had military backgrounds. Even
for the leaders in this sample who were in the presidential cabinet before
13
Vice presidents and prime ministers are functional equivalents within this particular
context. All the countries in my sample either have one or the other. No cabinet has
both a vice president and a prime minister, although a small number of countries
have multiple vice presidents.
14
In fact, many presidents were extremely hesitant to assign the second-in-command
position to another party elite. Jose Eduardo dos Santos, the first president of
Angola, named himself as his own vice president for a number of years while he
was in power. Felix Houphouet-Boigny, the first president of the Ivory Coast,
created a vice-president position, but kept the post empty the entire time he was in
power. These two cases (and other similar instances) are coded as not having the
second-in-command position filled by another elite.
107
Table 4.1 Previous positions of autocratic leaders (conditional on peaceful

leader entry)
Leader’s previous position Count Percent
Within the presidential cabinet

Vice president/ prime minister 25 23%
Minister of defense/security 15 14%
Minister of state/ interior 12 11%
Minister of finance 11 9%
Minister of planning 7 6%
Minister of foreign affairs 6 5%
Minister of natural resources 6 5%
Minister of development 5 5%
Minister of health 3 3%
Minister of education 3 3%
Miscellaneous portfolios 18 16%
Total (within cabinet) 111 100%
Outside of the presidential cabinet

Legislature 5 26%
Judicial system 5 26%
Military 5 26%
Opposition 3 16%
Unions 1 5%
Total (outside of cabinet) 19 100%
Notes: The leaders included in this table entered office through regular means, so coups and
irregular leader entries are excluded from this sample. The first president of each country is
not included in this sample because I am interested in the path of promotion after
independence was granted. Only leaders who were in office for three years or longer are
included in this sample. Leaders sometimes have multiple positions before entering office.
For example, Festus Mogae, who was the president of Botswana from 1998 to 2007 had
multiple cabinet positions from 1990 until 1997. He was the minister of finance from 1990
to 1997, minister of development and planning from 1990 to 1997, and vice president from
1992 to 1997. Each of these portfolios are recorded in this table.
their ascension to power, the most frequently held positions were the
minister of defense, vice president, or prime minister.
In addition, I document whether the person in the vice president or
prime minister position remains fairly constant over time, or whether the
person in this position is rotated very frequently. Accounting for cabinet
rotation rates is particularly important in the context of elite politics in
sub-Saharan Africa. African leaders routinely practiced the “revolving
108
Table 4.2 Previous positions of autocratic leaders (conditional on irregular

leader entry)

Military 41 93%
Opposition 3 7%

Minister of defense 8 44%
Minister of the interior 1 6%
Misc. minor portfolios 4 22%
Notes: The leaders included in this table entered office through irregular means – the modal
method of irregular leader entry is via a coup. The first president of each country is not included
in this sample because I am interested in the path of promotion after independence was granted.
Only leaders who were in office for three years or longer are included in this sample.
door” policy of constantly rotating important cabinet ministers in order to

prevent any one person from amassing too much power (Dickie and Rake
1973; Jackson and Rosberg 1982; Hassan 2017). Moreover, since vice
presidents are largely perceived to be de facto successors, presidents often
strategically rotate those in the second-in-command position to prevent
any particular elite from amassing too much power or influence. In order
to account for strategic rotation, I create a variable that measures the
stability of key appointments.
An example is helpful here. Consider Seretse Khama, who was the first
president of Botswana from 1966 to 1980. Quett Masire was appointed as
the vice president throughout Khama’s entire tenure, and in fact succeeded
Khama to become the next president. In this case, Masire had a stable vice
president for fourteen years during his tenure. On the other hand, Felix
Houphouet-Boigny, the first president of Cote d’Ivoire from 1960 to 1991,
did not appoint a vice president or prime minister the entire time he was in
office. Finally, Idriss Deby, who has been the president of Chad since 1991,
has been rotating his vice presidents practically on a yearly basis since
taking power. From 1991 to 2005, Deby named ten different vice
presidents. There have only been five years for which the vice presidency
was held by the same person as the previous year during Deby’s tenure.
109
In the context of my sample, if a president keeps the same person in the

vice president position for longer periods of time, this would be inter-
preted as a higher degree of de facto institutionalization compared with a
president who rotates the person in the vice president position every year.
In order to determine the stability of the second-in-command position, I
create a dummy variable that takes a value of 1 if the person in the vice
president or prime minister position is the same person as the previous
year. Cases where this is more frequently true can be interpreted as having
higher degrees of institutionalization.
Additionally, I document whether a minister of defense was appointed
for every country–year observation. The defense portfolio is an especially
important ministerial position because it represents control of military
force. As such, presidents were extremely reluctant to appoint other elites
as the defense minister. In fact, the defense portfolio is the most commonly
kept portfolio by the presidents themselves. Whether the president dele-
gates this position to someone other than himself is therefore a key
component of institutionalized power sharing. The distribution of this
portfolio has significant consequences. In Chapter 6, I show that having
an independent minister of defense significantly lowers the risk of coup
attempts.
Finally, I consider the extent to which the person appointed as the
minister of defense was rotated on a yearly basis. Similar to the second-
in-command position, I create a dummy variable that takes a value of 1 if
the person appointed to be the defense minister was the same person as the
year before. Cases with lower levels of rotation can be interpreted as
having higher degrees of institutionalization.
It is important to note that even when leaders have the ability to choose
who they appoint to them key positions – as they often do – the act of
delegating authority to any other elite serves as a constraint on the leader.
As outlined in Chapter 2, when elites are appointed to influential minis-
terial positions – and especially when the chain of command is formalized
in a constitutional succession order – this results in a shift in power away
from the leader in favor of the elite. This is true even if the elite was a
former ally of the leader. In fact, existing research shows that “allies” tend
to pose the most dangerous threats to autocratic leaders! European mon-
archs, for instance, were most frequently overthrown by their own family
members (Herz 1957; Kurrild-Klitgaard 2000; Kokkoknen and Sundell
2014; Tullock 1987) and African leaders are often deposed by their own
coethnics (Roessler 2016). In sum, the appointment of any elite to a
position of influence within the presidential cabinet constrains the leader
by empowering that particular elite with state resources and the visibility
of a position of power.
110
4.3 General Trends and Patterns
A clear advantage of this dataset is that it is objectively coded in a way

that can be verified and replicated. Rather than hand-coding cases accord-
ing to subjective criteria, this data was constructed by simply observing
whether certain constitutional amendments or cabinet appointments
existed, therefore the coding scheme did not require the researcher to
make judgment calls. In addition, the indicators distinguish between de
facto and de jure forms of institutionalization and allow us to disentangle
the regime from the leader. Another advantage is that this data is presented
as disaggregated indicators. Researchers can therefore decide which dimen-
sions of institutionalization are most relevant to the research question at
hand. Alternatively, the researcher can also choose to aggregate certain
indicators into an institutionalization “score,” although they are not obli-
gated to do so since the measures are presented separately.
To summarize, my dataset produces the following seven indicators that
serve as measures of regime institutionalization. Each variable is coded as
country–year units and takes the form of a dummy variable.
1. Successor policy: is there a constitutional amendment specifying the
rules of succession?
2. Successor (strict): is there a constitutional amendment specifying
exactly who would succeed the president in the case of his death?
3. Term limit: are there constitutional term limits?
4. VP/PM appointed: was a vice president or prime minister appointed?
5. VP/PM same: was the person appointed to the vice president or prime
minister position the same person as the year before?
6. Defense appointed: was a minister of defense appointed?
7. Defense same: was the person appointed as the minister of defense the
same person as the year before?
4.3 e xecutive constraints i n s ub-saharan

africa: general trends and p atterns
What does the empirical data on executive constraints within sub-Sharan
Africa look like? This section provides a descriptive summary of the data
and highlights some general trends and patterns both across countries and
over time.
Figure 4.1 displays cross-sectional differences in institutionalization
across all the countries in my sample.15 Each section of the pie chart
15
Because a number of the variables are highly correlated with each other (as I will
discuss below), I only include successor policy, term limits, VP/PM same, and
defense same in these graphs.
111
Angola Benin Botswana Burkina Faso Burundi
Cameroon Cape Verde CAR Chad Comoros
Congo DRC Ivory Coast Djibouti Eq. Guinea Eritrea
Ethiopia Gabon Gambia Ghana Guinea
Guinea Bissau Kenya Lesotho Liberia Madagascar
Number of constraints 0 1 2 3 4
Figure 4.1.a Cross-sectional differences in institutionalization
illustrates the proportion of years for which each country had one, two,
three, or all four dimensions of institutionalization in place. Black sections
represent the proportion of years for which the country had all four
dimensions in place. Dark grey sections represent the proportion of
years for which the countries had three dimensions in place, and so on.
White sections represent the portion of years for which the country had no
dimensions in place.
112
Malawi Mali Mauritania Mozambique Namibia
Niger Nigeria Rep. of Congo Rwanda Sao Tome
Senegal Sierra Leone Somalia South Africa Sudan
Swaziland Tanzania Togo Uganda Zambia
Zimbabwe
Figure 4.1.b Cross-sectional differences in institutionalization
Countries that have darker circles can be interpreted as being more

institutionalized on average, although these graphs do not take into
account change over time.16 Since these graphs show proportion of
years, the reader should be reminded that some countries gained indepen-
dence earlier than others, so the length of time countries are included in
the sample sometimes differs. Over 70 percent of the countries in my
sample either gained independence prior to 1970 or were never colonized
and such countries enter my dataset in the 1960s. A small number of
16
Appendix Table 4.2 includes a summary of institutionalization scores by country.
113
countries gained independence later – for instance, Zimbabwe did not

gain independence until 1980 and therefore has fewer years included in
the dataset.
It is clear from the graphs that countries across sub-Saharan Africa
differ widely in terms of level of institutionalization. Namibia, for
instance, is clearly a case with high levels of institutionalization. Since
gaining independence in 1990, Namibia has had formal succession rules
and term limits in the constitution without interruption since the start of
the regime. A vice president and minister of defense have been appointed
in the presidential cabinet every year, and these appointments have been
extremely stable. A constitutional leadership succession was carried out
peacefully in 2005 when the first president of Namibia, Sam Nujoma,
handed power to his successor, Hifikepunye Pohamba.
Contrast Namibia with Togo, a country that has had much lower levels
of institutionalization. Togo, which gained independence from France in
1960, has never had constitutional term limits, and formal succession
policies were in place only after 1992. No elites were appointed to the
vice president or minister of defense position for the first three decades
after independence. When elites were appointed to these cabinet positions
after the end of the Cold War, they were heavily rotated and replaced
every two to three years.
Besides cross-sectional differences, levels of regime institutionalization
have also changed over time. Figure 4.2 illustrates the proportion of
countries with executive constraints graphed over time.
The trends documented in this graph suggest that various dimensions
of institutionalization seem to rise together. Countries also appear to
become increasingly more institutionalized over time, although it’s impor-
tant to note that by the end of the data, 2010, only about half of countries
in sub-Saharan Africa have formal or informal executive constraints.
There is a sharp increase in the prevalence of term limits in the early
1990s, and in general, more countries became institutionalized after the
end of the Cold War.
Returning to the case of Togo, even though the country was almost
entirely un-institutionalized from 1960 to 1990, it has become more
institutionalized in recent decades. Following the end of the Cold War,
formal succession policies were created under then-president Eyadema
Gnassingbe, who also began to appoint elites to the vice president and
minister of defense cabinet positions. The regime in Togo today is much
more institutionalized that it was in previous decades.
How do various dimensions of institutionalization relate to each other?
Do the measures move together and therefore act as complements to each
other? Or do certain types of executive constraints substitute for other
114
1
.9
.8
.7
.6
.5
.4
.3
.2
.1
0
1960 1970 1980 1990 2000 2010
Year

Figure 4.2 Over time changes in institutionalization
types of constraints? Table 4.3 provides a correlation matrix of the seven

measures.
The first key takeaway is that no two measures are negatively corre-
lated, so there does not appear to be evidence of substitution between
different dimensions of institutionalization. In other words, it appears
unlikely that a leader would choose not to appoint a minister of defense
because she has appointed a vice president.
This observation is, in itself, an interesting finding because it remains an
open question whether informal institutions compete with or strengthen
existing formal institutions. Scholars have noted that when informal insti-
tutions arise, they can often undermine or weaken formal institutions
(Gryzmala-Busse 2010; Borocz 2000b). Corruption serves as such an
example. Grzymala-Busse argues that “informal institutions serve as tem-
plates and substitutes for formal choices. Even as they compensate for
formal institutions, informal rules can subvert them. Informal institutions
also reify formal rules by defining and expanding their domain, and by
providing incentives and information to follow formal institutions” (311).
On the other hand, informal institutions can also help strengthen formal
institutions. “Informal rules may generate precedents and prevalent prac-
tices that are then formalized for efficiency’s sake” (Gryzmala-Busse 2010,
321). They can support and reify formal institutions by “delineating the
domains where formal institutions rule – and by providing information and
enforcement that promote the functioning of formal institutions” (324).
115
Table 4.3 Correlation matrix of seven dimensions of institutionalization
Successor Successor Term VP/PM VP/PM Defense Defense

policy (strict) limits appoint same appoint same
Successor policy 1
Successor (strict) 0.855 1
Term limits 0.413 0.424 1
VP/PM appoint 0.187 0.238 0.292 1
VP/PM same 0.159 0.166 0.192 0.618 1
Defense appoint 0.209 0.221 0.186 0.203 0.122 1
Defense same 0.199 0.190 0.120 0.134 0.239 0.644 1
Note: Bolded numbers indicate two dimensions that are highly correlated (r > 0.6)
A second important takeaway is that some measures are highly corre-

lated with each other. Having succession rules is highly correlated with
having strict succession rules (r = 0.855), appointing the same VP/PM is
highly correlated with appointing a VP/PM (r = 0.618), and appointing
the same defense minister is highly correlated with appointing a defense
minister (r = 0.644). This is not surprising since each of these variables is
just a stricter version of the corresponding variable. In much of my
empirical analysis, I exclude measures that are highly correlated with
another indicator of institutionalization.
4.4 regime typologies

How do my measures of regime institutionalization compare with other
existing datasets of authoritarian institutions? Autocratic regime institu-
tionalization is difficult to operationalize, and there are not many existing
datasets that measure this concept directly. Perhaps the most direct way in
which regime institutionalization has been codified is through the regime
typologies framework.
In a seminal study, Geddes (1999a) classifies all autocratic regimes into
one of the following regime types: military, single-party, personalist, or
hybrids of these categories. These classifications are based on whether
control over “policy, leadership selection, and the security apparatus is in
the hands of a ruling party (dominant-party dictatorships), a royal family
(monarchy), the military (rule by the military institution), or a narrow
group centered around an individual dictator (personalist dictatorship)”
(Geddes, Wright, and Frantz 2014, p. 318).
116
4.4 Regime Typologies
Single-party regimes are defined as regimes in which the “party has

some influence over policy, controls most access to political power and
government jobs, and has functioning local-level organizations” (Geddes
1999a, 31). By contrast, in personalist regimes, leaders have “consoli-
dated control over policy and recruitment in his own hands, in the process
marginalizing other officers’ influence and/or reducing the influence and
functions of the party” (Geddes 1999a, 31). Military regimes are defined
as those “governed by an officer or retired officer, with the support of the
military establishment and some routine mechanism for high level officers
to influence policy choice and appointments” (Geddes 1999a, 31).
Geddes’ study and associated dataset have made immense contributions
to scholarship on authoritarian politics. It set the agenda for renewed
interest in the study of nondemocratic regimes outside of the industrialized
world and stimulated a large body of recent work on the policies, institu-
tions, and consequences of autocratic rule.17 However, the use of typologies
as an indicator for regime institutionalization has resulted in four central
problems – two theoretical and two measurement driven.
First, placing regimes into time-invariant categories obscures institu-
tional changes over time. This is a significant limitation because leaders
within the same regime often make very different decisions about the
extent to which they implement executive constraints. Founding leaders,
for instance, are often much more personalist than their successors
(Bienen and van de Walle 1989). Regimes tend to become less personalist
over time, especially after multiple leadership transitions.
Take the cases of Mexico under the rule of the PRI and China – perhaps
two of the most commonly cited examples of highly institutionalized
dictatorships. They are both coded as party-based regimes for the entire
duration of the party’s tenure.18 Although these regimes eventually
became strong and autonomous organizations, they were significantly
less institutionalized under their first leaders. The PRI under Plutarco
Elias Calles and the CCP under Mao Zedong resembled personalist
regimes, where each respective leader had consolidated control, reducing
the parties’ influence. Calles, the founder of the PRI,19 controlled the
actions and decisions of three presidents after him during what is referred
to as the Maximato Period. It was not until the fourth president, Lázaro
Cárdenas, that the regime established real leadership turnover norms
17
According to Google Scholar, Geddes (1999a) has been cited almost 300 times.
Geddes (1999b), an annual review article covering the same material as Geddes
(1999a), has been cited over 1,500 times.
18
The PRI was in power from 1929–2000 and the CCP has been in power since 1949.
19
Known as the Partido Nacional Revolucionario (PNR) at the time.
117
where power was handed over to the successor after the president’s term
ended (Buchenau 2006).
Mao was a similarly personalist founding leader. He promoted and
purged elite politicians at will and launched political campaigns, such as
the Hundred Flowers Movements and Cultural Revolution, during which
he arbitrarily purged officials who he deemed critical of the regime
(Meisner 1986). Over time, as Deng Xiaoping came into power, the CCP
eventually became an organization in which party promotion followed
established norms and procedures. In sum, many regimes change drastically
over time, and this is impossible to capture using a single time-invariant
regime category.20
Second, categorical distinctions between authoritarian regime types are
often insufficiently nuanced and obscure variation in institutional strength
within regime types. Take the category of dominant-party regimes. The
Soviet Union and China are coded as dominant-party regime types, and
they indeed were ruled by strong parties with organizational autonomy that
lasted through several leadership transitions. However, the category of
dominant-party regimes also includes cases such as the PDG in Guinea or
Communist Party of Kampuchea (CPK, also known as the Khmer Rouge)
in Cambodia. These parties, unlike the highly institutionalized communist
parties, were entirely concentrated around a single charismatic leader and
were not able to survive past the death and ousting of the leader.21
Third, regime typologies are composite indices that aggregate various
dimensions of leaders, institutions, and military structures into a single
category. As a result, it is difficult for researchers to identify the effects of
individual institutions (Pande and Udry 2006). Some regimes, for
instance, appear to be party-based, when in actuality the party is attached
to a strong and charismatic leader who merely exploits the party as a
personal vehicle to amplify his authority.
The Union Soudanaise – Rassemblement Democratique Africain (US-
RDA) under Modibo Keita in Mali, for instance, is coded as part of a
dominant-party regime. Yet national policies were determined entirely at
the discretion of Keita alone, and the US-RDA lacked institutionalized
rules and permanent structures. Eight years after taking power, Keita was
deposed in a coup and the party was banned. Although Keita, who was
20
It is possible for countries to be coded as different regime types over time, but most
countries are not coded as very many different types. The average country is coded
as two different types of regimes over an average time span of forty-four years, and
the average regime coding lasts about eighteen years.
21
The opposite scenario is sometimes true as well: not all cases coded as military or
personalist regimes are equally unconstrained.
118
also a self-proclaimed socialist, portrayed Mali as a one-party state, the

ruling party was actually extremely weak.
Fourth, although Geddes outlines a clear set of guidelines that were
employed to categorize regimes into different categories, a number of
these criteria used to code regime types require the researcher to make
subjective decisions about how to code the regime. Examples of subjective
criteria include: “Does the party have functioning local-level organiza-
tions that do something reasonably important, such as distribute seeds or
credit or organize local government?” or “has rule of law been main-
tained?”. The possibility of measurement error based on subjective coding
rules is also heightened by the fact that the dataset spans multiple regions
and time periods and often relies on information from various country
experts. An example of criteria that relies on different country sources
includes: “Does the country specialist literature describe the politburo-
equivalent as a rubber stamp for the leader?” Different country experts
from various regions may have different standards for evaluating institu-
tions making it difficult to know whether the criteria are being applied
uniformly. Moreover, it is not clear what the individual responses to these
criteria are, nor how the responses to these criteria are aggregated to
produce single regime categories.
4.4.1 Comparison of Institutionalization and Regime Typologies

In this section, I compare my indicators of regime institutionalization with
data on regime typologies. I find that there are a number of discrepancies
between my measures of institutionalization and regime typologies. Many
regimes that have been coded as dominant-party regimes are actually not
very institutionalized at all, and some regimes that have been coded as part
of personalist or military regimes actually have a number of executive
constraints in place.
The dataset that was introduced in Geddes (1999a) was updated and
rereleased by Geddes, Wright, and Frantz (2014). I will refer to the updated
dataset as the GWF dataset for the remainder of this chapter. GWF covers
all country–years with autocratic governments between 1946 and 2010 in
independent countries. From this country–year data, GWF creates a list of
280 autocratic regimes with their start and end dates, as detailed by their
codebook. Similar to the earlier regime datasets, GWF classifies all auto-
cratic regimes in their dataset into one of the following regime types:
monarchy, personal, military, party, party-personal, party-military, mili-
tary-personal, party-personal-military, oligarchy, indirect military.
Since GWF is cross-sectional, I collapse the country–year panel data
from my Africa sample into regime-level observations in order for my
119
dataset to be comparable with the GWF regime-level observations. The

resulting cross-sectional dataset has eighty-three regime-level observa-
tions.22 Since my original measures of regime institutionalization are
dummy variables, to transform them into cross-sectional indicators I
calculate the percentage of years for which a dimension of institutionali-
zation was implemented. For instance, a leader that was in power for
twenty years and had a vice president for ten of those years would score a
.50 on that particular dimension of institutionalization.23
Table 4.4 provides a comparison between my measures of institutio-
nalization and regime type for the Africa subsample. Shaded cells
represent disagreement between the assigned regime type and institutio-
nalization score on that particular dimension. For instance, a number of
regimes that are coded as party-based had a vice president or prime
minister less than 70 percent of the time in which the regime was in
power. This disagreement is highlighted by having shaded cells under
the dimension “second named” for those parties. In addition, a number
of regimes that are coded as military or personalist actually have a vice
president or prime minister more than 70 percent of the shaded cells.
A comparison of GWF and my measures of institutionalization reveals
a high number of discrepancies. Out of eighty-three regimes in Table 4.4,
only fifteen do not have any shaded cells. In other words, 82 percent of the
observations have an inconsistency between the assigned regime typology
and at least one dimension of institutionalization. This finding is consis-
tent with other studies that argue that the GWF typology better reflects the
later years of party institutionalization but is a poor representation of
earlier years (Lucardi 2017).
The comparison reveals that many regimes that are coded as dominant-
party by GWF are actually not very institutionalized at all. Mali under the
rule of the Sudanese Union scores zeros for all seven dimensions of party
institutionalization, making it the least institutionalized regime in the
dataset. Niger under the rule of the Nigeria Progressive Party scores
zeros on five out of seven dimensions. Party-based regimes in the
Gambia, Guinea, Guinea Bissau, and Madagascar each score zeros on
22
I only focus on comparisons of pure regime types, leaving out hybrid regime types, such
as party-personal, party-military, military-personal, or triple hybrid types. These types
of hybrid regimes are excluded because it is not clear whether these regimes are highly
institutionalized or not. For instance, should a researcher expect a party-personal
regime to be highly institutionalized (due to the presence of some party attributes) or
not institutionalized (due to the presence of some personalist attributes)?
23
Formally, the measure is calculated as following:
X
years with institutionalization
percent years with institutionalization ¼ X
years in power
120
four out of six dimensions of institutionalization. In sum, many regimes

that are coded as dominant-party perform very poorly when we examine
their institutionalization scores.
A closer examination of the cases that exhibit much disagreement also
highlights the ambiguity of the GWF coding scheme. Take the example of
Niger from 1960 until 1974. The Nigerian Progressive Party scored zeros
Table 4.4 Comparison between institutionalization and regime type
GWF case Regime type Success Success Term VP/PM VP/PM Defense Defense
rules (strict) limits appoint same appoint same
Party-based regimes and monarchies <50% <50% <50% <70% <50% <70% <50%
(institutionalized)
Angola 75–NA Party
Botswana 66–NA Party
Ethiopia 91–NA Party
Gambia 65–94 Party
Guinea 58–84 Party
G. Bissau 74–80 Party
Ivory Coast 60–99 Party
Kenya 63–02 Party
Madagascar 60–72 Party
Mali 60–68 Party
Mozambique 75–NA Party
Namibia 90–NA Party
Niger 60–74 Party
Rwanda 62–73 Party
Senegal 60–00 Party
Sierra Leone 68–92 Party
Tanzania 64–NA Party
Zambia 67–91 Party
Zambia 96–NA Party
Zimbabwe 80–NA Party
Ethiopia 89–74 Monarchy
Swaziland 68–NA Monarchy
Personalist and military regimes >50% >50% >50% >70% >50% >70% >50%
(not institutionalized)
Benin 60–63 Personal
Burkina Faso 60–66 Personal
Burkina Faso 87–NA Personal
Cameroon 83–NA Personal
CAR 03–NA Personal
CAR 60–65 Personal
Chad 82–90 Personal
Chad 90–NA Personal
Rep Congo 60–64 Personal
Rep Congo 97–NA Personal
DRC 60–97 Personal
DRC 97–NA Personal
Gambia 94–NA Personal
Ghana 81–00 Personal
Guinea 08–10 Personal
G. Bissau 02–03 Personal
Ivory Coast 00–NA Personal
Ivory Coast 99–00 Personal
121
Table 4.4 (continued)
Liberia 80–90 Personal
Madagascar 09–NA Personal
Madagascar 75–93 Personal
Malawi 64–94 Personal
Mali 68–91 Personal
Mauritania 08–NA Personal
Mauritania 60–78 Personal
Niger 96–99 Personal
Sierra Leone 97–98 Personal
Somalia 69–91 Personal
Sudan 69–85 Personal
Sudan 89–NA Personal
Togo 60–63 Personal
Togo 63–NA Personal
Uganda 66–71 Personal
Uganda 86–NA Personal
Benin 65–67 Military
Burkina Faso 80–82 Military
Burundi 87–93 Military
Chad 75–79 Military
Ghana 66–69 Military
Madagascar 72–75 Military
Mauritania 05–07 Military
Nigeria 66–79 Military
Sierra Leone 67–68 Military
Sudan 58–64 Military
Note: Shaded cells represent dimensions of party institutionalization that are in disagreement with the regime typology
coding. For example, for party-based regimes, “Second Appoint < 70%” means that the regimes listed in this column
are coded as party-based regimes, but have a second in command appointed less than 70 percent of the time.
For non party-based regimes, “Second Appoint >80%” means that the regimes listed in this column are NOT coded
as part of party-based regimes, but have a second in command appointed more than 80 percent of the time.
NA refers to regimes that were still in power as of 2010.
on six out of seven dimensions of institutionalization, making it one of the

weakest regimes in my sample, yet it is coded as part of a party-based
regime by GWF. Even when we look at the case description in the GWF
codebook, the leader does not appear to constrained at all.
Niger (1960–74)
Start: 8/3/1960 Independence under single-party rule. The PPN won the December
1958 preindependence election with help from the French. Prior to independence,
the most popular rival party was outlawed and its leaders jailed. Power was
centralized under Diori, who controlled ministerial appointments without parlia-
mentary scrutiny, could appoint and dismiss civil servants and military officers,
and could decree and veto laws […]. End: 4/15/1974 Coup led by the Army Chief
122
4.5 Comparisons with Other Existing Datasets
of Staff ousted the civilian government[…] . (Geddes, Wright, and Frantz 2012,
82)
It is difficult to know why the PPN in Niger was coded as part of a

party-based regime, rather than a personalist regime (or at least a party-
personal regime), in part because regime typologies are aggregate mea-
sures. The advantage of disaggregated indicators of institutionalization is
that the researcher can clearly observe the extent to which the regime
fulfills certain requirements (such as having leadership succession rules)
and make decisions based on transparent criteria.
The comparison of the two datasets also reveals that a number of
regimes that were coded as military or personalist are actually quite
institutionalized. Several regimes that are coded as personalist actually
exceed the threshold of institutionalization for many dominant-party
regimes. Although Burkina Faso, under the presidency of Blaise
Compaore, is coded as a personalist regime, it had an appointed prime
minister and minister of defense for the majority of Compaore’s tenure.24
The constitution of Burkina Faso also had succession procedures and term
limits during most of this period.
4. 5 comparisons with other e xisting datasets

The final section of this chapter presents graphical comparisons of my
institutionalization measures against two other commonly used datasets
of institutional strength – Polity (Marshall, Gurr, and Jaggers 2016) and
V-Dem (Coppedge et Al. 2016). The comparisons show that while exist-
ing variables can provide broad impressions of institutional strength, they
are often not sufficiently nuanced to accurately capture disaggregated
dimensions of executive constraints.
First I compare my disaggregated indicators of executive constraints
against the “XCONST” variable from the Polity dataset. “XCONST” is
described as a variable that
refers to the extent of institutionalized constraints on the decision- making powers
of chief executives, whether individuals or collectivities. 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 advisors 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. (Marshall, Gurr and Jaggers 2016, 24)
24
Recall that the data goes up until 2010, though Compaore was in power until 2014.
123
Proportion of observations with constraint (author’s data) 1.00
0.75
Type of constraint
Succession rules
Term limits
0.50 Stable VP appoint
Stable Defense appoint
0.25
1 2 3 4 5 6 7
POLITY "XCONST" score
Figure 4.3 Comparison between institutionalization and XCONST variable
“XCONST” is coded on a seven-point scale, where higher scores are

interpreted as more institutionalized and lower scores are interpreted as
less institutionalized.
Figure 4.3 provides a comparison of my institutionalization scores and
the “XCONST” variable from the Polity dataset. The graph reveals many
inconsistencies between the “XCONST” coding and observed executive
constraints in sub-Saharan Africa. For instance, while 100 percent of
countries that scored a six on the XCONST variable have constitutional
succession rules in place, less than 75 percent of countries that scored a
seven on the XCONST variable (a higher institutionalization score) have
constitutional succession rules in place.
I also provide comparisons between my institutionalization measures and
two V-Dem measures of executive oversight: legislative constraints on the
executive and judicial constraints on the executive. It is important to note
that these V-Dem measures should not be considered substitutes for my
measures of executive constraints. Instead, these comparisons reveal that
there are important differences across different types of executive constraints.
Simply having strong legislative oversight does not imply that the regime
automatically has strong constitutional rules that constrain the president.
I compare my regime institutionalization measures against the “legis-
lative constraints on the executive index” variable. This question was
124
Proportion of observations with constraint (author’s data)
0.8
Type of constraint
0.6
Succession rules
Term limits
Stable VP appoint
0.4
0.2
0 0.25 0.5 0.75 1

Level of legislative constraint on the executive
“Ordinal legislative constraints on the executive” variable
from V−DEM used. Five level ordered variable, with 0
being weakest constraints and 1 being strongest constraints.
Figure 4.4 Comparison between institutionalization and V-Dem measure of legislative

constraint
coded by asking experts the following question: “To what extent is the
legislature and government agencies (e.g. controller general, general pro-
secutor, or ombudsman) capable of questioning, investigating, and exer-
cising oversight over the executive?”
Figure 4.4 provides a comparison of the legislative constraint scores
against my regime institutionalization measures. While the two vari-
ables appear generally correlated, some important differences arise.
For instance, a greater proportion of countries in my sample that
score a 0.5 on the V-Dem legislative constraint index have constitu-
tional succession rules in place compared with countries that score a
0.75 on the V-Dem legislative constrain index. This suggests that
simply having legislative constraints in place does not automatically
guarantee the existence of certain executive constraints, such as con-
stitutional succession rules.
Finally, I also conduct a comparison of my institutionalization mea-
sures against the “judicial constraints on the executive index” variable.
This question was coded by asking experts the following question: “To
what extent does the executive respect the constitution and comply with
court rulings, and to what extent is the judiciary able to act in an inde-
pendent fashion?”
125

0.8
0.6
Type of constraint
Succession rules
Term limits
Stable VP appoint
0.4
0.2
0 0.25 0.5 0.75 1
Level of judicial constraint on the executive
“Ordinal judicial constraints on the executive” variable
Figure 4.5 Comparison between institutionalization and V-Dem measure of judicial

constraint
Figure 4.5 presents a comparison between my regime institutionaliza-

tion measures and the V-Dem measures of judicial constraint. Perhaps
unsurprisingly, the two variables do not appear to be very closely related,
which implies that judicial constraints should be analyzed separately from
This chapter presented an original dataset of autocratic regime insti-
tutionalization in all sub-Saharan regimes from 1960 to 2010. First I
discussed how I conceptualize and operationalize autocratic regime
institutionalization more generally, then I presented the dataset by
explaining my coding strategy and showing descriptive findings.
Finally I compared my dataset to other commonly used datasets of
authoritarian institutions, such as regime typologies, Polity scores,
and V-Dem measures.
The rest of the book will be primarily empirical and will use this dataset
to test my arguments about the causes and consequences of regime insti-
tutionalization. The next chapter will test the theoretical argument made
in Chapter 2 and show that there is a relationship between initial leader
126
4.6 Conclusion
strength and levels of institutionalization. Chapters 6 and 7 will demon-

strate that institutionalization does indeed have regime stabilizing effects
by examining outcomes such as the length of leader tenure, coup risk, and
leadership succession.
Appendix
Appendix Table 4.1 List of countries and leaders included in this study
Country Leader Years
Angola Antonio Agostinho Neto 1976–1979

Jose Eduardo dos Santos 1980–2010
Benin Hubert Maga 1960–1963
Sourou Migan Apithy 1964–1965
Christophe Soglo 1966–1967
Emile Derlin Zinsou 1968–1969
Hubert Maga 1970–1971
Mathieu Kerekou 1972–1990
Botswana Seretse Khama 1966–1980
Quet K. J Masire 1981–1997
Festus G Mogae 1998–2007
Seretse Khama Ian Khama 2008–2010
Burkina Faso Maurice Yameogo 1960–1965
Sangoule Lamizana 1966–1980
Saye Zerbo 1981–1982
Thomas Sankara 1984–1987
Blaise Compaore 1988–2010
Burundi Michel Micombero 1967–1976
Jean–Baptiste Bagaza 1977–1987
Pierre Buyoya 1988–1992
Sylvestre Ntibantunganya 1994–1995
Domitien Ndayizeye 2003–2004
Jean–Pierre Nkurunziza 2005–2010
Cameroon Ahmadou Ahidjo 1960–1982
Paul Biya 1983–2010
Cape Verde Aristides Maria Pereira 1976–1990
(continued)
127
Appendix Table 4.1 (continued)
CAR David Dacko 1960–1965

Jean–Bedel Bokassa 1966–1979
David Dacko 1980–1981
Andre Kolingba 1982–1993
Ange–Felix Patasse 1994–2002
Francois Bozize 2003–2010
Chad Francois Tombalbaye 1960–1974
Felix Malloum 1976–1978
Goukouni Oueddei 1980–1982
Hissene Habre 1983–1990
Idriss Deby 1991–2010
Comoros Ali Soilih 1976–1977
Ahmed Abdallah Abderemane 1979–1989
Said Mohamed Djohar 1990–1995
Mohamed Taki Abdoulkarim 1996–1998
Assoumani Azali 1999–2006
Ahmed Abdallah Sambi 2007–2010
Congo DRC Joseph Kasavubu 1961–1965
Mobutu Sese Seko 1966–1996
Laurent–Desire Kabila 1997–2000
Joseph Kabila 2001–2010
Cote d’Ivoire Felix Houphouet–Boigny 1960–1993
Henri Konan Bedie 1994–1999
Laurent Gbagbo 2001–2010
Djibouti Hassan Gouled Aptidon 1978–1998
Ismael Omar Gelleh 1999–2010
Equatorial Guinea Francisco Macias Nguema 1969–1979
Teodoro Obiang 1980–2010
Eritrea Issaias Afewerki 1993–2010
Ethiopia Haile Selassie I 1960–1973
Tafari Bante 1974–1976
Mengistu Haile Mariam 1977–1990
Meles Zenawi 1991–1994
Negasso Gidada 1995–2001
Girma Wolde Giorgis 2002–2010
Gabon Leon M’Ba 1960–1967
El Hadj Omar Bongo 1968–2009
(continued)
128
4.6 Conclusion
Gambia Dawada Jawara 1970–1993

Yahya Jammeh 1994–2010
Ghana Kwame Nkrumah 1960–1965
J.A. Ankrah 1966–1968
Edward Akufo–Addo 1970–1971
I.K. Acheampong 1972–1978
Hilla Limann 1980–1981
Jerry Rawlings 1982–2000
Guinea Ahmed Sekou Toure 1960–1983
Lansa Conte 1984–2008
Guinea Bissau Luis De Almeida Cabral 1974–1980
Joao Bernardo Vieira 1981–1998
Kumba Yala 2000–2002
Kenya Mzee Jomo Kenyatta 1964–1978
Daniel arap Moi 1979–2002
Lesotho Bethuel Pakalitha Mosisili 2000–2010
Liberia William Tubman 1960–1970
William Richard Tolbert 1971–1979
Samuel Kanyon Doe 1980–1990
Amos Claudius Sawyer 1991–1993
Charles Ghankay Taylor 1997–2002
Gyude Bryant 2004–2005
Ellen Johnson–Sirleaf 2006–2010
Madagascar Philibert Tsiranana 1960–1971
Gabriel Ramanantsoa 1972–1974
Didier Ratsiraka 1976–1992
Malawi Hastings Kamuzu Banda 1965–1993
Mali Modibo Keita 1960–1968
Moussa Traore 1969–1990
Mauritania Moktar Ould Daddah 1960–1978
Mohamed Khouna Ould Haidalla 1980–1984
Maawiya Ould Sid’Ahmed Taya 1985–2004
Ely Ould Mohamed Vall 2005–2006
Sidi Mohamed Ould Cheikh Abdellahi 2007–2008
Mozambique Samora Moises Machel 1976–1986
Joaquim Alberto Chissano 1987–2005
(continued)
129
Armando Emilio Guebuza 2006–2010

Namibia Samuel Daniel Nujoma 1990–2004
Hifikepunye Pohamba 2005–2010
Niger Hamani Diori 1960–1973
Seyni Kountche 1974–1987
Ali Saibou 1988–1992
Mahamane Ousmane 1993–1995
Ibrahim Bare Mainassara 1996–1998
Tandja Mamadou 2000–2004
Mamadou Tandja 2005–2009
Nigeria Alhaji Abubakah Tafawa Balewa 1960–1965
Yakubu Gowon 1966–1974
Olusegun Obasanjo 1976–1979
Alhaji Shehu Shagari 1980–1983
Muhammadu Buhari 1984–1985
Ibrahim Babangida 1986–1992
Sani Abacha 1994–1997
Republic of Congo Abbe Fulbert Youlou 1960–1963
Alphonse Massamba–Debat 1964–1967
Marien Ngouabi 1970–1976
Joachim Yhombi–Opango 1977–1978
Denis Sassou–Nguesso 1979–1992
Pascal Lissouba 1993–1997
Rwanda Gregoire Kayibanda 1962–1972
Juvenal Habyarimana 1973–1993
Pasteur Bizimungu 1994–1999
Paul Kagame 2000–2010
Sao Tome Manuel Pinto Da Costa 1976–1990
Senegal Leopold–Sedar Senghor 1963–1980
Abdou Diouf 1981–1999
Sierra Leone Milton Augustus Striery Margai 1961–1963
Albert Michael Margai 1964–1966
Banja Tejan–Sie 1968–1970
Siaka Probyn Stevens 1971–1985
Joseph Saidu Momoh 1986–1991
Valentine Esegragbo Melvine Strassar 1992–1995
(continued)
130
4.6 Conclusion
Somalia Aden Abdullah Osman 1961–1967

Abdirashid Ali Shermarke 1968–1969
Mohamed Siad Barre 1970–1990
Abdullahi Yussuf Ahmed 2006–2008
Sheikh Sharif Sheikh Ahmed 2009–2010
South Africa Nelson Rolihla Mandela 1994–1998
Thabo Mbeki 1999–2008
Jacob Zuma 2009–2010
Sudan Ibrahim Abboud 1960–1964
Ismail Al–Azhari 1966–1969
Jaafar Al Nemery 1970–1974
Gaafar Mohamed Nimeri 1976–1984
Ahmad Ali Al–Mirghani 1986–1987
Omar Hassan Ahmad Al–Bashir 1989–2010
Swaziland Barnabas Sibusiso Dlamini 2000–2010
Tanzania Julius Kambarage Nyerere 1962–1985
Ali Hassan Mwinyi 1986–1995
Benjamin William Mkapa 1996–2005
Jakaya Mrisho Kikwete 2006–2010
Togo Sylvanus Olympio 1960–1962
Nicolas Grunitzky 1963–1966
Gssingbe Eyadema 1967–2004
Faure Gssingbe 2005–2010
Uganda Milton Obote 1963–1970
Idi Amin 1971–1978
Milton Obote 1981–1985
Yoweri Museveni 1986–2010
Zambia Kenneth David Kaunda 1964–1991
Frederick Chiluba 1992–2001
Levy Patrick Mwanawasa 2002–2008
Rupiah Banda 2009–2010
Zimbabwe Robert Mugabe 1980–2010
131
Appendix Table 4.2 Summary of institutionalization scores by country
Term
Country Successor policy Successor (strict) limits VP/PM appoint VP/PM same Defense appoint Defense same
Angola 1 0.542 0.542 0.480 0.342 0.885 0.714

Benin 0.400 0.400 0.440 0.180 0.080 0.460 0.240
Botswana 1 0.311 1 1 0.866 0.088 0.044
Burkina Faso 0.440 0.440 0.440 0.500 0.340 0.820 0.500
Burundi 0.348 0.348 0.325 0.534 0.255 0.441 0.255
Cameroon 1 1 0.260 0.800 0.560 1 0.720
Cape Verde 0.857 0.857 0.857 1 0.857 0.971 0.714
CAR 0.440 0.440 0.260 0.500 0.180 0.320 0.100
Chad 0.580 0.280 0.280 0.560 0.220 0.680 0.300
Comoros 0.485 0.485 0.800 0.742 0.228 0.200 0.057
Congo DRC 0.166 0.166 0.625 0.395 0.145 0.395 0.145
Cote d’Ivoire 0.980 0.480 0.200 0.400 0.240 0.780 0.580
Djibouti 0.545 0.545 0.878 1 0.818 0.939 0.727

Equatorial Guinea 0.289 0.289 0.078 1 0.631 0.342 0.236
Eritrea 0.722 0.722 0.722 0 0 1 0.833
Ethiopia 0 0 0.200 0.880 0.680 0.960 0.680
Gabon 0.180 0.160 0.200 0.980 0.800 0.640 0.480
Gambia 0.350 0.350 0 0.975 0.775 0.125 0.075
Ghana 0.454 0.318 0.340 0.500 0.363 0.704 0.363
Guinea 0.420 0.020 0.020 0.480 0.340 0.540 0.320
Guinea Bissau 0.722 0.722 0.333 0.777 0.277 0.750 0.250
Kenya 1 1 0.413 0.934 0.739 0.413 0.326
(continued)
Lesotho 1 1 0 1 0.818 0 0
Liberia 0.860 0.860 0.560 0.640 0.460 1 0.680
Madagascar 0.380 0.380 0.260 0.920 0.620 0.700 0.420
Malawi 0.733 0.733 0.355 0.377 0.266 0.288 0.133
Mali 0.380 0.380 0.500 0.460 0.260 0.580 0.320
Mauritania 0.380 0.380 0.100 0.460 0.240 0.660 0.320
Mozambique 1 0.571 0.571 0.742 0.600 1 0.885
Namibia 1 1 1 1 0.857 1 0.761
Niger 0 0 0.140 0.580 0.300 0.600 0.280
Nigeria 0.560 0.500 0.420 0.600 0.460 0.660 0.380
Rep Congo 0.400 0.340 0.280 0.780 0.500 0.560 0.320
Rwanda 1 1 0.416 0.395 0.229 0.625 0.458
Sao Tome 1 1 0.571 0.771 0.285 0.828 0.514
Senegal 1 1 0.319 0.680 0.468 0.957 0.659
Sierra Leone 0.708 0.708 0.708 0.958 0.479 0.083 0.041
Somalia 0.176 0 0.235 0.970 0.617 1 0.529
South Africa 1 1 0.882 1 0.647 1 0.705
Sudan 0.440 0.440 0.240 0.860 0.420 0.660 0.340

Swaziland 1 1 0 1 0.636 0 0
Tanzania 0.723 0.723 0.723 1 0.765 0.702 0.510
Togo 0.360 0.360 0.060 0.440 0.180 0.400 0.260
Uganda 0.326 0.326 0.217 0.847 0.608 0.543 0.413
Zambia 0.434 0.434 0.413 0.934 0.500 0.739 0.413
Zimbabwe 1 1 0 0.741 0.516 0.709 0.548
4
How Should Institutionalization Be Measured?
One of the core arguments of this book is that autocratic regimes vary
drastically in the extent to which they are institutionalized. This chapter
lays the empirical groundwork for this claim by presenting detailed data on
regime institutionalization. In doing so, I also provide a discussion of how
we should define and measure autocratic regime institutionalization – a key
concept often used, but not well operationalized – in the literature.
I define regime institutionalization as the creation of rules and proce-
dures that structure the distribution of power and resources within the
ruling coalition. In an autocratic setting, regime institutionalization serves
to tie the leader’s hands. This chapter presents an original time-series
cross-sectional dataset measuring regime institutionalization in forty-six
countries in sub-Saharan Africa from 1960 to 2010. I discuss my coding
strategy in detail and present examples of each type of executive
constraint. I also summarize general trends and patterns in the data to
provide a broad perspective of what institutionalization looks like within
sub-Saharan Africa and how it differs across countries.
This chapter also compares this dataset against other commonly used
datasets of authoritarian institutions, such as regime typologies (Geddes,
Wright, and Frantz 2014). I show that regime typologies often do not reflect
the extent to which executive constraints are in place, and many countries
coded as party-based regimes are not very institutionalized at all. In addi-
tion, I also compare my measures of institutionalization against other
commonly used datasets of institutional strength – Polity (Marshall,
Gurr, and Jaggers 2016) and V-Dem (Coppedge et al. 2016).
Another important implication that emerges from this chapter is that
countries in sub-Saharan Africa vary widely in their level of institutiona-
lization, even during the decades of authoritarian rule following indepen-
dence. A small number of recent studies document that African countries
have become more institutionalized over time, especially after the end of
the Cold War (Posner and Young 2007). Others show that African party
93
systems in the multiparty era are not as uniformly un-institutionalized as

previous scholars have argued (Riedl 2014). Yet scholarly consensus
about the quality and presence of executive constraints in African auto-
cracies, especially during the decades immediately following indepen-
dence, has not changed. Scholars routinely write off institutions from
that period, and politics prior to the 1990s is believed to have been driven
primarily by “big man rule” and personalist dictatorships (Bienen 1978;
Decalo 1976; Jackson and Rosberg 1982; Zolberg 1966). By contrast, the
data presented here tells a different story. African politics was not uni-
formly un-institutionalized during the Cold War era – some incumbents
did indeed regulate power through formal or informal institutions. These
mechanisms help to explain why some dictatorships have lasted for
decades, even surviving the introduction of multiparty elections.
4 . 1 co n c e p t u a li z i n g au t h o r i ta r i a n r e g i m e
Before I define regime institutionalization it is first helpful to note what I
mean by “regime.” I consider a regime as the set of rules, norms, and
institutions that govern the way in which the government is run. Most
relevant for this study, these rules determine the way in which leaders
enter and exit office. A regime can have multiple leaders, as long as these
leaders enter and exit office according to established rules and norms. I
consider an irregular leadership turnover – often due to the forced exit of a
leader and/or the entry of a leader via a coup – as signaling a break in the
regime (Goemans, Gleditsch, and Chiozza 2009).1
For instance, the Gambia experienced a regime change in 1994 when
the armed forces staged a coup and deposed Gawda Jawara – the leader
who had been in power since 1962. Ghana also experienced a regime
change in 1993 when it became classified as a democracy (more on how I
define democracy later) – though the leader did not change. Flight lieute-
nant Jerry Rawlings took power via a coup in Ghana in 1979. In the early
1990s, in response to increasing pressures to democratize, Rawlings lega-
lized political parties and allowed for multiparty elections to occur in
1992. Although Rawlings won the elections, and remained the president
of Ghana, these elections were considered free and fair, and importantly -
there has been party and leadership turnover since then. Ghana is there-
fore coded as having a regime break in 1993. Finally, Tanzania has
remained the same regime since independence, despite having multiple
1
This definition of an autocratic regime is similar to Svolik’s (2012) definition of a
“ruling-coalition spell.”
94
leaders. Since independence was granted in 1961, Tanzania has had five
different presidents, but the same ruling coalition has remained in power,
and the ways in which the incumbent is selected has not drastically
changed.2
I define autocratic regime institutionalization as the creation of rules,
procedures, and hierarchies that structure the ways in which power and
resources are distributed within the ruling coalition. Examples of this
include establishing leadership succession procedures or creating rules
that govern elite promotion. When a leader institutionalizes the regime,
she takes actions that constrain her ability to take arbitrary actions in the
future. Autocratic regime institutionalization occurs when executive
constraints are created that tie the leader’s hands.
My conceptualization of institutionalization closely follows Huntington’s
(1968) definition of political institutionalization as “the process by
which organizations and procedures acquire value and stability” (12).
In particular, Huntington identifies several key features of institutiona-
lization that are most relevant for the survival of authoritarian regimes.
First, adaptability and the ability to outlive the founder is a key char-
acteristic of institutionalized organizations (13). Second, autonomy and
the ability of an organization to exist independently of particular actors
is a key feature of institutionalization. According to Huntington, an
organization that is merely an instrument of a leader lacks autonomy
and institutionalization (20).
This discussion is also related to Levitsky and Murillo’s (2009) discus-
sion of institutional strength.3 They argue that institutional strength
should be judged across two dimensions: enforcement and stability.
Strong institutions are those that political elites comply with in practice.
This conceptualization of institutional strength ties directly with my dis-
cussion of regime institutionalization. Regimes with institutionalized
2
Note that this conceptual discussion of “regime” does not affect the coding decisions
for this project because the measures of institutionalization I construct operate at the
country–year level, not at the regime level.
3
Much of the discussions of institutionalization has centered around the institutiona-
lization of political parties, rather than regimes. Huntington (1965), for instance,
defines party institutionalization as the process by which parties become established
and acquire value and stability. Levitsky (1998) likens institutionalization to the
entrenchment of the “rules of the game” that shape the expectations, payoffs, and
behavior of actors in his discussion of the Peronist party in Argentina (p. 80).
Panebianco (1988) argues that institutionalization transfers authority from the
leader to the party, and very few charismatic parties survive this transfer. Research
on party system institutionalization also describes strong party organizations as
having independent status and value, not subordinated to the interests of ambitious
leaders (Mainwaring and Scully 1995).
95
practices – those governed by stable rules that do not shift with changing
circumstances – are durable regimes that are likely to survive leadership
transitions precisely because they have rules that can outlast the circum-
stances in which they were initially created.4
In the context of an authoritarian regime, institutionalization has
several important implications. First, the creation of rules and procedures
depersonalizes the ways in which the regime operates by constraining the
autocratic leader’s ability to make arbitrary decisions in the future. This is
especially relevant in dictatorships, where power is often concentrated in
the hands of a single leader. When a leader establishes term limits, for
instance, she loses the ability to decide arbitrarily when and whether she
will step down from office. This kind of depersonalization can also hap-
pen through the creation of merit-based promotion within the executive.
Regimes with set rules, procedures, and hierarchies that regulate elite
promotion are more institutionalized compared with systems in which
a single incumbent can appoint or purge elites at any time without
constraint.
Second, institutionalization can empower other elites by establishing
hierarchical positions and rules regarding the appointment of elites to
important positions. A regime with a presidential cabinet in which elites
are appointed to important ministerial portfolios, such as defense or
finance, represents a regime in which power and resources are not entirely
concentrated within the hands of the leader. Such a regime is more
institutionalized compared with a presidential cabinet in which the incum-
bent holds all the important portfolios herself, and there are no estab-
lished rules or norms for the appointment of these positions.
However, it is important to note that institutionalization does not
always result in the empowerment of other elites. Some types of executive
constraints only limit the incumbent. Term limits on their own create
restrictions on the length of time that the incumbent can remain in office,
but do not specify who will replace the incumbent or how that replace-
ment process will occur.
Third, institutionalization separates the ruler from the regime.
Importantly, an institutionalized regime is one that can perpetuate its
own existence beyond the influence of an individual leader. A regime
that is not institutionalized is a government of one: the ruler. The ability
4
My conceptualization differs from Slater’s (2003) view that highly institutionalized
regimes do not necessarily have to contain institutions that constrain leaders, as long
as the institutions have “infrastructural power” – the ability to implement and
execute the leader’s demands. By contrast, I argue that the ability to keep the
executive in check is a necessary and central component of institutionalization.
96
to survive beyond a particular leader is of particular importance in

authoritarian regimes. Elections in democracies serve as a reliable vehi-
cle in which future incumbents are selected without interrupting the
regime. The absence of such procedures in autocracy is often the greatest
weakness of dictatorships; how a regime can perpetuate itself beyond a
particular leader is not an automatically clear process. This challenge is
often exacerbated by the fact that many leaders give themselves the title
of “president for life” or claim to have godlike qualities, implying that
they are irreplaceable. Unsurprisingly, many leadership transitions
within authoritarian regimes are violent and irregular, and the majority
of regimes are unable to survive past the death or departure of the first
leader.
Institutionalized regimes resemble autonomous organizations, capable
of functioning regardless of which leader is in power. They tend to have
mechanisms in place to regulate leadership succession. Regimes with high
levels of institutionalization also have rules and autonomous organiza-
tional structures that ensure that the regime can function after the current
incumbent departs office. When we think about the quality of autocratic
regimes, the extent to which there are structures and procedures in place
to guard against personalist rule and perpetuate the survival of the ruling
coalition are of critical importance.
4.1.1 How Do Institutions Differ from Institutionalization?

While institutions are rules, institutionalization is the entrenchment of
those rules. Existing scholarship generally defines formal institutions as
rules that shape and constrain the behavior of political actors. In the first
sentence of his book, North (1990) defines institutions as “the rules of
the game in a society, or more formally, the humanly devised constraints
that shape human interaction” (3). According to North, “institutions
reduce uncertainty by providing a structure to everyday life” (3).
Following North’s definition, Lauth (2015) further notes that “[i]nstitu-
tions are a set of rules … The set of rules also creates and shapes social
order in such a way that the behavior of all actors involved in that social
order is predictable. Institutions affect performance by voluntarily
following the rules or being motivated by the threat of sanctions”
(Lauth 2015, 57).
However, as Huntington (1968) stresses, all institutions vary in their
degree of institutionalization. In their review of the literature on institu-
tional strength in comparative politics, Levitsky and Murillo (2009) stress
that institutional strength should be treated as a variable, rather than an
assumption, especially in the context of developing countries or
97
authoritarian states, where rule-making environments are less established.

They argue that “institutions are stable to the degree that they survive not
only the passage of time but also changes in the conditions – i.e., under-
lying power and preference distributions – under which they were initially
created and reproduced” (117). In terms of my conceptualization, institu-
tions are stable when they are institutionalized.
It is important to note here that while institutions are most commonly
defined as rules that constrain and shape behavior, another approach is to
think about institutions as equilibrium behavior. Knight (1992), for
example, treats self-enforcement as the defining criterion of institutions.
In other words, according to this approach, institutions are simply equili-
bria in which actors do not have any incentives to deviate – regardless of
whether the equilibrium is beneficial to the parties involved and regardless
of whether they are formally codified. I do not conceptualize institutiona-
lization in line with this school of thought. In fact, a regime that is, in
equilibrium, highly predatory but stable, is an example of an un-institu-
tionalized regime.5 This book focuses on institutions as rules and con-
ceptualizes regime institutionalization as the establishment of rules that
constrain the incumbent.
4.1.2 Can Informal Institutions Constrain?

Much of the literature on authoritarian institutions has focused either on
“parchment institutions” (Carey 2000) – formal rules that are written
down such as laws, regulations, constitutions – or on nominally demo-
cratic institutions such as parties, legislatures, or elections. But does
institutionalization have to occur on paper? Can informal rules and
norms constrain the behavior of executives? Helmke and Levitsky
(2004, 725) rightly emphasize that informal institutions have “remained
at the margins of the institutionalist turn in comparative politics,” even
though informal norms structure political behavior and are highly influ-
ential for political outcomes.
I follow Helmke and Levitsky’s definition of informal institutions as
“socially shared rules, usually unwritten, that are created, communicated,
and enforced outside of officially sanctioned channels” (727). This defini-
tion builds on North’s (1990) argument that norms can be considered
institutions in that they create rules that incentivize or constrain behavior.
They contrast this concept against formal institutions, which are defined
5
As a parallel example, a country might be stuck in an equilibrium where there are
high levels of corruption that operate in defiance of formal regulatory institutions,
but I would not consider corruption practices to be a set of institutions.
98
as “rules and procedures that are created communicated, and enforced

through channels widely accepted as official” (727).6
The rules of the game that structure political outcomes are often
informal – even in democracies – but this is especially the case in author-
itarian regimes where politics is often weakly institutionalized. A com-
monly cited example of a highly influential informal institution is the way
in which incumbents chose future presidents in twentieth-century Mexico.
Under the rule of the PRI, sitting presidents had the right to choose their
successors, and this practice was known as the dedazo (“big finger”).
From 1934 through 2000, this mechanism facilitated eleven regular lea-
dership turnovers (Langston 2006). Other common examples of informal
institutions include corruption and clientelism (see Hicken 2011 and
Treisman 2007 for surveys on these topics), although informal institutions
do not always produce negative outcomes. Tsai (2007), for instance,
illustrates how rural villages in China provide public goods where village
temple associations reinforce norms of social obligation.
When discussing informal institutions in developing and undemocratic
countries, we often think of clientelistic and patronage-based practices
that expand executive power. Presidents in Africa – especially in the
period immediately following independence – have frequently been
labeled patrimonial (or neo-patrimonial) leaders (Zolberg 1966).
Scholars agree that African politics is often driven by clientelistic practices
in which the president has complete authority over distributive decisions.
Leaders often use cabinet portfolios as patronage to buy elite support
(Arriola 2009), and they can even be used to coopt opposition politicians
(Arriola, DeVaro, and Meng 2018; Buckles and Gandhi 2017). Presidents
in many Latin American countries also have informal authority over state
institutions that lead to relatively unconstrained executives (O’Donnell
1994; Hartlyn 1994).
What often goes overlooked, however, is that informal institutions
can also be used to constrain executives. In fact, autocratic leaders are
often informally constrained in the extent to which they can monopolize
rents by their commitments to share resources with elites. This type of
informal constraint is especially effective when backed by credible
threats of deposition. I argue that regime institutionalization can indeed
be carried out via informal institutions. The primary function of institu-
tionalization in autocratic regimes is to constrain the executive, and the
mechanism of tying the leader’s hands can happen formally or
informally.
6
Formal rules include state institutions (parties, legislatures, and courts) as well as
state-enforced rules (constitutions and legislation).
99
In present day China, for instance, the appointment of the politburo

standing committee – the highest echelon of power within the ruling
Chinese Communist Party – has consistently followed three unwritten
norms. First, officials within the politburo and the politburo standing
committee retire by a set age limit. Officials are expected to retire by the
time they are sixty-eight, and generally do not receive appointments after
they turn sixty-seven. Minister-level officials are expected to retire at the
age of sixty-five, and lower-level officials are expected to retire at the age
of sixty. Second, officials who are appointed to the politburo standing
committee are routinely drawn from existing officials within the polit-
buro. Appointments follow a specific path to promotion. Third, among
eligible politburo members, appointments to the standing committee are
generally made on the basis of age (Miller 2017). Since appointment
are made according to these norms, the president faces limitations on
who he can appoint to the standing committee.
At the same time, as Thelen (1999a) cautions, norms and informal
institutions should “exert some independent power over individual beha-
vior” beyond simply reinforcing formal institutions (377). In Chapter 6,
I show that the appointment of stable vice presidents, prime ministers, and
defense ministers secures autocratic rule by lengthening incumbent
tenures and alleviating coup threats. Chapter 7 demonstrates that the
appointment of a de facto successor through the office of the vice
president/prime minister promotes peaceful leadership succession beyond
the existence of constitutional succession rules.
4.1.3 When Does Institutionalization Take Hold?

The reader may also be wondering when institutionalization actually
takes hold: when institutions are created or when they are tested? For
instance, if a leader implements leadership succession policies is the regime
institutionalized along this dimension when the constitutional rule is first
created? Or does the regime become institutionalized at the moment of the
leadership transition when the succession procedure follows the constitu-
tional rule?
I argue that institutionalization takes hold soon after the institutions
are created, since rules that credibly constrain leaders shift the distribution
of power in favor of elites.7 The longer these rules are kept in place, the
more time elites have to consolidate their authority by using the resources
7
Recall that in the formal model presented in Chapter 2, positive levels of institutio-
nalization increase the likelihood that the autocrat would be deposed if fighting were
to occur in the second period.
100
and influence afforded by positions of power. Moments where constitu-

tional rules are tested allow researchers to observe whether institutiona-
lization has actually taken place. As Chapter 6 will show, most instances
of leadership transitions do actually follow the procedures laid out in the
constitution (if such a rule existed). Chapters 6 and 7 will also discuss how
leaders are more likely to violate term limits, making such rules a weaker
form of regime institutionalization.
4.2 measuring regime institutionalization

Now that we’ve conceptualized regime institutionalization, how should
this variable be measured? Developing high-quality indicators of author-
itarian institutions poses some real challenges. Dictatorships are
frequently closed off, therefore restricting or completely eliminating
access to reliable and accurate information. Moreover, conventional
measures of institutional strength in democracies cannot simply be
imported to autocracies due to the lack of free and fair political competi-
tion. For example, while electoral results from presidential or legislative
elections can serve as a credible measure of incumbent or party strength
in democracies, the same approach cannot be imported to autocracies
because election results are often either falsified or do not reflect the true
preferences of citizens.
This section introduces my original dataset on regime institutionaliza-
tion in all sub-Saharan African countries, from 1960 to 2010. I present
measures of institutionalization that reflect the extent to which the regime
is autonomous from the leader. Since institutionalized regimes are those
that can outlive individual leaders, several of my measures focus on the
extent to which formal leadership succession policies exist. A regime with
clear and specific succession procedures represents a more institutiona-
lized organization because it has internal mechanisms to perpetuate itself
beyond the lifespan of a single leader. Another important dimension of
institutionalization is the extent to which the regime is able to function
independently of the leader. To gauge this dimension of institutionaliza-
tion, I present additional measures reflecting the extent to which other
elites fill key cabinet positions. Such cases where decision-making author-
ity is not concentrated entirely within the hands of the leader also repre-
sent regimes with higher levels of institutionalization. I collect historical
data on state constitutions and presidential cabinets, for which I have
comprehensive records for every country and year in my dataset. From
these records, I document the creation of executive constraints, hierarch-
ical positions, and implementation of rules and procedures that structure
the distribution of power.
101
4.2.1 Defining the Sample: What Qualifies as an

Authoritarian Regime?
Before I discuss my measurement strategy, it is useful to first define the
sample of countries that fall within the scope of this study. What qualifies
as an authoritarian regime? Many scholars have recently reevaluated this
question.8 There are two general camps. On one hand, the minimalist
definition of democracy requires political actors to acquire positions of
power through “competitive struggle for the people’s vote” (Shumpeter
1947, 269). Democracies are therefore defined through competitive
elections and regimes that fall short of this requirement are considered
autocracies. On the other hand, one can also consider a broader notion of
“democracy” by focusing not only on free and fair elections but also on
citizen access to impartial information, and the extent to which freedoms
(such as the freedom of speech or assembly) are upheld (Dahl 1972).
For my analysis in this book, I define an authoritarian regime as a country
in which executives do not come to power through competitive national
elections. For the coding of regimes, I follow Cheibub, Gandhi, and
Vreeland’s (2010) criteria. They define an autocracy as a country in which
any of the following criteria are violated: (1) the executive is selected either by
a popular vote or the election of a committee for the purposes of executive
selection (such as the Electoral College in the United States); (2) the legislative
is elected either directly or indirectly by popular election; (3) multiple poli-
tical parties are legally allowed to exist in the regime and the legislature; and
(4) there has been alternation in the party that is currently in power.
The last criterion, referred to as the “Botswana Rule,” is particularly
relevant when studying regimes within sub-Saharan Africa. As Przeworski et
al. (2000) and Cheibub, Gandhi, and Vreeland (2010) discuss, some regimes
may allow for multiparty elections only because they are confident that the
opposition will lose or because the opposition will not be allowed to take
office if they win. If a ruling party has never lost an election or has not
stepped down after losing an election, then it is difficult to ascertain whether
the regime is a true democracy or whether it is simply an autocracy that
allows for multiparty elections without the real intention of ceding power.
We can therefore only be certain that a regime will indeed allow for party
alternation in cases where the opposition is allowed to take office after
winning an election.
Within sub-Saharan Africa a number of countries led by ruling parties
have never experienced party alternation since the ruling party took power
after independence. Angola, for instance, has always been ruled by the
8
See Diamond 2002; Gandhi 2008; Geddes 2003; Pzeworski 2000; Svolik 2012.
102
Moviemento Popular de Libertacao de Angola (MPLA), despite holding

multiparty elections in 1992, 2008, and 2012. Although multiparty elections
were introduced in 1992, Tanzania has never had party alternation, and the
ruling Chama Cha Mapinduzi (CCM)9 has always won presidential elections
and legislative majorities. The Polity scores for Angola and Tanzania during
the period of multiparty elections remain low – ranging between −6 and −1.
In sum, simply having the appearance of multiparty elections is not sufficient
in guaranteeing that a country has democratized, especially when the incum-
bent and ruling party have never been tested by an electoral loss. Party
alternation is therefore a key criterion in defining regime type.
In addition to the criteria delineated by Cheibub, Gandhi and Vreeland
(2010) I further require that a country must be coded as a democracy for
over ten years in order to be considered a truly consolidated democracy.
There are a number of countries that have brief periods in which they are
coded as democracies by Cheibub, Gandhi, and Vreeland (2010), but then
revert back to authoritarianism. Ghana, for instance, is coded as a democ-
racy from 1969–1971 and 1979–1980, but was considered an autocracy
for long periods before and after those intervals. A truly consolidated
democracy should not be easily dismantled and requiring that countries
must be coded as democracies for over ten years eliminates countries that
only appear democratic for short periods of time.10
In sum, to derive my sample of authoritarian regimes I start with all
countries in sub-Saharan Africa. Countries enter the dataset either at the
year of independence or in 1960,11 and remain in the dataset either until
they democratize (following the criteria discussed above) or until 2010.
The final sample includes forty-six countries from 1960 to 2010.
Appendix Table 4.1 lists all the countries, leaders, and time periods that
qualify as authoritarian regimes and therefore are included in this study.
An alternative way of creating my sample of cases would have been to
utilize Polity scores (Marshall, Jaggers, and Gurr 2011) and drop cases that
fall below a certain numerical threshold. This procedure more closely reflects
the Dahl conceptualization of democracy in which the quality of political
9
The CCM was created in 1977 after the former ruling party, the Tanganyika
African National Union (TANU), merged with the ruling party of Zanzibar, follow-
ing the merger of the two countries.
10
Using this coding scheme, the following countries and time periods are considered
democratic and therefore dropped from the sample: Benin after 1991, Cape Verde
after 1990, Ghana after 1993, Kenya after 1998, Madagascar after 1993, Malawi
after 1994, Mali after 1992, Mauritius for the whole period, Nigeria after 1999, Sao
Tome and Principe after 1991, Senegal after 2000, and Sierra Leone after 1998.
11
Only Ghana and Guinea gained independence a few years prior to 1960. Liberia and
Ethiopia were never colonized. Each of these three countries enter the dataset in 1960.
103
competition and institutions are taken into account. However, I do not base
my coding of authoritarian regimes on Polity scores for three main reasons.
First, the vast majority of countries in Africa do not meet the minimalist
definition of contestation of power through competitive elections, at least
not until the 1990s. The “less stringent” criterion is sufficient to identify the
authoritarian sample of cases in this region. Second, Polity uses criteria
about the quality of executive constraints and competitiveness to create an
aggregated score.12 Since I am examining constraints on executive power, I
do not want to eliminate cases by using a coding scheme that takes into
account the quality of institutions and constraints on the executive. Doing
so would risk eliminating cases of highly institutionalized authoritarian
regimes in my final sample. Third, Polity scores do not reflect party alter-
nation; therefore, a case such as Botswana can score exceptionally high
despite not having had party alternation in the executive.
4.2.2 The Data

I define institutionalization as the creation of hierarchies, rules, and proce-
dures that structure the distribution of power and resources within the ruling
coalition. An important implication of institutionalization within authoritar-
ian regimes is that it functions to constrain the leader’s ability to make
arbitrary decisions. In order to operationalize this concept, I create measures
that focus on the extent to which the leader’s hands are tied. Such constraints
can take the form of formal constitutional rules that explicitly limit the leader’s
authority. In particular, I focus on the creation of constitutional leadership
succession procedures that dictate how a peaceful transition of power should
occur, thereby prolonging the tenure of the regime beyond a single leader.
I also focus on the establishment of clear hierarchical positions by
examining the extent to which the president is willing to distribute impor-
tant cabinet positions (such as vice president, prime minister, and defense
minister) to other elites. The creation of stable cabinet appointments limits
the incumbent’s power by placing other elites in a position of power or
influence within the presidential cabinet. It also reflects the process of
institutionalization by creating an independent hierarchy and decision-
making apparatus separate from the leader. In sum, these measures pro-
mote the creation of an independent functioning organization and also
serve to limit the authority of a single leader.
12
One of the component variables of Polity scores is “XCONST,” which reflects the
extent to which there are institutionalized constraints on the decision-making
powers of the executive. A comparison of my measures of executive constraints
and XCONST is discussed later in this chapter.
104
4.2.3 Formal Institutions: Autocratic Constitutions

To construct my measures of regime institutionalization, I utilize the
Europa World Year Book (1960–2010), which has yearly records of all
executive posts, ministerial positions, and constitutions for all countries in
sub-Saharan Africa. First, I examine the creation of leadership succession
rules within constitutions. Scholars have identified constitutions as useful
mechanisms that can establish formal rules and procedures in a public
way that holds autocrats accountable (Albertus and Menaldo 2012;
Ginsburg and Simpser 2013). In fact, autocratic constitutions are also a
type of “focal point” for elites, creating a universally observed set of rules
and enforcement behavior surrounding the incumbent.
For every country–year observation, I document whether the constitu-
tion had an amendment outlining procedures governing presidential
succession. The Kenyan constitution, which has been in effect since inde-
pendence in 1963, includes specific instructions for succession. Chapter II,
Part I, Section 6 reads:

resignation of the President … an election of a President shall be held
The transfer of power from Jomo Kenyatta to Daniel arap Moi during
the 1978 presidential succession in Kenya illustrates the importance of
succession rules. Near the end of Kenyatta’s rule, a faction within the
ruling KANU party tried to contest the authority of then-vice president
Moi on the grounds that he was not a member of the dominant ethnic
group. Moi and his supporters were able to effectively dispute their claims
by utilizing the policy governing presidential succession outlined in the
constitution (Tamarkin, 1979, 21–26).
Beyond this particular case, constitutional succession rules play a signifi-
cant role in promoting successful leadership transitions. In the chapter 7,
I show that regimes with constitutions that outline specific succession pro-
cedures are significantly more likely to undergo peaceful leadership succes-
sion. Furthermore, once in place, succession rules are not easily overturned.
In my sample, conditional on creating a constitutional amendment governing
the rules of succession, 74 percent of these procedures have been left intact.
105
If succession rules do exist, I also distinguish whether the procedures

specify who would succeed the president, rather than simply providing
vague rules about nominating a successor. Unlike the Kenyan constitu-
tion, which specifies that the vice president should become the interim
president, the constitution of Angola provides guidelines that are much
vaguer. Article 33 of the constitution states: “In case of the death, resig-
nation, or permanent impediment of the President of the Republic, the
Council of the Revolution will designate from among its members
the person who will exercise temporarily the charge of President of the
Republic” (Blaustein and Flanz 1976, 6).
In addition to succession rules, I examine the creation of constitutional
term limits. Scholars have increasingly pointed to term limits as a key type
of executive constraint in authoritarian regimes, particularly in African
states after the end of the Cold War (Posner and Young 2007). According
to Svolik (2012), “a term limit on a leader’s tenure therefore amounts to a
line in the sand: Compliance is easily and publicly observable. Term limits
thus both embody a compromise about the limited authority of any single
leader and provide an unambiguous signal of commitment to such a
constraint” (198). Scholars and practitioners within sub-Saharan Africa
pay particular attention to the creation of term limits, arguing that “term
limits are a necessary bulwark against abuse of power, especially when
electoral systems are weak” (Louw-Vaudran 2016).
For every country–year observation, I record whether the constitution
includes term limits. The constitution of the Democratic Republic of
Congo, for instance, from 1978–1996 included a clause stating that the
President would be elected for a seven-year term, renewable only once.
Article 66 of the constitution of Ghana has included the following provi-
sion since 1993: “(1) A person elected as President shall … hold office for a
term of four years … (2) A person shall not be elected to hold office as
President of Ghana for more than two terms” (Peaslee and Peaslee Xydis
1974). Executives with formal term limits written into the constitution
represent leaders who are operating within more highly institutionalized
regimes.
4.2.4 Informal Institutions: Presidential Cabinets

In order to codify informal constraints on presidential power, I turn to the
appointment of key positions within the presidential cabinet. For every
country–year observation, I document the name of the president, the name
of the vice president, prime minister, and minister of defense if one had
been designated.
106
The appointment of a vice president (VP) or prime minister (PM)13 –

so-called “second-in-command” positions – represents the creation of a
hierarchical structure within the executive as well as a distribution of
authority to other elites. Designating a second-in-command, a very visible
national position, casts another elite as a potential focus of power and
possible successor to the executive. In fact, we can even think of the act of
naming a second-in-command as the autocrat solving the collective action
problem for other elites as an alternative leader to rally around.14 The
creation of these positions, therefore, reduces reliance on a single autocrat
and promotes the survival of the regime past the tenure of the first leader.
Moreover, within African presidential cabinets, most presidential
successors were former vice presidents or prime ministers. Under the
Cameroon Union (UC) in Cameroon, for instance, Paul Biya had been
the prime minister in 1975 under the presidency of Ahmadou Ahidjo
before becoming president himself in 1982. To verify that this was indeed
the larger pattern in my data, I code the previous positions of presidential
successors. The data reveals that the position of vice president or prime
minister is often a stepping stone for the presidency. Table 4.1 provides a
full summary of previous positions held by presidents in my sample,
conditional on a peaceful transfer of power. The data show that 23
percent of all former presidents held the position of vice president or
prime minister. Minister of defense, the state/interior, and finance are
also commonly held portfolios by elites who later became the incumbent.
For successions that were peaceful, the vast majority of successors were
chosen within the president’s cabinet.
I also analyze past positions of incumbents who came into power via
irregular means. Table 4.2 provides a full summary of previous positions
held by presidents in my sample, conditional on a nonpeaceful transfer of
power. Most of the leaders in this sample came into power via a coup.
Unsurprisingly, these leaders commonly had military backgrounds. Even
for the leaders in this sample who were in the presidential cabinet before
13
Vice presidents and prime ministers are functional equivalents within this particular
context. All the countries in my sample either have one or the other. No cabinet has
both a vice president and a prime minister, although a small number of countries
have multiple vice presidents.
14
In fact, many presidents were extremely hesitant to assign the second-in-command
position to another party elite. Jose Eduardo dos Santos, the first president of
Angola, named himself as his own vice president for a number of years while he
was in power. Felix Houphouet-Boigny, the first president of the Ivory Coast,
created a vice-president position, but kept the post empty the entire time he was in
power. These two cases (and other similar instances) are coded as not having the
second-in-command position filled by another elite.
107
Table 4.1 Previous positions of autocratic leaders (conditional on peaceful

leader entry)

Minister of defense/security 15 14%
Minister of state/ interior 12 11%
Minister of finance 11 9%
Minister of foreign affairs 6 5%
Minister of natural resources 6 5%
Minister of development 5 5%
Minister of health 3 3%
Minister of education 3 3%
Miscellaneous portfolios 18 16%

Legislature 5 26%
Judicial system 5 26%
Military 5 26%
Opposition 3 16%
Unions 1 5%
Notes: The leaders included in this table entered office through regular means, so coups and
irregular leader entries are excluded from this sample. The first president of each country is
not included in this sample because I am interested in the path of promotion after
independence was granted. Only leaders who were in office for three years or longer are
included in this sample. Leaders sometimes have multiple positions before entering office.
For example, Festus Mogae, who was the president of Botswana from 1998 to 2007 had
multiple cabinet positions from 1990 until 1997. He was the minister of finance from 1990
to 1997, minister of development and planning from 1990 to 1997, and vice president from
1992 to 1997. Each of these portfolios are recorded in this table.
their ascension to power, the most frequently held positions were the
minister of defense, vice president, or prime minister.
In addition, I document whether the person in the vice president or
prime minister position remains fairly constant over time, or whether the
person in this position is rotated very frequently. Accounting for cabinet
rotation rates is particularly important in the context of elite politics in
sub-Saharan Africa. African leaders routinely practiced the “revolving
108
Table 4.2 Previous positions of autocratic leaders (conditional on irregular

leader entry)

Military 41 93%
Opposition 3 7%

Minister of defense 8 44%
Minister of the interior 1 6%
Misc. minor portfolios 4 22%
Notes: The leaders included in this table entered office through irregular means – the modal
method of irregular leader entry is via a coup. The first president of each country is not included
in this sample because I am interested in the path of promotion after independence was granted.
Only leaders who were in office for three years or longer are included in this sample.
door” policy of constantly rotating important cabinet ministers in order to

prevent any one person from amassing too much power (Dickie and Rake
1973; Jackson and Rosberg 1982; Hassan 2017). Moreover, since vice
presidents are largely perceived to be de facto successors, presidents often
strategically rotate those in the second-in-command position to prevent
any particular elite from amassing too much power or influence. In order
to account for strategic rotation, I create a variable that measures the
stability of key appointments.
An example is helpful here. Consider Seretse Khama, who was the first
president of Botswana from 1966 to 1980. Quett Masire was appointed as
the vice president throughout Khama’s entire tenure, and in fact succeeded
Khama to become the next president. In this case, Masire had a stable vice
president for fourteen years during his tenure. On the other hand, Felix
Houphouet-Boigny, the first president of Cote d’Ivoire from 1960 to 1991,
did not appoint a vice president or prime minister the entire time he was in
office. Finally, Idriss Deby, who has been the president of Chad since 1991,
has been rotating his vice presidents practically on a yearly basis since
taking power. From 1991 to 2005, Deby named ten different vice
presidents. There have only been five years for which the vice presidency
was held by the same person as the previous year during Deby’s tenure.
109
In the context of my sample, if a president keeps the same person in the

vice president position for longer periods of time, this would be inter-
preted as a higher degree of de facto institutionalization compared with a
president who rotates the person in the vice president position every year.
In order to determine the stability of the second-in-command position, I
create a dummy variable that takes a value of 1 if the person in the vice
president or prime minister position is the same person as the previous
year. Cases where this is more frequently true can be interpreted as having
higher degrees of institutionalization.
Additionally, I document whether a minister of defense was appointed
for every country–year observation. The defense portfolio is an especially
important ministerial position because it represents control of military
force. As such, presidents were extremely reluctant to appoint other elites
as the defense minister. In fact, the defense portfolio is the most commonly
kept portfolio by the presidents themselves. Whether the president dele-
gates this position to someone other than himself is therefore a key
component of institutionalized power sharing. The distribution of this
portfolio has significant consequences. In Chapter 6, I show that having
an independent minister of defense significantly lowers the risk of coup
attempts.
Finally, I consider the extent to which the person appointed as the
minister of defense was rotated on a yearly basis. Similar to the second-
in-command position, I create a dummy variable that takes a value of 1 if
the person appointed to be the defense minister was the same person as the
year before. Cases with lower levels of rotation can be interpreted as
having higher degrees of institutionalization.
It is important to note that even when leaders have the ability to choose
who they appoint to them key positions – as they often do – the act of
delegating authority to any other elite serves as a constraint on the leader.
As outlined in Chapter 2, when elites are appointed to influential minis-
terial positions – and especially when the chain of command is formalized
in a constitutional succession order – this results in a shift in power away
from the leader in favor of the elite. This is true even if the elite was a
former ally of the leader. In fact, existing research shows that “allies” tend
to pose the most dangerous threats to autocratic leaders! European mon-
archs, for instance, were most frequently overthrown by their own family
members (Herz 1957; Kurrild-Klitgaard 2000; Kokkoknen and Sundell
2014; Tullock 1987) and African leaders are often deposed by their own
coethnics (Roessler 2016). In sum, the appointment of any elite to a
position of influence within the presidential cabinet constrains the leader
by empowering that particular elite with state resources and the visibility
of a position of power.
110

that can be verified and replicated. Rather than hand-coding cases accord-
ing to subjective criteria, this data was constructed by simply observing
whether certain constitutional amendments or cabinet appointments
existed, therefore the coding scheme did not require the researcher to
make judgment calls. In addition, the indicators distinguish between de
facto and de jure forms of institutionalization and allow us to disentangle
the regime from the leader. Another advantage is that this data is presented
as disaggregated indicators. Researchers can therefore decide which dimen-
sions of institutionalization are most relevant to the research question at
hand. Alternatively, the researcher can also choose to aggregate certain
indicators into an institutionalization “score,” although they are not obli-
gated to do so since the measures are presented separately.
To summarize, my dataset produces the following seven indicators that
serve as measures of regime institutionalization. Each variable is coded as
country–year units and takes the form of a dummy variable.
1. Successor policy: is there a constitutional amendment specifying the
rules of succession?
2. Successor (strict): is there a constitutional amendment specifying
exactly who would succeed the president in the case of his death?
3. Term limit: are there constitutional term limits?
4. VP/PM appointed: was a vice president or prime minister appointed?
5. VP/PM same: was the person appointed to the vice president or prime
minister position the same person as the year before?
6. Defense appointed: was a minister of defense appointed?
7. Defense same: was the person appointed as the minister of defense the
same person as the year before?
4.3 e xecutive constraints i n s ub-saharan

africa: general trends and p atterns
What does the empirical data on executive constraints within sub-Sharan
Africa look like? This section provides a descriptive summary of the data
and highlights some general trends and patterns both across countries and
over time.
Figure 4.1 displays cross-sectional differences in institutionalization
across all the countries in my sample.15 Each section of the pie chart
15
Because a number of the variables are highly correlated with each other (as I will
discuss below), I only include successor policy, term limits, VP/PM same, and
defense same in these graphs.
111
Angola Benin Botswana Burkina Faso Burundi
Cameroon Cape Verde CAR Chad Comoros
Congo DRC Ivory Coast Djibouti Eq. Guinea Eritrea
Ethiopia Gabon Gambia Ghana Guinea
Guinea Bissau Kenya Lesotho Liberia Madagascar
Number of constraints 0 1 2 3 4
Figure 4.1.a Cross-sectional differences in institutionalization
illustrates the proportion of years for which each country had one, two,
three, or all four dimensions of institutionalization in place. Black sections
represent the proportion of years for which the country had all four
dimensions in place. Dark grey sections represent the proportion of
years for which the countries had three dimensions in place, and so on.
White sections represent the portion of years for which the country had no
dimensions in place.
112
Malawi Mali Mauritania Mozambique Namibia
Niger Nigeria Rep. of Congo Rwanda Sao Tome
Senegal Sierra Leone Somalia South Africa Sudan
Swaziland Tanzania Togo Uganda Zambia
Zimbabwe
Figure 4.1.b Cross-sectional differences in institutionalization
Countries that have darker circles can be interpreted as being more

institutionalized on average, although these graphs do not take into
account change over time.16 Since these graphs show proportion of
years, the reader should be reminded that some countries gained indepen-
dence earlier than others, so the length of time countries are included in
the sample sometimes differs. Over 70 percent of the countries in my
sample either gained independence prior to 1970 or were never colonized
and such countries enter my dataset in the 1960s. A small number of
16
Appendix Table 4.2 includes a summary of institutionalization scores by country.
113
countries gained independence later – for instance, Zimbabwe did not

gain independence until 1980 and therefore has fewer years included in
the dataset.
It is clear from the graphs that countries across sub-Saharan Africa
differ widely in terms of level of institutionalization. Namibia, for
instance, is clearly a case with high levels of institutionalization. Since
gaining independence in 1990, Namibia has had formal succession rules
and term limits in the constitution without interruption since the start of
the regime. A vice president and minister of defense have been appointed
in the presidential cabinet every year, and these appointments have been
extremely stable. A constitutional leadership succession was carried out
peacefully in 2005 when the first president of Namibia, Sam Nujoma,
handed power to his successor, Hifikepunye Pohamba.
Contrast Namibia with Togo, a country that has had much lower levels
of institutionalization. Togo, which gained independence from France in
1960, has never had constitutional term limits, and formal succession
policies were in place only after 1992. No elites were appointed to the
vice president or minister of defense position for the first three decades
after independence. When elites were appointed to these cabinet positions
after the end of the Cold War, they were heavily rotated and replaced
every two to three years.
Besides cross-sectional differences, levels of regime institutionalization
have also changed over time. Figure 4.2 illustrates the proportion of
countries with executive constraints graphed over time.
The trends documented in this graph suggest that various dimensions
of institutionalization seem to rise together. Countries also appear to
become increasingly more institutionalized over time, although it’s impor-
tant to note that by the end of the data, 2010, only about half of countries
in sub-Saharan Africa have formal or informal executive constraints.
There is a sharp increase in the prevalence of term limits in the early
1990s, and in general, more countries became institutionalized after the
end of the Cold War.
Returning to the case of Togo, even though the country was almost
entirely un-institutionalized from 1960 to 1990, it has become more
institutionalized in recent decades. Following the end of the Cold War,
formal succession policies were created under then-president Eyadema
Gnassingbe, who also began to appoint elites to the vice president and
minister of defense cabinet positions. The regime in Togo today is much
more institutionalized that it was in previous decades.
How do various dimensions of institutionalization relate to each other?
Do the measures move together and therefore act as complements to each
other? Or do certain types of executive constraints substitute for other
114
1
.9
.8
.7
.6
.5
.4
.3
.2
.1
0
1960 1970 1980 1990 2000 2010
Year

Figure 4.2 Over time changes in institutionalization
types of constraints? Table 4.3 provides a correlation matrix of the seven

measures.
The first key takeaway is that no two measures are negatively corre-
lated, so there does not appear to be evidence of substitution between
different dimensions of institutionalization. In other words, it appears
unlikely that a leader would choose not to appoint a minister of defense
because she has appointed a vice president.
This observation is, in itself, an interesting finding because it remains an
open question whether informal institutions compete with or strengthen
existing formal institutions. Scholars have noted that when informal insti-
tutions arise, they can often undermine or weaken formal institutions
(Gryzmala-Busse 2010; Borocz 2000b). Corruption serves as such an
example. Grzymala-Busse argues that “informal institutions serve as tem-
plates and substitutes for formal choices. Even as they compensate for
formal institutions, informal rules can subvert them. Informal institutions
also reify formal rules by defining and expanding their domain, and by
providing incentives and information to follow formal institutions” (311).
On the other hand, informal institutions can also help strengthen formal
institutions. “Informal rules may generate precedents and prevalent prac-
tices that are then formalized for efficiency’s sake” (Gryzmala-Busse 2010,
321). They can support and reify formal institutions by “delineating the
domains where formal institutions rule – and by providing information and
enforcement that promote the functioning of formal institutions” (324).
115
Table 4.3 Correlation matrix of seven dimensions of institutionalization
Successor Successor Term VP/PM VP/PM Defense Defense

policy (strict) limits appoint same appoint same
Successor policy 1
Successor (strict) 0.855 1
Term limits 0.413 0.424 1
VP/PM appoint 0.187 0.238 0.292 1
VP/PM same 0.159 0.166 0.192 0.618 1
Defense appoint 0.209 0.221 0.186 0.203 0.122 1
Defense same 0.199 0.190 0.120 0.134 0.239 0.644 1
Note: Bolded numbers indicate two dimensions that are highly correlated (r > 0.6)
A second important takeaway is that some measures are highly corre-

lated with each other. Having succession rules is highly correlated with
having strict succession rules (r = 0.855), appointing the same VP/PM is
highly correlated with appointing a VP/PM (r = 0.618), and appointing
the same defense minister is highly correlated with appointing a defense
minister (r = 0.644). This is not surprising since each of these variables is
just a stricter version of the corresponding variable. In much of my
empirical analysis, I exclude measures that are highly correlated with
another indicator of institutionalization.
4.4 regime typologies

How do my measures of regime institutionalization compare with other
existing datasets of authoritarian institutions? Autocratic regime institu-
tionalization is difficult to operationalize, and there are not many existing
datasets that measure this concept directly. Perhaps the most direct way in
which regime institutionalization has been codified is through the regime
typologies framework.
In a seminal study, Geddes (1999a) classifies all autocratic regimes into
one of the following regime types: military, single-party, personalist, or
hybrids of these categories. These classifications are based on whether
control over “policy, leadership selection, and the security apparatus is in
the hands of a ruling party (dominant-party dictatorships), a royal family
(monarchy), the military (rule by the military institution), or a narrow
group centered around an individual dictator (personalist dictatorship)”
(Geddes, Wright, and Frantz 2014, p. 318).
116
Single-party regimes are defined as regimes in which the “party has

some influence over policy, controls most access to political power and
government jobs, and has functioning local-level organizations” (Geddes
1999a, 31). By contrast, in personalist regimes, leaders have “consoli-
dated control over policy and recruitment in his own hands, in the process
marginalizing other officers’ influence and/or reducing the influence and
functions of the party” (Geddes 1999a, 31). Military regimes are defined
as those “governed by an officer or retired officer, with the support of the
military establishment and some routine mechanism for high level officers
to influence policy choice and appointments” (Geddes 1999a, 31).
Geddes’ study and associated dataset have made immense contributions
to scholarship on authoritarian politics. It set the agenda for renewed
interest in the study of nondemocratic regimes outside of the industrialized
world and stimulated a large body of recent work on the policies, institu-
tions, and consequences of autocratic rule.17 However, the use of typologies
as an indicator for regime institutionalization has resulted in four central
problems – two theoretical and two measurement driven.
First, placing regimes into time-invariant categories obscures institu-
tional changes over time. This is a significant limitation because leaders
within the same regime often make very different decisions about the
extent to which they implement executive constraints. Founding leaders,
for instance, are often much more personalist than their successors
(Bienen and van de Walle 1989). Regimes tend to become less personalist
over time, especially after multiple leadership transitions.
Take the cases of Mexico under the rule of the PRI and China – perhaps
two of the most commonly cited examples of highly institutionalized
dictatorships. They are both coded as party-based regimes for the entire
duration of the party’s tenure.18 Although these regimes eventually
became strong and autonomous organizations, they were significantly
less institutionalized under their first leaders. The PRI under Plutarco
Elias Calles and the CCP under Mao Zedong resembled personalist
regimes, where each respective leader had consolidated control, reducing
the parties’ influence. Calles, the founder of the PRI,19 controlled the
actions and decisions of three presidents after him during what is referred
to as the Maximato Period. It was not until the fourth president, Lázaro
Cárdenas, that the regime established real leadership turnover norms
17
According to Google Scholar, Geddes (1999a) has been cited almost 300 times.
Geddes (1999b), an annual review article covering the same material as Geddes
(1999a), has been cited over 1,500 times.
18
The PRI was in power from 1929–2000 and the CCP has been in power since 1949.
19
Known as the Partido Nacional Revolucionario (PNR) at the time.
117
where power was handed over to the successor after the president’s term
ended (Buchenau 2006).
Mao was a similarly personalist founding leader. He promoted and
purged elite politicians at will and launched political campaigns, such as
the Hundred Flowers Movements and Cultural Revolution, during which
he arbitrarily purged officials who he deemed critical of the regime
(Meisner 1986). Over time, as Deng Xiaoping came into power, the CCP
eventually became an organization in which party promotion followed
established norms and procedures. In sum, many regimes change drastically
over time, and this is impossible to capture using a single time-invariant
regime category.20
Second, categorical distinctions between authoritarian regime types are
often insufficiently nuanced and obscure variation in institutional strength
within regime types. Take the category of dominant-party regimes. The
Soviet Union and China are coded as dominant-party regime types, and
they indeed were ruled by strong parties with organizational autonomy that
lasted through several leadership transitions. However, the category of
dominant-party regimes also includes cases such as the PDG in Guinea or
Communist Party of Kampuchea (CPK, also known as the Khmer Rouge)
in Cambodia. These parties, unlike the highly institutionalized communist
parties, were entirely concentrated around a single charismatic leader and
were not able to survive past the death and ousting of the leader.21
Third, regime typologies are composite indices that aggregate various
dimensions of leaders, institutions, and military structures into a single
category. As a result, it is difficult for researchers to identify the effects of
individual institutions (Pande and Udry 2006). Some regimes, for
instance, appear to be party-based, when in actuality the party is attached
to a strong and charismatic leader who merely exploits the party as a
personal vehicle to amplify his authority.
The Union Soudanaise – Rassemblement Democratique Africain (US-
RDA) under Modibo Keita in Mali, for instance, is coded as part of a
dominant-party regime. Yet national policies were determined entirely at
the discretion of Keita alone, and the US-RDA lacked institutionalized
rules and permanent structures. Eight years after taking power, Keita was
deposed in a coup and the party was banned. Although Keita, who was
20
It is possible for countries to be coded as different regime types over time, but most
countries are not coded as very many different types. The average country is coded
as two different types of regimes over an average time span of forty-four years, and
the average regime coding lasts about eighteen years.
21
The opposite scenario is sometimes true as well: not all cases coded as military or
personalist regimes are equally unconstrained.
118
also a self-proclaimed socialist, portrayed Mali as a one-party state, the

ruling party was actually extremely weak.
Fourth, although Geddes outlines a clear set of guidelines that were
employed to categorize regimes into different categories, a number of
these criteria used to code regime types require the researcher to make
subjective decisions about how to code the regime. Examples of subjective
criteria include: “Does the party have functioning local-level organiza-
tions that do something reasonably important, such as distribute seeds or
credit or organize local government?” or “has rule of law been main-
tained?”. The possibility of measurement error based on subjective coding
rules is also heightened by the fact that the dataset spans multiple regions
and time periods and often relies on information from various country
experts. An example of criteria that relies on different country sources
includes: “Does the country specialist literature describe the politburo-
equivalent as a rubber stamp for the leader?” Different country experts
from various regions may have different standards for evaluating institu-
tions making it difficult to know whether the criteria are being applied
uniformly. Moreover, it is not clear what the individual responses to these
criteria are, nor how the responses to these criteria are aggregated to
produce single regime categories.
4.4.1 Comparison of Institutionalization and Regime Typologies

In this section, I compare my indicators of regime institutionalization with
data on regime typologies. I find that there are a number of discrepancies
between my measures of institutionalization and regime typologies. Many
regimes that have been coded as dominant-party regimes are actually not
very institutionalized at all, and some regimes that have been coded as part
of personalist or military regimes actually have a number of executive
constraints in place.
The dataset that was introduced in Geddes (1999a) was updated and
rereleased by Geddes, Wright, and Frantz (2014). I will refer to the updated
dataset as the GWF dataset for the remainder of this chapter. GWF covers
all country–years with autocratic governments between 1946 and 2010 in
independent countries. From this country–year data, GWF creates a list of
280 autocratic regimes with their start and end dates, as detailed by their
codebook. Similar to the earlier regime datasets, GWF classifies all auto-
cratic regimes in their dataset into one of the following regime types:
monarchy, personal, military, party, party-personal, party-military, mili-
tary-personal, party-personal-military, oligarchy, indirect military.
Since GWF is cross-sectional, I collapse the country–year panel data
from my Africa sample into regime-level observations in order for my
119
dataset to be comparable with the GWF regime-level observations. The

resulting cross-sectional dataset has eighty-three regime-level observa-
tions.22 Since my original measures of regime institutionalization are
dummy variables, to transform them into cross-sectional indicators I
calculate the percentage of years for which a dimension of institutionali-
zation was implemented. For instance, a leader that was in power for
twenty years and had a vice president for ten of those years would score a
.50 on that particular dimension of institutionalization.23
Table 4.4 provides a comparison between my measures of institutio-
nalization and regime type for the Africa subsample. Shaded cells
represent disagreement between the assigned regime type and institutio-
nalization score on that particular dimension. For instance, a number of
regimes that are coded as party-based had a vice president or prime
minister less than 70 percent of the time in which the regime was in
power. This disagreement is highlighted by having shaded cells under
the dimension “second named” for those parties. In addition, a number
of regimes that are coded as military or personalist actually have a vice
president or prime minister more than 70 percent of the shaded cells.
A comparison of GWF and my measures of institutionalization reveals
a high number of discrepancies. Out of eighty-three regimes in Table 4.4,
only fifteen do not have any shaded cells. In other words, 82 percent of the
observations have an inconsistency between the assigned regime typology
and at least one dimension of institutionalization. This finding is consis-
tent with other studies that argue that the GWF typology better reflects the
later years of party institutionalization but is a poor representation of
earlier years (Lucardi 2017).
The comparison reveals that many regimes that are coded as dominant-
party by GWF are actually not very institutionalized at all. Mali under the
rule of the Sudanese Union scores zeros for all seven dimensions of party
institutionalization, making it the least institutionalized regime in the
dataset. Niger under the rule of the Nigeria Progressive Party scores
zeros on five out of seven dimensions. Party-based regimes in the
Gambia, Guinea, Guinea Bissau, and Madagascar each score zeros on
22
I only focus on comparisons of pure regime types, leaving out hybrid regime types, such
as party-personal, party-military, military-personal, or triple hybrid types. These types
of hybrid regimes are excluded because it is not clear whether these regimes are highly
institutionalized or not. For instance, should a researcher expect a party-personal
regime to be highly institutionalized (due to the presence of some party attributes) or
not institutionalized (due to the presence of some personalist attributes)?
23
X
years in power
120
four out of six dimensions of institutionalization. In sum, many regimes

that are coded as dominant-party perform very poorly when we examine
their institutionalization scores.
A closer examination of the cases that exhibit much disagreement also
highlights the ambiguity of the GWF coding scheme. Take the example of
Niger from 1960 until 1974. The Nigerian Progressive Party scored zeros
Table 4.4 Comparison between institutionalization and regime type
Party-based regimes and monarchies <50% <50% <50% <70% <50% <70% <50%
(institutionalized)
Angola 75–NA Party
Botswana 66–NA Party
Ethiopia 91–NA Party
Gambia 65–94 Party
Guinea 58–84 Party
G. Bissau 74–80 Party
Ivory Coast 60–99 Party
Kenya 63–02 Party
Madagascar 60–72 Party
Mali 60–68 Party
Mozambique 75–NA Party
Namibia 90–NA Party
Niger 60–74 Party
Rwanda 62–73 Party
Senegal 60–00 Party
Sierra Leone 68–92 Party
Tanzania 64–NA Party
Zambia 67–91 Party
Zambia 96–NA Party
Zimbabwe 80–NA Party
Ethiopia 89–74 Monarchy
Swaziland 68–NA Monarchy
Burkina Faso 87–NA Personal
Cameroon 83–NA Personal
CAR 03–NA Personal
Chad 82–90 Personal
Chad 90–NA Personal
Rep Congo 60–64 Personal
Rep Congo 97–NA Personal
DRC 60–97 Personal
DRC 97–NA Personal
Gambia 94–NA Personal
Ghana 81–00 Personal
Ivory Coast 00–NA Personal
Ivory Coast 99–00 Personal
121
Table 4.4 (continued)
Madagascar 09–NA Personal
Madagascar 75–93 Personal
Malawi 64–94 Personal
Mali 68–91 Personal
Mauritania 08–NA Personal
Niger 96–99 Personal
Sierra Leone 97–98 Personal
Somalia 69–91 Personal
Sudan 69–85 Personal
Sudan 89–NA Personal
Togo 60–63 Personal
Togo 63–NA Personal
Uganda 86–NA Personal
Burkina Faso 80–82 Military
Burundi 87–93 Military
Chad 75–79 Military
Madagascar 72–75 Military
Mauritania 05–07 Military
Sierra Leone 67–68 Military
Note: Shaded cells represent dimensions of party institutionalization that are in disagreement with the regime typology
coding. For example, for party-based regimes, “Second Appoint < 70%” means that the regimes listed in this column
are coded as party-based regimes, but have a second in command appointed less than 70 percent of the time.
For non party-based regimes, “Second Appoint >80%” means that the regimes listed in this column are NOT coded
as part of party-based regimes, but have a second in command appointed more than 80 percent of the time.
NA refers to regimes that were still in power as of 2010.
on six out of seven dimensions of institutionalization, making it one of the

weakest regimes in my sample, yet it is coded as part of a party-based
regime by GWF. Even when we look at the case description in the GWF
codebook, the leader does not appear to constrained at all.
Niger (1960–74)
Start: 8/3/1960 Independence under single-party rule. The PPN won the December
1958 preindependence election with help from the French. Prior to independence,
the most popular rival party was outlawed and its leaders jailed. Power was
centralized under Diori, who controlled ministerial appointments without parlia-
mentary scrutiny, could appoint and dismiss civil servants and military officers,
and could decree and veto laws […]. End: 4/15/1974 Coup led by the Army Chief
122
of Staff ousted the civilian government[…] . (Geddes, Wright, and Frantz 2012,
82)
It is difficult to know why the PPN in Niger was coded as part of a

party-based regime, rather than a personalist regime (or at least a party-
personal regime), in part because regime typologies are aggregate mea-
sures. The advantage of disaggregated indicators of institutionalization is
that the researcher can clearly observe the extent to which the regime
fulfills certain requirements (such as having leadership succession rules)
and make decisions based on transparent criteria.
The comparison of the two datasets also reveals that a number of
regimes that were coded as military or personalist are actually quite
institutionalized. Several regimes that are coded as personalist actually
exceed the threshold of institutionalization for many dominant-party
regimes. Although Burkina Faso, under the presidency of Blaise
Compaore, is coded as a personalist regime, it had an appointed prime
minister and minister of defense for the majority of Compaore’s tenure.24
The constitution of Burkina Faso also had succession procedures and term
limits during most of this period.
4. 5 comparisons with other e xisting datasets

The final section of this chapter presents graphical comparisons of my
institutionalization measures against two other commonly used datasets
of institutional strength – Polity (Marshall, Gurr, and Jaggers 2016) and
V-Dem (Coppedge et Al. 2016). The comparisons show that while exist-
ing variables can provide broad impressions of institutional strength, they
are often not sufficiently nuanced to accurately capture disaggregated
dimensions of executive constraints.
First I compare my disaggregated indicators of executive constraints
against the “XCONST” variable from the Polity dataset. “XCONST” is
described as a variable that
refers to the extent of institutionalized constraints on the decision- making powers
of chief executives, whether individuals or collectivities. 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 advisors 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. (Marshall, Gurr and Jaggers 2016, 24)
24
Recall that the data goes up until 2010, though Compaore was in power until 2014.
123
Proportion of observations with constraint (author’s data) 1.00
0.75
Type of constraint
Succession rules
Term limits
0.50 Stable VP appoint
0.25
1 2 3 4 5 6 7
POLITY "XCONST" score
Figure 4.3 Comparison between institutionalization and XCONST variable
“XCONST” is coded on a seven-point scale, where higher scores are

interpreted as more institutionalized and lower scores are interpreted as
less institutionalized.
Figure 4.3 provides a comparison of my institutionalization scores and
the “XCONST” variable from the Polity dataset. The graph reveals many
inconsistencies between the “XCONST” coding and observed executive
constraints in sub-Saharan Africa. For instance, while 100 percent of
countries that scored a six on the XCONST variable have constitutional
succession rules in place, less than 75 percent of countries that scored a
seven on the XCONST variable (a higher institutionalization score) have
constitutional succession rules in place.
I also provide comparisons between my institutionalization measures and
two V-Dem measures of executive oversight: legislative constraints on the
executive and judicial constraints on the executive. It is important to note
that these V-Dem measures should not be considered substitutes for my
measures of executive constraints. Instead, these comparisons reveal that
there are important differences across different types of executive constraints.
Simply having strong legislative oversight does not imply that the regime
automatically has strong constitutional rules that constrain the president.
I compare my regime institutionalization measures against the “legis-
lative constraints on the executive index” variable. This question was
124
0.8
Type of constraint
0.6
Succession rules
Term limits
Stable VP appoint
0.4
0.2
0 0.25 0.5 0.75 1

Level of legislative constraint on the executive
“Ordinal legislative constraints on the executive” variable
Figure 4.4 Comparison between institutionalization and V-Dem measure of legislative

constraint
coded by asking experts the following question: “To what extent is the
legislature and government agencies (e.g. controller general, general pro-
secutor, or ombudsman) capable of questioning, investigating, and exer-
cising oversight over the executive?”
Figure 4.4 provides a comparison of the legislative constraint scores
against my regime institutionalization measures. While the two vari-
ables appear generally correlated, some important differences arise.
For instance, a greater proportion of countries in my sample that
score a 0.5 on the V-Dem legislative constraint index have constitu-
tional succession rules in place compared with countries that score a
0.75 on the V-Dem legislative constrain index. This suggests that
simply having legislative constraints in place does not automatically
guarantee the existence of certain executive constraints, such as con-
stitutional succession rules.
Finally, I also conduct a comparison of my institutionalization mea-
sures against the “judicial constraints on the executive index” variable.
This question was coded by asking experts the following question: “To
what extent does the executive respect the constitution and comply with
court rulings, and to what extent is the judiciary able to act in an inde-
pendent fashion?”
125

0.8
0.6
Type of constraint
Succession rules
Term limits
Stable VP appoint
0.4
0.2
0 0.25 0.5 0.75 1
Level of judicial constraint on the executive
“Ordinal judicial constraints on the executive” variable
Figure 4.5 Comparison between institutionalization and V-Dem measure of judicial

constraint
Figure 4.5 presents a comparison between my regime institutionaliza-

tion measures and the V-Dem measures of judicial constraint. Perhaps
unsurprisingly, the two variables do not appear to be very closely related,
which implies that judicial constraints should be analyzed separately from
This chapter presented an original dataset of autocratic regime insti-
tutionalization in all sub-Saharan regimes from 1960 to 2010. First I
discussed how I conceptualize and operationalize autocratic regime
institutionalization more generally, then I presented the dataset by
explaining my coding strategy and showing descriptive findings.
Finally I compared my dataset to other commonly used datasets of
authoritarian institutions, such as regime typologies, Polity scores,
and V-Dem measures.
The rest of the book will be primarily empirical and will use this dataset
to test my arguments about the causes and consequences of regime insti-
tutionalization. The next chapter will test the theoretical argument made
in Chapter 2 and show that there is a relationship between initial leader
126
4.6 Conclusion
strength and levels of institutionalization. Chapters 6 and 7 will demon-

strate that institutionalization does indeed have regime stabilizing effects
by examining outcomes such as the length of leader tenure, coup risk, and
leadership succession.
Appendix
Appendix Table 4.1 List of countries and leaders included in this study
Angola Antonio Agostinho Neto 1976–1979

Jose Eduardo dos Santos 1980–2010
Benin Hubert Maga 1960–1963
Sourou Migan Apithy 1964–1965
Christophe Soglo 1966–1967
Emile Derlin Zinsou 1968–1969
Hubert Maga 1970–1971
Mathieu Kerekou 1972–1990
Botswana Seretse Khama 1966–1980
Quet K. J Masire 1981–1997
Festus G Mogae 1998–2007
Seretse Khama Ian Khama 2008–2010
Burkina Faso Maurice Yameogo 1960–1965
Sangoule Lamizana 1966–1980
Saye Zerbo 1981–1982
Thomas Sankara 1984–1987
Blaise Compaore 1988–2010
Burundi Michel Micombero 1967–1976
Jean–Baptiste Bagaza 1977–1987
Sylvestre Ntibantunganya 1994–1995
Domitien Ndayizeye 2003–2004
Jean–Pierre Nkurunziza 2005–2010
Cameroon Ahmadou Ahidjo 1960–1982
Paul Biya 1983–2010
Cape Verde Aristides Maria Pereira 1976–1990
(continued)
127
CAR David Dacko 1960–1965

Jean–Bedel Bokassa 1966–1979
David Dacko 1980–1981
Andre Kolingba 1982–1993
Ange–Felix Patasse 1994–2002
Francois Bozize 2003–2010
Chad Francois Tombalbaye 1960–1974
Felix Malloum 1976–1978
Goukouni Oueddei 1980–1982
Hissene Habre 1983–1990
Idriss Deby 1991–2010
Comoros Ali Soilih 1976–1977
Ahmed Abdallah Abderemane 1979–1989
Said Mohamed Djohar 1990–1995
Mohamed Taki Abdoulkarim 1996–1998
Assoumani Azali 1999–2006
Ahmed Abdallah Sambi 2007–2010
Congo DRC Joseph Kasavubu 1961–1965
Mobutu Sese Seko 1966–1996
Laurent–Desire Kabila 1997–2000
Joseph Kabila 2001–2010
Cote d’Ivoire Felix Houphouet–Boigny 1960–1993
Henri Konan Bedie 1994–1999
Laurent Gbagbo 2001–2010
Djibouti Hassan Gouled Aptidon 1978–1998
Ismael Omar Gelleh 1999–2010
Equatorial Guinea Francisco Macias Nguema 1969–1979
Teodoro Obiang 1980–2010
Eritrea Issaias Afewerki 1993–2010
Ethiopia Haile Selassie I 1960–1973
Tafari Bante 1974–1976
Mengistu Haile Mariam 1977–1990
Meles Zenawi 1991–1994
Negasso Gidada 1995–2001
Girma Wolde Giorgis 2002–2010
Gabon Leon M’Ba 1960–1967
El Hadj Omar Bongo 1968–2009
(continued)
128
4.6 Conclusion
Gambia Dawada Jawara 1970–1993

Yahya Jammeh 1994–2010
Ghana Kwame Nkrumah 1960–1965
J.A. Ankrah 1966–1968
Edward Akufo–Addo 1970–1971
I.K. Acheampong 1972–1978
Hilla Limann 1980–1981
Jerry Rawlings 1982–2000
Guinea Ahmed Sekou Toure 1960–1983
Lansa Conte 1984–2008
Guinea Bissau Luis De Almeida Cabral 1974–1980
Kumba Yala 2000–2002
Kenya Mzee Jomo Kenyatta 1964–1978
Daniel arap Moi 1979–2002
Lesotho Bethuel Pakalitha Mosisili 2000–2010
Liberia William Tubman 1960–1970
William Richard Tolbert 1971–1979
Samuel Kanyon Doe 1980–1990
Amos Claudius Sawyer 1991–1993
Charles Ghankay Taylor 1997–2002
Gyude Bryant 2004–2005
Ellen Johnson–Sirleaf 2006–2010
Madagascar Philibert Tsiranana 1960–1971
Gabriel Ramanantsoa 1972–1974
Didier Ratsiraka 1976–1992
Malawi Hastings Kamuzu Banda 1965–1993
Mali Modibo Keita 1960–1968
Moussa Traore 1969–1990
Mauritania Moktar Ould Daddah 1960–1978
Mohamed Khouna Ould Haidalla 1980–1984
Maawiya Ould Sid’Ahmed Taya 1985–2004
Ely Ould Mohamed Vall 2005–2006
Sidi Mohamed Ould Cheikh Abdellahi 2007–2008
Mozambique Samora Moises Machel 1976–1986
Joaquim Alberto Chissano 1987–2005
(continued)
129
Armando Emilio Guebuza 2006–2010

Namibia Samuel Daniel Nujoma 1990–2004
Hifikepunye Pohamba 2005–2010
Niger Hamani Diori 1960–1973
Seyni Kountche 1974–1987
Ali Saibou 1988–1992
Mahamane Ousmane 1993–1995
Ibrahim Bare Mainassara 1996–1998
Tandja Mamadou 2000–2004
Mamadou Tandja 2005–2009
Nigeria Alhaji Abubakah Tafawa Balewa 1960–1965
Yakubu Gowon 1966–1974
Olusegun Obasanjo 1976–1979
Alhaji Shehu Shagari 1980–1983
Muhammadu Buhari 1984–1985
Ibrahim Babangida 1986–1992
Sani Abacha 1994–1997
Republic of Congo Abbe Fulbert Youlou 1960–1963
Alphonse Massamba–Debat 1964–1967
Marien Ngouabi 1970–1976
Joachim Yhombi–Opango 1977–1978
Pascal Lissouba 1993–1997
Rwanda Gregoire Kayibanda 1962–1972
Juvenal Habyarimana 1973–1993
Pasteur Bizimungu 1994–1999
Paul Kagame 2000–2010
Sao Tome Manuel Pinto Da Costa 1976–1990
Senegal Leopold–Sedar Senghor 1963–1980
Abdou Diouf 1981–1999
Sierra Leone Milton Augustus Striery Margai 1961–1963
Albert Michael Margai 1964–1966
Banja Tejan–Sie 1968–1970
Siaka Probyn Stevens 1971–1985
Joseph Saidu Momoh 1986–1991
Valentine Esegragbo Melvine Strassar 1992–1995
(continued)
130
4.6 Conclusion
Somalia Aden Abdullah Osman 1961–1967

Abdirashid Ali Shermarke 1968–1969
Mohamed Siad Barre 1970–1990
Abdullahi Yussuf Ahmed 2006–2008
Sheikh Sharif Sheikh Ahmed 2009–2010
South Africa Nelson Rolihla Mandela 1994–1998
Thabo Mbeki 1999–2008
Jacob Zuma 2009–2010
Sudan Ibrahim Abboud 1960–1964
Ismail Al–Azhari 1966–1969
Jaafar Al Nemery 1970–1974
Gaafar Mohamed Nimeri 1976–1984
Ahmad Ali Al–Mirghani 1986–1987
Omar Hassan Ahmad Al–Bashir 1989–2010
Swaziland Barnabas Sibusiso Dlamini 2000–2010
Tanzania Julius Kambarage Nyerere 1962–1985
Ali Hassan Mwinyi 1986–1995
Benjamin William Mkapa 1996–2005
Jakaya Mrisho Kikwete 2006–2010
Togo Sylvanus Olympio 1960–1962
Nicolas Grunitzky 1963–1966
Gssingbe Eyadema 1967–2004
Faure Gssingbe 2005–2010
Uganda Milton Obote 1963–1970
Idi Amin 1971–1978
Milton Obote 1981–1985
Yoweri Museveni 1986–2010
Zambia Kenneth David Kaunda 1964–1991
Frederick Chiluba 1992–2001
Levy Patrick Mwanawasa 2002–2008
Rupiah Banda 2009–2010
Zimbabwe Robert Mugabe 1980–2010
131
Appendix Table 4.2 Summary of institutionalization scores by country
Term
Country Successor policy Successor (strict) limits VP/PM appoint VP/PM same Defense appoint Defense same
Angola 1 0.542 0.542 0.480 0.342 0.885 0.714

Benin 0.400 0.400 0.440 0.180 0.080 0.460 0.240

Botswana 1 0.311 1 1 0.866 0.088 0.044
Burkina Faso 0.440 0.440 0.440 0.500 0.340 0.820 0.500
Burundi 0.348 0.348 0.325 0.534 0.255 0.441 0.255
Cameroon 1 1 0.260 0.800 0.560 1 0.720
Cape Verde 0.857 0.857 0.857 1 0.857 0.971 0.714
CAR 0.440 0.440 0.260 0.500 0.180 0.320 0.100
Chad 0.580 0.280 0.280 0.560 0.220 0.680 0.300
Comoros 0.485 0.485 0.800 0.742 0.228 0.200 0.057
Congo DRC 0.166 0.166 0.625 0.395 0.145 0.395 0.145
Cote d’Ivoire 0.980 0.480 0.200 0.400 0.240 0.780 0.580
Djibouti 0.545 0.545 0.878 1 0.818 0.939 0.727
Equatorial Guinea 0.289 0.289 0.078 1 0.631 0.342 0.236
Eritrea 0.722 0.722 0.722 0 0 1 0.833
Ethiopia 0 0 0.200 0.880 0.680 0.960 0.680
Gabon 0.180 0.160 0.200 0.980 0.800 0.640 0.480
Gambia 0.350 0.350 0 0.975 0.775 0.125 0.075
Ghana 0.454 0.318 0.340 0.500 0.363 0.704 0.363
Guinea 0.420 0.020 0.020 0.480 0.340 0.540 0.320
Guinea Bissau 0.722 0.722 0.333 0.777 0.277 0.750 0.250
Kenya 1 1 0.413 0.934 0.739 0.413 0.326
(continued)
Lesotho 1 1 0 1 0.818 0 0
Liberia 0.860 0.860 0.560 0.640 0.460 1 0.680
Madagascar 0.380 0.380 0.260 0.920 0.620 0.700 0.420
Malawi 0.733 0.733 0.355 0.377 0.266 0.288 0.133
Mali 0.380 0.380 0.500 0.460 0.260 0.580 0.320
Mauritania 0.380 0.380 0.100 0.460 0.240 0.660 0.320

Mozambique 1 0.571 0.571 0.742 0.600 1 0.885
Namibia 1 1 1 1 0.857 1 0.761
Niger 0 0 0.140 0.580 0.300 0.600 0.280
Nigeria 0.560 0.500 0.420 0.600 0.460 0.660 0.380
Rep Congo 0.400 0.340 0.280 0.780 0.500 0.560 0.320
Rwanda 1 1 0.416 0.395 0.229 0.625 0.458
Sao Tome 1 1 0.571 0.771 0.285 0.828 0.514
Senegal 1 1 0.319 0.680 0.468 0.957 0.659
Sierra Leone 0.708 0.708 0.708 0.958 0.479 0.083 0.041
Somalia 0.176 0 0.235 0.970 0.617 1 0.529
South Africa 1 1 0.882 1 0.647 1 0.705
Sudan 0.440 0.440 0.240 0.860 0.420 0.660 0.340
Swaziland 1 1 0 1 0.636 0 0
Tanzania 0.723 0.723 0.723 1 0.765 0.702 0.510
Togo 0.360 0.360 0.060 0.440 0.180 0.400 0.260
Uganda 0.326 0.326 0.217 0.847 0.608 0.543 0.413
Zambia 0.434 0.434 0.413 0.934 0.500 0.739 0.413
Zimbabwe 1 1 0 0.741 0.516 0.709 0.548
5
What Are the Causes of Regime
Institutionalization?
Under what conditions will autocratic leaders institutionalize their

regimes? Chapter 2 presented a theoretical model demonstrating how
institutionalization ties the hands of the autocrat by shifting the future
distribution of power between elites and the leader. It also made
a prediction that leaders who enter power initially strong are less likely
to institutionalize their regimes because they do not need to create
institutionalized mechanisms of rent distribution to remain in power.
Chapter 3 illustrated these dynamics through case studies of Cameroon
and Côte d’Ivoire.
This chapter tests the theoretical argument made in Chapter 2 on
the full set of countries within sub-Saharan Africa. I test the argument
using two different research designs. First, I will present a series of
cross-sectional regressions focusing on the ways in which leaders
obtained power. Second, I will present a differences-in-differences
design using my time-series data on regime institutionalization.
The formal model presented in Chapter 2 argued that the initial
distribution of strength between the leader and elites affected the
leader’s decision to institutionalize the regime. Using a series of cross-
sectional regression, I will demonstrate that initial leader strength
is indeed associated with levels of regime institutionalization. To
measure leader strength, I propose that the ways in which leaders
obtain power determine their initial level strength vis-à-vis elites once
they enter office. In particular, I focus on the following three
comparisons:
First, I compare founding presidents with their successors. When we
examine all autocratic leaders within sub-Saharan Africa, we expect the
first president immediately following independence to be stronger than the
leaders who subsequently took power, largely due to the “founding father”
persona many of these first leaders adopted, which made them extremely
popular.
134
5.1 Reviewing the Theoretical Argument
Second, I compare leaders within the set of postindependence regimes.

Within the set of leaders who came into power immediately following inde-
pendence, those who founded mass movements (such as Felix Houphouet-
Boigny of Côte d’Ivoire) tended to be much stronger compared with leaders
who essentially inherited their position of power through a close alliance with
the outgoing colonial authorities (such as Ahmadou Ahidjo of Cameroon).1
Third, I compare leaders who took power through coup d’etats against
leaders who do not. Leaders who successfully enter power via coups have
coercive control upon taking power and can be interpreted as stronger
leaders compared with elites who do not exert the same kind of control
over the military.
Using these three strategies, I show that leaders who enter power
already strong are less likely to institutionalize the regime at the start of
their tenure, even if this initial advantage does not last indefinitely.
In addition, I leverage the end of the Cold War as a causal identification
strategy by showing that leaders with access to natural resources institu-
tionalized their regimes at lower rates compared with leaders without
access to natural resources. I first show that autocratic regimes in Africa
generally became more institutionalized following the end of the Cold
War. This trend reflects the increasing demands leaders faced in the 1990s
to implement democratic-seeming institutions, though as I argue, these
power-sharing measures likely buttressed authoritarian durability, rather
than usher in true democracy. Then, through a differences-in-differences
design, I show that leaders who had access to natural resource rents, such
as oil, institutionalized at lower rates compared to leaders who did not
have access to oil rents. This provides additional empirical support that in
light of exogenous shocks to the distribution of power, leaders who are
stronger do not need to institutionalize.
5.1 observable i mplications of the

theoretical argument
The formal model presented in Chapter 2 illustrates how institutionaliza-
tion empowers elites and outlines the conditions under which autocrats
choose to implement executive constraints. I will quickly summarize the
main finding from the model that is most relevant to the empirical tests
presented in this chapter.
1
It is helpful to remind the reader that we are interested in the leader’s relative level of
strength compared with elites. While the first strategy compares founders and successors
in the entire sample of leaders, the second strategy focuses on variation in leader strength
within the subset of leaders who took power immediately following independence.
135
5 What Are the Causes of Regime Institutionalization?
The model showed that leaders who enter power initially strong vis-à-
vis other elites are less likely to institutionalize the regime. This is
because initially strong leaders do not have to worry about commitment
problems arising in the future bargaining round. Commitment problems
arise only when elites are very strong relative to the leader in the first
period because are likely to succeed in a rebellion if they were to try to
depose the leader. In such circumstances, leaders need to institutionalize
the regime in order to remain in power because doing so effectively
transfers access to the state to elites, therefore guaranteeing their future
rents.
If instead, the leader enters power relatively strong, then elites cannot
demand a large share of rents because they do not have a credible threat of
rebellion in the first period. In this case, the leader does not have to
institutionalize the regime in order to remain in power because she can
make a distributive offer that elites will accept in the first period of
the game.
In short, the model argues that levels of regime institutionalization are
determined primarily by initial leader strength. This chapter will test the
following hypothesis:
Hypothesis 1: Initially strong autocratic leaders are less likely to
institutionalize their regimes.
5 . 2 j u s t i f y i n g t h e m o d e l ’ s as s ump ti ons
Before turning to the empirical test, I will first provide some empirical
evidence of two key assumptions made in the model. The first key assump-
tion the model makes is that leaders institutionalize at the beginning of
their tenure. Recall that the model assumes that leaders make decisions
about institutionalization after observing the initial draw of power. If this
is the case, then we should expect to see institutionalization occur near the
beginning of leaders’ tenures.
I provide empirical evidence that the creation of constitutional succes-
sion procedures and term limits do indeed occur within the first few years of
leaders coming into power. In fact, about half of these rules are created
within the leader’s first year of entering power. To show this, I calculate the
number of years into the leader’s tenure constitutional succession rules or
term limits were created. For example, Agostinho Neto, the first leader of
Angola, implemented formal succession rules the first year he came to
power, in 1976. Aristides Pereira, who became the first president of Cape
Verde in 1975, created formal succession rules in 1981 – six years after
taking power.
136
5.2 Justifying the Model’s Assumptions
I limit the time frame to rules created between 1960 and 1990,
since the end of the Cold War provided an exogenous shock to the
distribution of power of leaders, resulting in renewed incentives to
institutionalize (as this chapter will later discuss). I also focus primar-
ily on the creation of formal rules in this subsection because informal
power-sharing measures – the designation of vice presidents, prime
ministers, and defense ministers – are much more likely to have
interruptions or gaps. For example, Hamani Diori, the first president
of Niger, appointed a defense minister for about 60 percent of the
time he was in office. He began to appoint a defense minister six
years after the start of his term, but there was a one-year gap in 1971
in which a defense minister was not appointed. Similarly, Denis
Sassou Nguesso, who was president of the Republic of Congo from
1979 to 1992, appointed a defense minister from 1981 to 1984, and
again from 1991 to 1992.
From 1960 to 1990, I recorded all instances of “institutional
spells” for constitutional succession procedures and term limits. For
instance, succession procedures have existed within the Cameroon
constitution since 1960. These rules remained in place through
2010. Succession rules in Cameroon from 1960 to 2010 constitute
one institutional spell. For each institutional spell, I then recorded
when during the leader’s tenure that particular rule was created.
Returning to the Cameroon example, succession rules were created
in 1960, the first year of Ahmadou Ahidjo’s rule. That observation
would therefore score a 1 because the rule was created in the
first year of Ahidjo’s tenure.
Table 5.1 displays summary statistics of the number of years into the
autocrat’s tenure the rule was implemented.
The data shows that where formal executive constraints exist, they
are usually created within the first few years of the leader’s tenure.
From 1960 to 1990, there were fifty-three instances in which
Table 5.1 How long into their tenure do leaders create formal executive
constraints?
Number of Mean number Median number

institutional spells of years of years
Leadership succession 34 3.8 1

Term limits 19 4.3 3
Total 53 4 2
Note: Sample includes rules created between 1960 and 1990.
137
a constitutional succession rule or term limits were created. In 49 per-

cent of those instances, the rule was created in the first year of the
leader’s tenure. Conditional on being created, the average rule was
implemented in the fourth year of the leader’s tenure. The median
leader creates a formal rule in the second year of taking power. In
other words, most leaders institutionalize their regimes near the
beginning of their term. This is consistent with the theoretical argu-
ment that leaders make institutionalization decisions based on the
initial draw of power, which is generally determined by the ways in
which they obtain power.
A second key assumption the model makes is that once
implemented, institutionalization persists into the future. Recall that
the model assumes that the leader’s decision to institutionalize at the
start of the game affects the future distribution of power between the
leader and elite. For institutionalization to shift the future draw of
power, these rules should remain in place through the second period
of the game.
It is important to note that my argument about institutional per-
sistence pertains mostly to uninterrupted regime spells. When regimes
fall – particularly when leaders are deposed in coups – new leaders set
up their own institutional arrangements. In cases where the leader
institutionalized the regime, therefore creating lasting regimes that
survive multiple leadership transitions, power-sharing arrangements
constructed in the 1960s and 1970s have persisted over time. For
example, the regime in Tanzania (which has been in power since
independence) has maintained leadership succession policies and
stable cabinet appointments since the 1960s and this continues
today. However in cases where a personalist leader did not institu-
tionalize, leading to the downfall of the regime (either after the
leader’s death or due to a coup), the institutional makeup of the
regime changes after one regime ends and the next one begins. For
example, Ghana’s first president pursued very limited amounts of
institutionalization and he was deposed in a coup six years after
taking power. After the coup, the military suspended the constitution
therefore eliminating the leadership succession policy that had been in
place. In this case, the institutional arrangements that were initially
created did not persist after the fall of the regime.
I provide empirical evidence that constitutional rules, once created, do
generally tend to remain in place. The majority of constitutional
succession rules and term limits are not removed or revoked once they
are created. To demonstrate this, I return to my analysis of institutional
138
5.2 Justifying the Model’s Assumptions
Table 5.2 Do constitutional rules persist, once created?
Institution Number created Number remaining Percent remaining
Constitution 83 42 51%
Conditional on the constitution not being revoked

Successor 43 32 74%
Term limits 37 27 73%
spells. Once again, I focus primarily on the creation of formal constitu-

tional rules because the appointment of elites to key cabinet positions
frequently has interruptions or gaps.
From 1960 to 2010, I recorded all instances of “institutional spells” for
constitutions, constitutional succession rules, and term limits. During that
time period, there were eighty-three instances in which a constitution was
created. Out of these eighty-three constitutional spells, the constitution
remained in place through the end of the dataset 51 percent of the time.
Descriptive statistics are summarized in Table 5.2. Constitutions are fre-
quently suspended or revoked during periods of regime change or in the
aftermath of a coup.
Given that leaders often suspend constitutions after regime change,
I argue that in order to answer the question of whether leaders revoke
constitutional succession rules, we should condition on the constitution
not having been suspended or revoked. In cases where the constitution
was revoked, it is difficult to know whether the leader would have kept
succession rules intact if the entire constitution had not been suspended.
Therefore, we are able to observe whether succession rules were kept in
place only for succession rules in constitutions that were not eventually
suspended.
Conditional on the constitution not being suspended, there were forty-
three instances where a constitutional succession rule was created. In
74 percent of these cases, once created, the succession rule was kept in
place through the end of 2010. Moreover, conditional on the constitution
not being suspended, there were thirty-seven instances where term limits
were created. In 73 percent of these cases, once created, term limits were
kept in place through the end of 2010. In other words, constitutional
succession rules and term limits generally remain in place, when they are
created.
139
5.3 cross-sectional design: data

an d me as u re m e nt
Now we turn to testing the hypothesis that leaders who enter power
initially strong are less likely to institutionalize the regime. This
section summarizes the measurement approach I take – especially
with regards to measuring the distribution of power between leaders
and elites.
The analyses in this section are cross-sectional, and the unit of obser-
vation is the leader, since the primary independent variable is initial
leader strength. I will return to the time-series structure of the dataset
later in the chapter when I present a differences-in-differences research
design.
5.3.1 Independent Variables: Measuring Leader Strength

The independent variable for the empirical tests in this chapter is the
incumbent’s risk of removal if challenged by elites. In other words, we
are interested in measuring the leader’s initial strength, vis-à-vis other
elites. Measuring the distribution of power between political actors in
authoritarian contexts is a nontrivial challenge. The absence of electoral
and formal political competition (such as presidential or legislative elec-
tions) obscures accurate information about the true strength of incum-
bents relative to other elites.
The main relationship that is explored in the model is the leader’s
strength when she first enters office and how this affects her decisions
to institutionalize at the start of the game. (Recall that the model’s
main findings are centered around p1, the draw of the leader’s
relative power in the first period, rather than the distribution of pt.)
To measure this initial draw of power, I examine the ways in
which postindependence leaders came to power within sub-Saharan
Africa.
I generate three strategies, based on this historical approach, to
approximate the distribution of power between incumbents and elites.
The first strategy compares the first president of each country against
their successors. First presidents were generally bolstered by initial exci-
tement over independence, and were therefore quite popular. Moreover,
presidents often strategically designate weak elites as their successor, due
to fears of being deposed by their appointees. We should therefore
expect founding leaders to be strong leaders compared with their
successors.
140
5.3 Cross-sectional Design: Data and Measurement
It is important to note that all my measures reflect relative leader

strength, not absolute leader strength. I am arguing that founding leaders
are, in general, stronger compared to their successors. This does not mean
that all founding leaders are, in absolute terms, extremely strong, nor does
it assume that all founding leaders are uniformly strong. It simply means
that when we compare founding leaders against nonfounding leaders,
founding leaders are stronger than their counterparts.
The second strategy focuses on the set of leaders who were the first
presidents of the newly independent African states after decolonization.
Within these postindependence regimes, leaders who had led robust social
movements were extremely popular and perceived as the “founding
father” of the newly independent states. Such leaders were much stronger
compared with leaders who essentially inherited their positions of power
through close relationships with the outgoing colonial power.
Third, I focus on leaders who came into power through coups. Such
incumbents have coercive control upon taking power and therefore can be
interpreted as stronger leaders compared with elites who do not have
control over the military.
The strategy of focusing on the initial distribution of power has
a distinct empirical advantage – it lessens the concern that the relationship
between institutionalization and the distribution of power is endogenous.
Since the process of institutionalization shifts the distribution of power in
future periods, it would be difficult to untangle the effect of the incum-
bent’s average level of strength because institutionalization affects the
future distribution of power.
By contrast, when we focus on the initial distribution of power that
exists as the leader enters office, the incumbent has not made her decision
to institutionalize yet. This initial measure of leader strength is therefore
plausibly exogenous to the degree of institutionalization that is implemen-
ted after the leader takes office. I now discuss each of the strategies in turn.
5.3.1.1 Founding Presidents. When founding presidents first take

power, they tend to be extremely influential and popular. In their study
of African leaders Bienen and van de Walle (1989) argue that the first
leader following independence enjoyed a special legitimacy and mass
support during their tenure. Other scholars have also observed that foun-
ders of authoritarian regimes tend to be highly influential and charismatic,
often becoming personalist dictators during their rule (Chehabi 1998;
Panebianco 1988). Being a founding father of an autocracy therefore
serves as an indicator of a strong leader.
Within sub-Saharan Africa, the leader’s initial popularity was
often also boosted by the fact that the country had just gained
141
independence. Jomo Kenyatta, the first president of Kenya, is a prime

example of the immense power he acquired simply by being asso-
ciated with the founding of the independent Kenya state.
[Kenyatta] served as a popular symbol of the new Kenyan nation: in the
public mind it was difficult to distinguish between the two. For almost three
decades Kenyatta and Kenya were as one . . . The Kiswahili term of respect
Mzee, meaning “wise old man” became Kenyatta’s unofficial title as father of
his country, with the connotation of paternal authority and sanctity. But
more than the nation or its people he symbolized the Kenyan state: the
patriarch of Kenya . . . his position was virtually unassailable. (Jackson and
Rosberg 1982, 99)
On the other hand, leaders who take power by succeeding another

president are frequently much weaker and less popular compared
with their predecessors. By the end of the tenure of the founding
leader, initial excitement over independence and the creation of
a new republic faded quickly in many African states (Bienen and
van de Walle 1991; Boone 2014; Jackson and Rosberg 1982).
Therefore, leaders who subsequently come to power following the
first president involuntarily lose the advantages that were endowed
to the “founding fathers” as a result of the passage of time away from
independence.
In addition, presidents sometimes strategically choose successors
who they believe do not have the charisma and influence to preemp-
tively depose them (Tullock 1987). Therefore, successors often start
out less influential then their previous incumbents by design. Daniel
arap Moi, for instance, was Jomo Kenyatta’s designated successor for
many years before Kenyatta died. While Moi was vice president and
at the start of his tenure as president, he was not perceived to be
a strong leader. “Before Jomo Kenyatta died in 1978, many observers
of Kenya’s politics considered his vice president, Daniel arap Moi, to
be unintelligent. ‘Moi’ jokes made the rounds of Kenya’s political
circles. Yet Moi has been in power more than a decade and is now
considered a canny political leader” (Bienen and van de Walle
1991, 6).
Moreover, would-be successors often reinforce this dynamic while
their predecessors are still in power because it is dangerous for
appointed heirs to amass too much power while their predecessor is
still in office. “Although Moi served as vice president for twelve
years, [he] had acted primarily as Kenyatta’s agent in building bridges
between the country’s different cultural communities and had little
opportunity, perhaps little inclination, to articulate his own views”
(Widner 1992, 134). In his ascension to the presidency, Moi also did
142
not attempt to frame himself as a nationalist figure in the fight for

independence, despite the fact that he had been an active figure in
Kenya’s preindependence politics.2 In sum, often by design, successors
are generally much weaker than their predecessors when they first
take office.
To summarize, the founding president of the country following
independence was often quite strong relative to other elites. We
should therefore expect first presidents to be less likely to institutio-
nalize their regimes. For this strategy, I create a dummy variable that
takes a value of 1 if the leader was the first president of the regime
and a 0 otherwise.
5.3.1.2 Strong Nationalist Leaders. My second strategy focuses on

variation within the sample of leaders who were founding presidents.
In the two decades leading up to independence, political movements
emerged in virtually all colonized territories within Africa. On one
hand, many of these organizations were active independence move-
ments that lobbied for liberation from the European powers. The
leaders of these independence movements, such as Jomo Kenyatta of
Kenya or Sekou Toure of Guinea, enjoyed high levels of popularity
and mobilized large bases of support (Boone 2014; Collier 1982;
Levitsky and Way 2012; Mamdani 1996). A number of these leaders
participated in armed struggles against the colonial powers, which
afforded them extremely high levels of legitimacy upon taking power
(Huntington 1968; Levitsky 2012).
Felix Houphouet-Boigny of the Ivory Coast is an example of such
a figure. Houphouet-Boigny had been a famed advocate for indepen-
dence early on in the movement for self-government. In 1946, he
founded the Rassemblement Democratique Africain (RDA) – a pan-
African organization consisting of multiple political parties from
French colonies that lobbied for independence. That same year,
Houphouet-Boigny also founded the Parti Democratique de la Cote
d’Ivoire (PDCI) to fight for self-determination within the Ivory Coast.
Even before independence was granted, the PDCI – controlled by
Houphouet-Boigny – dominated politics. Jackson and Rosberg (1982)
describe Houphouet-Boigny as dominating all aspects of political life
2
In fact, prior to independence, Moi cofounded the Kenya African Democratic Union
(KADU), a rival political party to KANU, which advocated for a federal system. He
also harbored rebel fighters on his farm during the Mau Mau Uprising that was
fought against the British in the 1950s (Widner 1992, 135).
143
“since the first stirrings of nationalist politics in 1946,” controlling the state
bureaucracy and National Assembly with “stern and unrelenting control”
(145–147). The government of the Ivory Coast under Houphouet-Boigny
was the government of “virtually one man” (145).
Conversely, in states in which active independence movements did not
emerge, preindependence parties existed to participate in elections that were
overseen by the colonial authorities. In such organizations, high-ranking
politicians were often perceived as part of the colonial bureaucracy. Such
politicians essentially inherited their positions of power after decolonization
but were not shielded by the same influence and mass following that
surrounded pro-independence leaders. I interpret these preindependence
statesmen who did not lead mass movements as weak autocrats compared
to independence leaders.
Ahmadou Ahidjo, the first president of Cameroon, is a prime example
of such a case. In the decades prior to independence, Ahidjo was encour-
aged to run for office by the French authorities, who viewed him as
a politician who would act in the interest of the colonial state. In fact,
the colonial administration even referred to the “Ahidjo option” – given
that independence seemed to be increasingly inevitable, the French
authorities preferred to have Ahidjo as head of state (Joseph 1978, 46).
Ahidjo did indeed take the office of the presidency upon independence,
but experienced very low levels of popular support. “In spite of the
efforts of the propagandists of the regime, President Ahidjo cannot be
considered a heroic figure from the nationalist past . . . it is a matter of
historical fact that he was initially “fabricated” and brought to power by
the colonial regime” (Joseph 1978, 76).3 As a result, upon taking power
Ahidjo needed to create official structures that would allow him to buy
the support of key elites – the lamidos (chiefs) who still controlled much
of the northern regions. He created a constitutional stipulation that
stated traditional chiefs would be protected, and filled key cabinet
appointments with elites that the lamidos favored. However, Ahidjo
made it clear that in exchange for these benefits, he expected the lamidos
to support his rule.
To sum, within the sample of founding presidents, there is
variation in terms of the strength of these leaders upon taking office.
Presidents such as Ahmadou Ahidjo who came to power by inheriting
their positions from the colonial state tended to lack the popularity
3
144
and influence of nationalist figures. He therefore had to institutiona-

lize the regime in order to buy support from other elites. Strong
nationalist figures, such as Felix Houphouet-Boigny, were exceedingly
popular upon taking office at independence. Such leaders did not
need to institutionalize their regimes because they did not face
a credible threat from other elites and had very low likelihoods of
being deposed. Within the sample of first presidents, we should there-
fore expect that strong nationalist leaders are less likely to institutio-
nalize the regime after coming into power.
To code this variable, I looked for leaders who fought in
independence wars, those who were jailed by the colonial state for
pro-independence activities, or those who spearheaded wide-reaching
pro-independence movements.4 Such leaders entered power strong,
compared with postindependence leaders who inherited their positions
from the outgoing colonial power. I create a dummy variable that takes
a value of 1 if the leader was a strong independence movement figure and
a 0 otherwise.
5.3.1.3 Coup Leaders. For my third strategy, I focus on leaders who

came into power through coups. One of the most precarious challenges
to any civilian incumbent is the threat of military takeover. When the
leader is the head of the military, however, this threat is diminished.
Moreover, autocrats who successfully come into power through a coup
demonstrate that they have control over the coercive apparatus that
allowed them to launch a successful coup in the first place (Gehlbach
and Keefer 2011). We should therefore expect successful coup leaders to
be less likely to institutionalize their regimes because they have control
over the military.
This argument builds on existing studies of coups and military govern-
ments, although the literature on this topic is somewhat limited. Geddes
(2009) examines variation in the creation of institutions in military
regimes that emerge out of successful coups. She argues that leaders of
military regimes that are run by highly professionalized militaries are less
likely to rely on civilian institutions in order to share power since the
threat of being ousted is credible. In un-professionalized militaries, leaders
cannot be certain that lower-ranked officers will comply with their orders.
They are therefore more likely to empower civilian institutions in order to
4
For my coding of independence leaders and preindependence statesmen, I rely on
Morrison (1989), Clodfelter (1992), and historical accounts.
145
act as a counterbalancing force against the threat posed by other military

officers. Geddes uses the ranking of the officer as a proxy for the level of
professionalism in the military. She argues that coups led by junior officers
demonstrate a lack of discipline in the military. Coups led by generals, on
the other hand, tend to have the unified support of the military.
Unfortunately, the study does not break down how Geddes codes junior
and senior officers, so it is unclear what officer rankings are associated
with the categories of seniority in this study.
Singh (2014) makes the case that the success of coups depends on the level
of seniority of the officer, primarily because senior officers have more access
to resources that facilitate the execution of a successful coup. Coups that are
carried out from the top – namely by generals – are most likely to succeed
because senior officers have more influence and authority that allow them to
shape expectations about the likelihood of a successful coup early on in the
process. Coups that are carried out by mid-ranking officers – majors, lieute-
nants-colonels, and colonels – succeed about half the time because mid-level
officers can usually seize symbolic targets and gain access to broadcast
capabilities, such as the radio. Finally, coups that come from the very bottom
of the officer corps – those organized by enlisted men or very junior officers –
are least likely to succeed because these officers lack the influence and
personal resources to convince others that their coup will succeed.
Singh’s argument resonates with the officer rankings available in my
data. Out of seventy coup leaders who became the president of the regime,
I have data on the officer rankings of fifty-three leaders. Generals make up
38 percent of coup leaders and 40 percent are mid-level officers (such as
majors and lieutenant colonels). Only 22 percent had ambiguous titles or
were clearly very low ranked (such as a squadron commander). In other
words, the majority of the coup leaders in my dataset (78 percent) were
mid- to high-ranking military officers who had reasonable control over the
military upon seizing power in a coup.
Returning to Geddes’s (2009) thesis, we should therefore expect that the
majority of coup leaders in my sample would be less likely to institutionalize
the regime because they are high-ranking officers who have already demon-
strated that they have control over the military and coercive apparatus of the
state. They therefore do not need to rely on civilian institutions in order to stay
in power. In sum, coup leaders in my sample should be associated with lower
levels of regime institutionalization. For this strategy, I create a dummy vari-
able that takes a value of 1 if the leader entered power through a coup and a 0
otherwise.5
5
To code this variable, I relied on Goemans, Gleditsch, and Chiozza 2009, Powell and
Thyne 2011, and various leaders biographies.
146
5.4 Empirical Strategy and Findings
5.3.2 Dependent Variables: Regime Institutionalization

The primary dependent variable (DV) for the empirical analyses is the
level of regime institutionalization. Since we are primarily concerned with
the leader’s initial draw of power, our units of analysis are cross-sectional
and the analysis conducted at the leader level. I transform my time-series
institutionalization measures into cross-sectional units at the leader level.
Since my original measures of regime institutionalization are dummy
variables, to transform them into cross-sectional indicators I calculate
the percentage of years for which a dimension of institutionalization
was implemented. For instance, a leader who was in power for twenty
years and had a vice president for ten of those years would score a .50 on
that particular dimension of institutionalization.6
To sum, the cross-sectional version of the institutionalization scores
report the proportion of years for which a leader implemented a particular
dimension of institutionalization. These scores range from 0 to 1, with
higher scores indicating higher levels of institutionalization. I present
findings using both my seven disaggregated measures of institutionaliza-
tion, as well as an aggregate institutionalization score that was calculated
using Item Response Theory (IRT). The IRT institutionalization score is
a continuous measure, ranging from 0 to 1.7
5.4 cross-sectional design: empirical strategy

and fin d i n g s
I now present the findings from empirical analyses focusing on how
leaders enter power as an indicator of their initial draw of strength.
The analyses in this section are cross-sectional, and the unit of obser-
vation is the leader, since the primary independent variable is initial
leader strength. In the following section, I will present an alternative
test which returns to the time-series structure of the dataset by exam-
ining changes in levels of institutionalization before and after the end
of the Cold War.
The estimating equation used for the empirical models in this section is
the following:
6
X
years in power
7
Details on this method are included in the Appendix at the end of this chapter.
147
Y i ¼ α þ βleader strengthi þ δXi þ γW c þ i

where Y i are my measures of institutionalization, estimated separately.
The unit of analysis is the leader, and the outcome variables are cross-
sectional, calculated as the percentage of years in which the leader
implemented a particular dimension of institutionalization. Since
the outcome variable is a continuous institutionalization score, I use
ordinary least squares (OLS) for the analysis in this section.
leader strengthi is a dummy variable indicating whether the leader is
coded as a strong or weak leader using the strategies detailed above,
and β is the main parameter of interest. Xi is a set of leader-specific
characteristics, and W c is a set of country-specific characteristics. i is
the error term. All models include robust standard errors clustered at the
country level.
Should the results in this section be interpreted as causal or corre-
lational? Since the analysis is based on cross-sectional observational
data and ordinary least squares regressions, there are limits to the
extent with which we can interpret the results as causal. In all of the
models, I included controls for the most likely confounders or alter-
native explanations. These controls include leader-specific character-
istics, such as the number of years the leader was in office, whether
the leader has a military background, and whether the leader has
a ruling party. I also control for country-specific characteristics,
such as GDP per capita, oil production per capita, the size of the
country, the size of the population, a measure of ethnic fractionaliza-
tion, and colonial controls (identity of the colonizer, size of settler
population). However, we cannot completely eliminate the possibility
of reverse causality or omitted variable bias, so the reader is encour-
aged to interpret the results from the cross-sectional analysis as
correlational. The goal of this chapter is to provide empirical support
for the theoretical argument made in Chapters 2 and 3 that initially
strong leaders are less likely to institutionalize.
5.4.1 Founding Presidents

My first strategy uses founding presidents as a proxy for strong
leaders. I restrict the time period for this sample to include only
observations from 1961 to 1991, because the end of the Cold War
resulted in a shift in the distribution of power for all leaders in the
continent (I explore this event later in the chapter). The results are
actually stronger when we include post-Cold War observations, but
148
Figure 5.1 Founding leaders versus subsequent leaders
these findings are inflated by exogenous changes that occurred after

1991. This sample also excludes leaders that came into power via
coups. I make this restriction because founding presidents never came
into power via coups – they are all civilian leaders. I therefore only
include successors who were also civilian leaders as well in order to
facilitate a more consistent comparison.8
Figure 5.1 illustrates the results from this analysis, and the full
regression table is reported in Appendix Table 5.1. Recall that the
main independent variable for this analysis is a dummy variable that
takes a value of 1 if the president was a founding leader. The results
show that founding presidents are significantly less likely to institu-
tionalize along formal dimensions – they are less likely to create strict
succession policies and constitutional term limits. Although the coef-
ficients for informal institutions are also negative for first presidents,
the difference is not statistically significant. This suggests that some
founding leaders may have used appointments to key cabinet posi-
tions as a way of maintaining support from elites, especially as their
initial popularity from independence began to wane.
8
I only exclude leaders who came into power via coups – I do not exclude leaders who
have a military background who did not come into power via coups.
149
5.4.2 Strong Nationalist Leaders

My second strategy looks at the subsample of first presidents after
independence, and argues that leaders who were strong nationalist
figures experienced high levels of influence and popularity upon tak-
ing office. These leaders therefore did not need to institutionalize the
regime in order to stay in power. This sample is also restricted to
the pre-1991 time period. The sample only includes leaders from
the first regime. I define a first regime as the first continuous regime
after independence. As soon as a regime experiences a violent deposi-
tion, it drops out of the sample. Leaders are defined as a “strong
independence movement leader” if they started a mass movement or
party during the independence era to lobby against the colonial
powers. Leaders who were jailed for pro-independence activities or
who were important actors in armed combat are also coded as strong
independence movement leaders.
These results are illustrated in Figure 5.2 and the full regression
results are reported in Appendix Table 5.2. Recall that the main
independent variable for this analysis is a dummy variable that
takes a value of 1 if the president was a strong nationalist leader.
Strong nationalist leaders are significantly less likely to have formal
strict succession rules, and they are also significantly less likely to
Figure 5.2 Strong nationalist leaders versus presidents allied with outgoing colonial
powers
150
appoint a vice president or prime minister. The coefficients on most

other dimensions of institutionalization are also negative, though not
statistically significant.
A potential concern is that nationalist leaders who fought in inde-
pendence wars might be systematically different from nationalist lea-
ders who led peaceful independence movements. To account for this
possibility, I run a robustness test that controls for a leader’s partici-
pation in an independence war. The results from this robustness
check are reported in Appendix Table 5.3. Even when controlling
for participation in an independence war, nationalist leaders institu-
tionalize less on the aggregate institutionalization score and they are
also less likely to create leadership succession policies. The coefficients
on most other dimensions of institutionalization are also negative,
though not statistically significant.
5.4.3 Coup Leaders

My third strategy uses coup leaders as a proxy for initial leader strength.
Leaders that successfully come into power via coups have control over the
military and coercive apparatus of the state. They therefore face a lower
probability of being overthrown and therefore are less likely to
institutionalize.
These results are illustrated in Figure 5.3, and the full regression results
are reported in Appendix Table 5.4. Recall that the main independent
variable for this analysis is a dummy variable that takes a value of 1 if the
president came into power via a coup. The results show that coup leaders
are much less likely to institutionalize along all dimensions of institutio-
nalization, and these differences are statistically significant across all
dimensions. These results still generally hold when we break down the
sample into different time periods. Appendix Table 5.5 reports the results
on a subsample of the data that is restricted from 1960–1991, and the
results remain the same. Appendix Table 5.6 reports results on
a subsample of the data that is restricted to the post-Cold War period
(1992–2010), and postcoup leaders remain less likely to institutionalize,
although some results lose their significance. This may be a result of the
fact that coups have generally become less common in the post-Cold War
period.
I also examine variation within coup leaders. Geddes (2009) makes
the claim that high-ranking officers should be less likely to rely on
civilian institutions because they have also consolidated control over
the military apparatus upon taking power. Low-ranking officers who
151
Figure 5.3 Coup leaders versus noncoup leaders
come into power via a coup should be more likely to institutionalize

because their grip on authority is less established. I examine levels of
institutionalization within the subset of coup leaders and find that offi-
cers who are ranked at the level of “general” are less likely to keep formal
constitutions in place, compared with coup leaders who have lower
ranking military statuses (such as majors or colonels). I find some weak
evidence for this claim. Most of the dimensions of institutionalization
are negatively correlated with rank, though most fail to be statistically
significant. Interestingly, I find the opposite to be true of defense minister
appointments (and stable defense appointments). Both of these are posi-
tively correlated with high-ranking officers, and these relationships are
statistically significant. Results from these regressions are reported in
Appendix Table 5.7.
In addition, I run a robustness check that takes into account the
concern that low-ranking coup leaders might not be strong at all.
I create a dummy variable, “High-ranking coup leader,” that takes
a value of 1 for leaders who come into power via a coup and held
a rank of general at the time of the coup. The variable takes a value
of 0 for all other observations, including coup leaders who hold a low
ranking. The analysis is reported in Appendix Table 5.8 and the
results remain consistent.
152
5.4.3.1 Alternative Hypotheses: Coup Traps and Military

Commissions. Are coup leaders less likely to institutionalize simply
because they fall into coup traps and therefore are not in power for very
long? When we compare the mean and distribution of the number of years
in office of coup leaders and noncoup leaders, it is clear that there is not
a significant difference between the two groups. Coup leaders remain in
office for an average of 8.21 years and noncoup leaders remain in office for
an average of 8.62 years. This difference is not statistically significant
(p-value=0.74). Moreover, the distributions of time in office between the
two groups do not look very different either. The number of years in office
for all coup leaders ranges from one to thirty-seven, and for noncoup
leaders, this variable ranges from one to forty-one. It is therefore not the
case that coup leaders institutionalize less simply because they are in
power for significantly shorter periods of time.9
A second possible alternative explanation is that coup leaders and
military governments do not rely on civilian institutions at all; even if
they wanted to create an institutional power-sharing mechanism, they
would do so through a military commission, rather than through the
presidential cabinet. Therefore, the observation that coup leaders do
not make key cabinet appointments such as vice president, prime
minister, or defense minister does not imply that the regime is weakly
institutionalized; military leaders may simply institutionalize through
different channels.
I consider this alternative hypothesis by collecting data on the creation
of military commissions. Since the alternative argument is that coup
leaders are actually institutionalizing through other mechanisms rather
than cabinet appointments, I restrict the observations in this section only
to coup leaders who did not appoint a vice president or prime minister.
The goal is to find out whether coup leaders who did not name a second in
command in the presidential cabinet did so through a military commis-
sion. For every leader–year observation for which a coup leader did not
appoint a vice president or prime minister in the cabinet, I record whether
the leader established a military commission. I define a military commis-
sion as a committee of military elites whose positions take precedence over
the state presidential cabinet.10
9
In all the models, I also control for the number of years in power.
10
This information is available from the Europa World Year Book (1960–2010).
Military commissions may have various names: “revolutionary council,” “national
redemption government.” However, as long as it is clearly a body composed only of
military elites whose authority supersedes the state executive, I consider this body
a type of military commission.
153
I find that the majority of coup leaders who do not appoint a vice
president or prime minister also do not create military commissions. Out
of ninety-one total leader–year units, military commissions are created
only 36 percent of the time. Out of twenty-nine leaders in this sample who
were in power for more than one year, only nine had commissions the
entire time.11
Importantly, for all of these observations, cabinets were never comple-
tely suspended. In other words, coup leaders who did decide to establish
military commissions kept the executive cabinets alongside the commis-
sions – although the military commissions appear to have taken prece-
dence over civilian institutions. However, coup leaders are more likely to
have removed key appointments within the cabinet – suggesting a lower
level of institutionalization along these dimensions.
5.4.4 What About Rebel-Group Regimes?

A possible additional explanation for variation in regime institutionaliza-
tion is that regimes that were led by rebel-group parties that fought for
independence, such as the Mozambique Liberation Front (FRELIMO) or
the MPLA in Angola, tend to be strong and institutionalized organizations.
Huntington (1968) first introduced this argument, claiming that party-
based regimes that emerge out of sustained revolutionary or nationalist
struggles are more durable compared with regimes that came into power
without such a struggle.
Recent studies have extended this theory, arguing that parties and elites
who face strongly organized enemies – often emerging out of revolutions
or liberation conflicts – tend to build strong parties and regimes to manage
such threats (Smith 2005). Levitsky and Way (2012; 2013) argue that
armed conflict, usually taking the form of revolutionary struggles or
independence wars, provide ruling parties with a crucial source of cohe-
sion that facilitates stable autocratic rule. Other scholars have also argued
that rebel-group governments that emerge out of civil war often benefit
from high levels of elite solidarity, organizational structures developed
during wartime, and legitimacy from winning the conflict. This provides
such postconflict governments with distinct incumbency advantages,
allowing them to consolidate autocratic rule (Dresden 2017; Lyons 2016).
In contrast with this dominant view, some scholars suggest that
organizations that are built for fighting and coming to power are not
11
The average size of a military commission is sixteen members.
154
necessarily well equipped to govern. In his discussion of independence

parties in Africa, Welch (1970) asserts that the guerrilla style tactics and
organization of parties that were engaged in anti-colonial activities
became considerably less effective for creating a self-governing state
after independence. In fact, Lyons (2016) notes that “insurgent groups
are often violent, fractious, and support all-or-nothing policies, all char-
acteristics that make them unlikely candidates for successful peacetime
regimes” (2–3). As Slater points out “revolutionary parties tend to frag-
ment once their shared enemy is vanquished – especially when that enemy
is a departed colonial power, as in Burma or Indonesia” (2010, 52).
In fact, detailed case studies of mass-based independence move-
ments in Africa suggest that even during the height of independence
struggles, many organizations were plagued with weak organizational
structures. Mulford (1967), who details the creation and organization
of the United National Independence Party (UNIP) in Zambia, argues
that the party had “almost no funds and lacked sufficient numbers of
able and dedicated prepared to devote themselves to the laborious
task of party organization” (143). Bienen (1970) observes similar
organizational weakness in Tanzania, noting that “TANU’s rapid
growth into a mass movement in the 1960’s was characterized by
an absence of central direction, due largely to the fact that there
existed neither a central staff nor a firm base for central finances”
(43).
These organizations were also frequently strung together by tem-
porary goals of gaining independence that were soon replaced by
challenges of how to distribute patronage and implement power
sharing in the new regime. Bienen notes that a central problem facing
Julius Nyerere, the first leader in Tanzania after independence, was
that not all party elites from the preindependence era could be
afforded top government positions – a source of dissatisfaction that
his opponents could exploit. “It was one thing for TANU to organize
and become the dominant national movement; it was quite another to
establish an effective government over many small-scale and dispersed
communities after independence . . . overcoming parochialisms for the
sake of organizing a national movement is very different from ruling
a society” (1970, 43). In sum, these studies cast some doubt on the
claim that all revolutionary or pro-independence organizations were
strong, well organized, and fully capable of dealing with the chal-
lenges of governing postcolonial states.
So are rebel-group regimes likely or unlikely to be institutionalized upon
taking power? My theory suggests a more nuanced answer. On one hand,
regimes that emerge out of independence wars have already eliminated
155
armed rivals within the government by the time they take power.12
Moreover, the elites who take control of the regime after the conflict ends
tend to be the military leaders who were responsible for winning the
independence wars; in these cases, civilian elites either do not exist or are
irrelevant. These leaders tend to be extremely popular and influential,
similar to strong nationalist figures that emerge out of peaceful indepen-
dence movements. Levitsky and Way (2012) note that armed conflict
produces revolutionary leaders with “extreme legitimacy and unquestioned
authority.” It is precisely this “unquestioned authority” that allows leaders
that emerge out of independence wars to be strong relative to other elites.
Based on this logic, we would therefore expect these leaders to be less likely
to institutionalize their regimes.
However, unlike nationalist figures or founding presidents that came
into power after peaceful independence movements, former rebel-group
leaders have a distinct weakness – the existence of other military elites
who have the capacity to overthrow them. Unlike coup leaders, who tend
to act with the assistance of very few other actors, independence wars
require extensive rebel groups that are led by multiple military actors.
Leaders of rebel-group regimes need to ensure that they retain the sup-
port of military leaders who participated in the independence wars and
do so by institutionalizing along dimensions particularly relevant for
coercive actors: the defense minister cabinet post (Meng and Paine
2020).
I argue that leaders of regimes that came into power by fighting
independence wars should be more likely to institutionalize their regime
along two particular dimensions: the appointment of a defense minister
and the appointment of a stable defense minister. I test this argument by
creating a dummy variable, “rebel group,” that takes a value of 1 if the
regime came into power by fighting an independence war. For my coding
of rebel-group parties and independence wars, I refer to Clodfelter
(1992) and the UCDP/PRIO Armed Conflict Dataset (Gledditsch et al.
2002). I refer to the UCDP definition of conflict as “a contested incom-
patibility that concerns government and/or territory where the use of
armed force between two parties, of which at least one is the government
of a state, results in at least 25 battle-related deaths” (Gledditsch et al.
2002). Countries with rebel-group governments include Angola, Cape
12
Although in some of these countries, there are multiple rebel-groups that fought for
independence. Angola, for instance, had UNITA (National Union for the Total
Independence of Angola) in addition to the MPLA and Mozambique had
REMANO (Mozambican National Resistance) in addition to FRELIMO. These
other rebel-groups, however, did not take office or have any say in the government
after independence was granted.
156
5.5 Differences-in-differences Design
Figure 5.4 Rebel-group leaders versus nonrebel-group leaders
Verde, Eritrea, Guinea-Bissau, Mozambique, Namibia, South Africa,

and Zimbabwe.
The empirical analysis supports my argument. The results are illustrated
in Figure 5.4, and the full regression results are reported in Appendix Table
5.9. Recall that the main independent variable for this analysis is a dummy
variable that takes a value of 1 if the regime came into power by fighting an
independence war. Former rebel-group leaders are significantly more likely
to make defense minister appointments and are also more likely to keep
these appointments stable. Such leaders are also more likely to establish
formal succession rules, which supports studies that claim that revolution-
ary groups are institutionalized. These results remain robust when we split
the sample into different time periods. Appendix Table 5.10 reports results
on the sample that is restricted to the 1960–1991 time period, and
Appendix Table 5.11 reports results on a sample that is restricted to the
post-Cold War period.
5 . 5 di f f e r e n ce s - i n- d i f f er e n c e s d e s i g n : the e n d
of the cold war
I also use the end of the Cold War as an exogenous shock to test
my argument about institutionalization and leader strength using
a differences-in-differences approach. While the regressions reported
157
in Section 5.4 are correlational, the research design presented in this

section can be interpreted more confidently as identifying a causal
effect. After the end of the Cold War, international norms shifted
abruptly in favor of democratization and political liberalization.
Power abruptly and exogenously shifted against autocratic leaders.
In order to remain in power, leaders had to institutionalize. However,
I demonstrate that leaders with access to natural resources institutio-
nalized less compared with leaders without access to natural resources
following the end of the Cold War.
Before presenting results from the differences-in-differences design,
I first show evidence that all autocratic regimes became more institu-
tionalized beginning in the 1990s. Following the end of the Cold War,
autocratic leaders felt increased pressures to implement institutions
that appeared democratic as international norms shifted in favor of
democratization and political liberalization (Bratton and van de Walle
1997; Levitsky and Way 2010).
For the analyses in this section, I retain the time-series structure of my
data. Observations are country–year units, and the main dependent vari-
able is a dummy variable that takes a value of 1 if a particular dimension of
institutionalization was in place for that year. I create a dummy variable,
“cold war”, that takes a value of 1 for years prior to 1991, and a 0
otherwise. For each model, I include country fixed effects, so that I am
comparing each country to itself pre- and post-Cold War.
Results from the baseline analysis are reported in Appendix
Table 5.12. This sample includes all observations from 1960 to
2010. Across the board, leaders were significantly less likely to insti-
tutionalize the regime during the Cold War. However, beginning in
the 1990s, autocratic leaders were much more likely to implement
formal and informal executive constraints. This presents an alterna-
tive perspective on democratization in Africa. As the results show,
one of the striking changes that occurred at the end of the Cold War
is that authoritarian regimes became increasingly institutionalized.
These measures allowed autocratic leaders to stabilize their regimes
in light of changing geopolitical circumstances. Even with the intro-
duction of multiparty elections, many African regimes remain
staunchly nondemocratic today.
I also use the end of the Cold War as an exogenous shock to test my
argument about institutionalization and leader strength using
a differences-in-differences approach. In the 1990s, international
norms abruptly shifted toward favoring democratization and political
liberalization. While few countries became true consolidated
158
5.5 Differences-in-differences Design
democracies during this time, many autocrats suddenly became faced

with increasing pressures to appear more democratic.
In the language of the model, leaders faced an exogenous shock to their
authority at the end of the Cold War, therefore shifting the distribution of
power in favor of elites. Leaders who became weaker institutionalized
their regime in order to remain in power.
Autocrats with access to natural resource rents, however, were
more insulated from this shock compared with their counterparts
who did not have access to such rents. This is because access to oil
provides leaders with a constant source of revenue that does not rely
on the collection of taxes (Dunning 2008). We can therefore expect
leaders with access to natural resources to be stronger compared with
leaders without access to resource rents.
The estimating equation used for the empirical models in this section is
the following:
Y it ¼ α þ βoili þ γpostCold Wart þ δoil postColdWarit þ i þ it

where Y i are my measures of institutionalization, estimated separately.
oili is a dummy variable that takes a value of 1 if the country has access to
oil from 1990 to 2000 – the period immediately after the shift in the
distribution of power following the end of the Cold War – and a 0
otherwise. postColdWart is a dummy variable that takes a value of 1 for
years between 1960 and 1990, and a 1 otherwise. i represents country
fixed effects, and it is the error term. All models include robust standard
errors clustered at the country level. δ is the main parameter of interest, as
it estimates differences in levels of institutionalization before and after the
Cold War for countries with and without oil. We should expect leaders
with access to oil rents to institutionalize at lower rates following the end
of the Cold War, compared to leaders who did not have access to oil rents
after the end of the Cold War.
Figure 5.5 illustrates the results for the successor variable. It shows
that leaders who did not have access to oil institutionalized at much
higher rates compared with leaders who did have access to oil after the
end of the Cold War (along the particular dimension of constitutional
succession rules). As the graphs illustrate, there was a general trend
toward higher levels of institutionalization for both oil rich and nonoil
rich leaders. However, even taking into account generally higher levels
of institutionalization after 1990, nonoil rich leaders, who can be inter-
preted as weaker, institutionalized at much higher rates. The full regres-
sion results are reported in Appendix Table 5.13, and the results show
that the interaction term between oil and post-Cold War is consistently
159
Figure 5.5 Were leaders with access to oil revenues less likely to institutionalize after the end
of the Cold War?
negative. In light of exogenous shocks to the distribution of power,

leaders who are stronger do not need to institutionalize their regimes.
Readers may wonder whether the same difference-in-differences
approach may be applied using changes in access to foreign aid, especially
from the United States or the Soviet Union. The conventional wisdom is
that many African countries received economic and military support from
the major powers during the Cold War, and this foreign aid ceased after
the end of the Cold War. It turns out that this perceived trend is not
empirically supported when we look at the data. Virtually all countries in
sub-Saharan Africa received military and economic aid from the United
States during and after the Cold War. In fact, the amount of aid the United
States provides to most African countries has actually increased
since 1990.
To verify this empirically, I rely on data from USAID,13 which provides
annual reports on economic and military assistance the United States has
provided to other countries since 1945.14 I calculated the total amount of
military and economic aid each country within my dataset received from
the United States from 1960–1990 and from 1990–2010.
13
USAID, https://explorer.usaid.gov/.
14
To the best of my knowledge reliable aid data from the Soviet Union is not available.
160
5.6 Conclusion
The first striking finding is that virtually every country within sub-
Saharan Africa has received foreign aid from the United States during
the Cold War period, and these countries continue to receive foreign aid
from the United States. The second striking finding is that aid flows from
the United States have actually increased since the end of the Cold War,
and this is true for all but two countries in my sample.15 On average, for
countries that received more aid following the end of the Cold War, the
rate of foreign aid received increased by 30 percent. To sum, the United
States provided foreign aid to all countries in sub-Saharan Africa during
the Cold War and continues to provide aid to these countries. Due to the
steady flow of aid from the United States, we cannot use a shock in access
to foreign aid as a proxy for a shift in the distribution of power because an
abrupt shock did not occur.
5.6 conclusion
This chapter provided empirical evidence of the theoretical claims made in
Chapter 2 that there is a systematic relationship between initial leader
strength and levels of regime institutionalization. Using my dataset of
regime institutionalization and several proxies of leader strength based
on the way in which leaders came to power, I tested my theoretical
argument using a cross-sectional approach and a differences-in-
differences approach. I showed that founding presidents, nationalist lea-
ders, and postcoup leaders (all of whom can be interpreted as initially
strong) were less likely to institutionalize their regimes. I also showed that
following the end of the Cold War, leaders with access to oil rents were
less vulnerable in facing pressures to institutionalize.
This chapter treated regime institutionalization as the main depen-
dent variable and addressed the question of under what conditions
leaders will institutionalize. By contrast, the next two chapters will
treat regime institutionalization as the main independent variable and
ask the question of whether institutionalized regimes actually perform
better on a number of key outcomes. We now turn to examining the
effects of regime institutionalization.
15
Togo and Mauritius are the only two countries for which the total amount of aid
received from 1991 to 2010 was lower compared with total amount of aid received
from 1960 to 1990. However, both of these countries continue to receive substantial
foreign aid from the United States.
161
Appendix Description of Aggregate Institutionalization Score

Using my measures of party institutionalization I create an aggregate
institutionalization score using Item Response Theory (IRT) (de Ayala
2009; Embretson and Reise 2000; Hambleton, Swaminathan, and
Rogers 1991; McDonald 1999). IRT is used to measure
unobservable characteristics; in this case, party institutionalization is
the latent variable of interest. Latent traits cannot be measured
directly, but they can be quantified using several items designed to
measure the level of the latent trait; in this case, my six indicators of
institutionalization are used to constraint the latent variable. To
create my score of party institutionalization, I use the “irt” function
in Stata (14.2). The process produces an aggregate score for every
party-year. Summary statistics are presented below.
Observations Mean Standard Min Max

Deviation
Institutionalization score 1,293 .445 .115 .284 .648
162
Appendix Table 5.1 Founding leaders
(1) (2) (3) (4) (5) (6) (7) (8)

VARIABLES Aggregate score Successor Successor (strict) Term limit Second appoint Second same Defense appoint Defense same
First president −0.054** −0.152 −0.287** −0.240** −0.092 −0.075 −0.106 −0.103
(0.025) (0.158) (0.140) (0.093) (0.110) (0.112) (0.136) (0.110)
Military −0.108*** −0.434*** −0.351** −0.224* 0.049 0.001 −0.398** −0.250**
(0.024) (0.151) (0.156) (0.116) (0.193) (0.159) (0.160) (0.119)
Party 0.074*** 0.330* 0.063 0.120 −0.024 −0.069 −0.027 0.041
(0.025) (0.163) (0.199) (0.143) (0.140) (0.125) (0.143) (0.130)
Years in power 0.001 0.011 0.013* −0.004 −0.006 0.003 0.007 0.010

(0.001) (0.009) (0.007) (0.004) (0.009) (0.008) (0.008) (0.006)
GDP per capita 0.041*** 0.201*** 0.029 0.107 0.105* 0.105* −0.010 −0.034
(0.013) (0.063) (0.090) (0.066) (0.054) (0.054) (0.088) (0.048)
Oil production −0.040*** −0.194*** −0.053 −0.123* −0.051 −0.042 0.010 0.027
(0.012) (0.048) (0.082) (0.065) (0.048) (0.049) (0.080) (0.042)
(continued)
(1) (2) (3) (4) (5) (6) (7) (8)

Population 0.015* 0.101 0.054 −0.016 0.056 0.077* 0.110* 0.071

(0.009) (0.063) (0.059) (0.055) (0.053) (0.044) (0.063) (0.045)
Ethnic fractionalization −0.026 0.056 0.149 −0.150 0.302 0.107 −0.103 −0.056
(0.035) (0.277) (0.267) (0.244) (0.260) (0.199) (0.251) (0.178)
Constant 0.268*** −0.671 −0.216 0.462 −0.049 −0.412 −0.290 −0.228
(0.070) (0.596) (0.563) (0.440) (0.533) (0.460) (0.625) (0.426)
Observations 93 62 62 62 62 62 62 62

R-squared 0.433 0.363 0.217 0.194 0.131 0.147 0.204 0.213
Notes: Robust standard errors clustered by country in parentheses. The sample only includes observations from 1960 to 1991 and coup leaders are excluded for comparability
since first presidents were never coup leaders. All models use OLS. The dependent variable for model 1 is a continuous score that was created by the IRT model. The dependent
variable for models 2–8 are the disaggregated dimensions of institutionalization. All the dependent variables are the percentage of years in which the leader institutionalized.
Significance levels: ***p < 0.01, ** p < 0.05, * p < 0.10
Appendix Table 5.2 Strong nationalist leaders
(1) (2) (3) (4) (5) (6) (7) (8)

Nationalist leaders −0.050** −0.200 −0.229* −0.094 −0.306* −0.240* 0.056 −0.022
(0.020) (0.124) (0.125) (0.071) (0.154) (0.132) (0.140) (0.104)
Party 0.057** 0.241 −0.083 0.142 −0.037 0.019 0.064 0.132
(0.025) (0.159) (0.198) (0.120) (0.149) (0.114) (0.199) (0.163)
Years in power 0.001 0.010 0.013* −0.002 −0.005 0.001 0.001 0.004
(0.002) (0.009) (0.006) (0.003) (0.009) (0.007) (0.008) (0.006)
GDP per capita 0.051*** 0.194*** 0.017 0.146* 0.113* 0.127** 0.032 −0.006
(0.016) (0.061) (0.066) (0.078) (0.060) (0.052) (0.082) (0.037)

Oil production −0.045*** −0.169*** −0.013 −0.145* −0.068 −0.074 0.001 0.027
(0.016) (0.049) (0.056) (0.082) (0.049) (0.044) (0.076) (0.032)
Population 0.042*** 0.221*** 0.177** 0.027 0.087 0.090 0.175** 0.133**
(0.013) (0.070) (0.071) (0.049) (0.069) (0.056) (0.083) (0.057)
(continued)
(1) (2) (3) (4) (5) (6) (7) (8)

Ethnic fractionalization −0.043 0.015 0.142 −0.364 0.235 0.067 −0.163 −0.015
(0.077) (0.447) (0.406) (0.370) (0.404) (0.298) (0.470) (0.320)
Constant 0.028 −1.618*** −1.323** 0.018 −0.227 −0.474 −0.925 −0.828*
(0.101) (0.563) (0.607) (0.424) (0.600) (0.501) (0.667) (0.453)
Observations 41 41 41 41 41 41 41 41
R-squared 0.465 0.384 0.262 0.270 0.245 0.265 0.214 0.301

Notes: Robust standard errors clustered by country in parentheses. Sample includes only observations from 1960 to 1991 and only includes leaders from the first regime.
I define first regime as the first continuous regime after independence. As soon as a regime experiences a violent deposition, it drops out of the sample. Leaders are defined as
a “strong independence movement leader” if they started a mass movement or party during the independence era to lobby against the colonial powers. Leaders who were
jailed for pro-independence activities or who were important actors in armed combat are also coded as strong independence movement leaders. This sample does not
include leaders who are coup leaders (by definition this is not possible) or those with a military background. All models use OLS. The dependent variable for model 1 is
a continuous score that was created by the IRT model. The dependent variable for models 2–8 are the disaggregated dimensions of institutionalization. All the dependent
variables are the percentage of years in which the leader institutionalized. Significance levels: *** p < 0.01, ** p < 0.05, * p < 0.10
Appendix Table 5.3 Strong nationalist leaders, controlling for participation in war
(1) (2) (3) (4) (5) (6) (7) (8)

Nationalist leaders −0.058** −0.275** −0.218* −0.064 −0.263 −0.218 −0.011 −0.092
(0.022) (0.128) (0.125) (0.067) (0.155) (0.132) (0.126) (0.093)
Participation in war 0.064* 0.499* 0.060 −0.072 −0.164 −0.020 0.071 0.140
(0.035) (0.290) (0.492) (0.160) (0.257) (0.222) (0.249) (0.264)
Party 0.018 0.034 0.012 0.144 −0.082 −0.082 −0.131 −0.082
(0.039) (0.189) (0.289) (0.219) (0.174) (0.138) (0.166) (0.143)
Years in power 0.002 0.016* 0.010 −0.004 −0.007 0.001 0.009 0.012***
(0.002) (0.009) (0.008) (0.006) (0.009) (0.007) (0.007) (0.004)

GDP per capita 0.049*** 0.215*** 0.023 0.118* 0.067 0.087** 0.060 0.019
(0.015) (0.065) (0.062) (0.069) (0.051) (0.041) (0.076) (0.032)
Oil production −0.044*** −0.203*** −0.017 −0.108 −0.008 −0.023 −0.042 −0.013
(0.015) (0.054) (0.053) (0.068) (0.035) (0.028) (0.071) (0.030)
Population 0.041** 0.237*** 0.188** 0.007 0.050 0.057 0.184** 0.141***
(0.016) (0.076) (0.074) (0.053) (0.058) (0.042) (0.072) (0.043)
(continued)
(1) (2) (3) (4) (5) (6) (7) (8)

Ethnic fractionalization −0.023 0.128 0.070 −0.389 0.235 0.095 0.006 0.155
(0.078) (0.450) (0.441) (0.307) (0.406) (0.274) (0.502) (0.265)
Constant 0.018 −1.936** −1.541* 0.176 0.150 −0.193 −0.805 −0.750*
(0.131) (0.718) (0.861) (0.488) (0.521) (0.388) (0.593) (0.370)
Observations 41 41 41 41 41 41 41 41
R-squared 0.513 0.473 0.283 0.404 0.429 0.458 0.393 0.554
Notes: Robust standard errors clustered by country in parentheses. Sample includes only observations from 1960–1991 and only includes leaders from the first regime.

I define first regime as the first continuous regime after independence. As soon as a regime experiences a violent deposition, it drops out of the sample. Leaders are defined as
a “strong independence movement leader” if they started a mass movement or party during the independence era to lobby against the colonial powers. Leaders who were
jailed for pro-independence activities or who were important actors in armed combat are also coded as strong independence movement leaders. “Participation in war” is
a dummy variable that takes a value of 1 if the leader participated in an independence war, and 0 otherwise. This sample does not include leaders who are coup leaders (by
definition this is not possible) or those with a military background. All models use OLS. The dependent variable for model 1 is a continuous score that was created by the
IRT model. The dependent variable for models 2–8 are the disaggregated dimensions of institutionalization. All the dependent variables are the percentage of years in
which the leader institutionalized. Significance levels: *** p < 0.01, ** p < 0.05, * p < 0.10
Appendix Table 5.4 Coup leaders
(1) (2) (3) (4) (5) (6) (7) (8)

Coup leader −0.08*** −0.31*** −0.23*** −0.15** −0.20** −0.18*** −0.19*** −0.16***
(0.020) (0.080) (0.080) (0.070) (0.080) (0.064) (0.061) (0.045)
Military (non-coup) −0.026 −0.198** −0.106 0.020 0.045 0.079 −0.133 −0.046
(0.025) (0.091) (0.088) (0.111) (0.091) (0.072) (0.093) (0.068)
Party −0.005 0.103 0.010 −0.045 −0.144 −0.157 −0.157 −0.152*
(0.034) (0.131) (0.162) (0.135) (0.115) (0.097) (0.146) (0.086)
Years in power 0.001 0.010** 0.007* −0.004 −0.005 0.004 −0.003 0.006**
(0.001) (0.004) (0.004) (0.005) (0.004) (0.003) (0.004) (0.002)

GDP per capita 0.023*** 0.081*** 0.071*** 0.070*** 0.039*** 0.041*** −0.004 −0.009
(0.006) (0.020) (0.024) (0.020) (0.009) (0.013) (0.016) (0.014)
Oil production −0.019* −0.097** −0.085** −0.045 −0.004 −0.004 0.028 0.019
(0.010) (0.039) (0.036) (0.030) (0.017) (0.016) (0.026) (0.019)
Population 0.013 0.017 0.022 0.046 0.022 0.029 0.106** 0.074**
(0.011) (0.043) (0.044) (0.034) (0.033) (0.026) (0.048) (0.033)
(continued)
(1) (2) (3) (4) (5) (6) (7) (8)

Ethnic fractionalization 0.003 0.025 0.027 −0.018 −0.012 0.038 −0.091 0.001
(0.059) (0.254) (0.269) (0.174) (0.123) (0.092) (0.195) (0.129)
Constant 0.313*** 0.236 0.139 −0.034 0.592** 0.104 −0.104 −0.261
(0.087) (0.349) (0.366) (0.283) (0.273) (0.233) (0.418) (0.284)
Observations 158 158 158 158 158 158 158 158
R-squared 0.210 0.224 0.136 0.117 0.136 0.195 0.144 0.166

Notes: Robust standard errors clustered by country in parentheses. This sample includes the entire time period. All models use OLS. The dependent variable for model 1 is
a continuous score that was created by the IRT model. The dependent variable for models 2–8 are the disaggregated dimensions of institutionalization. All the dependent
Appendix Table 5.5 Coup leaders (Cold War sample)
(1) (2) (3) (4) (5) (6) (7) (8)

Coup leader −0.06*** −0.29*** −0.17** −0.03 −0.25*** −0.20*** −0.18** −0.18***
(0.014) (0.073) (0.073) (0.068) (0.085) (0.064) (0.079) (0.052)
Military (non-coup) −0.062*** −0.347*** −0.196* −0.058 0.059 0.058 −0.313* −0.183*
(0.020) (0.099) (0.108) (0.100) (0.158) (0.138) (0.157) (0.095)
Party 0.081*** 0.429*** 0.206 0.155 0.049 −0.002 −0.054 0.011
(0.025) (0.116) (0.160) (0.118) (0.121) (0.109) (0.104) (0.095)
Years in power −0.000 0.004 0.004 −0.006 −0.012** −0.002 −0.005 0.003
(0.001) (0.007) (0.005) (0.004) (0.006) (0.004) (0.005) (0.004)

GDP per capita 0.054*** 0.211*** 0.102 0.169** 0.087* 0.074 0.012 −0.024
(0.015) (0.055) (0.086) (0.066) (0.051) (0.051) (0.080) (0.042)
Oil production −0.048*** −0.194*** −0.100 −0.164** −0.033 −0.009 0.007 0.032
(0.013) (0.043) (0.080) (0.063) (0.048) (0.050) (0.071) (0.036)
Population 0.013 0.056 0.038 0.013 0.012 0.032 0.112** 0.081**
(0.009) (0.039) (0.035) (0.029) (0.043) (0.030) (0.045) (0.032)
(continued)
(1) (2) (3) (4) (5) (6) (7) (8)

Ethnic fractionalization −0.006 −0.006 0.058 −0.040 0.109 0.006 −0.037 −0.020
(0.038) (0.227) (0.244) (0.171) (0.240) (0.160) (0.271) (0.148)
Constant 0.231*** −0.323 −0.224 −0.068 0.478 0.062 −0.298 −0.328
(0.069) (0.353) (0.311) (0.279) (0.404) (0.299) (0.439) (0.278)
Observations 107 107 107 107 107 107 107 107
R-squared 0.375 0.385 0.158 0.145 0.162 0.176 0.153 0.192

Notes: Robust standard errors clustered by country in parentheses. This sample includes observations from 1960 to 1991. All models use OLS. The dependent variable for
model 1 is a continuous score that was created by the IRT model. The dependent variable for models 2–8 are the disaggregated dimensions of institutionalization. All the
dependent variables are the percentage of years in which the leader institutionalized. Significance levels: *** p < 0.01, ** p < 0.05, * p < 0.10
Appendix Table 5.6 Coup leaders (post-1991 sample)
(1) (2) (3) (4) (5) (6) (7) (8)

Coup leader −0.068** −0.249** −0.211* −0.196 −0.009 −0.067 −0.152 −0.123
(0.028) (0.111) (0.112) (0.131) (0.081) (0.095) (0.104) (0.085)
Military (non−coup) −0.040 −0.225* −0.191 −0.066 0.017 0.033 0.048 −0.035
(0.036) (0.114) (0.117) (0.173) (0.056) (0.076) (0.097) (0.100)
Party −0.059 −0.005 0.025 −0.275 −0.276 −0.248** −0.208 −0.334***
(0.048) (0.183) (0.182) (0.204) (0.196) (0.113) (0.229) (0.120)
Years in power 0.004* 0.018** 0.017** −0.000 −0.000 0.013** 0.000 0.015**
(0.002) (0.008) (0.008) (0.008) (0.007) (0.006) (0.006) (0.006)

GDP per capita 0.011** 0.038** 0.027 0.035* 0.015 0.029*** −0.029 −0.018
(0.004) (0.018) (0.024) (0.019) (0.011) (0.010) (0.022) (0.011)
Oil production −0.019*** −0.105*** −0.092*** −0.028 −0.025 −0.027** 0.038** 0.030**
(0.005) (0.025) (0.026) (0.019) (0.016) (0.012) (0.019) (0.012)
Population −0.006 −0.071 −0.067 0.023 −0.037 −0.002 0.018 0.031
(0.014) (0.058) (0.060) (0.055) (0.039) (0.035) (0.048) (0.033)
(continued)
(1) (2) (3) (4) (5) (6) (7) (8)

Ethnic fractionalization 0.040 0.086 0.074 0.126 0.071 0.137 0.004 0.152
(0.080) (0.274) (0.274) (0.335) (0.124) (0.101) (0.162) (0.150)
Constant 0.529*** 1.233** 1.184** 0.366 1.203*** 0.369 0.706 0.054
(0.118) (0.472) (0.494) (0.471) (0.327) (0.283) (0.424) (0.284)
Observations 80 80 80 80 80 80 80 80
R-squared 0.234 0.272 0.199 0.120 0.136 0.231 0.100 0.236

model 1 is a continuous score that was created by the IRT model. The dependent variable for models 2–8 is the disaggregated dimensions of institutionalization. All the
Appendix Table 5.7 Variation within coup leaders
(1) (2) (3) (4) (5) (6) (7) (8)

High rank −0.034 −0.167 −0.168 −0.085 −0.012 −0.036 0.166* 0.085*
(0.022) (0.106) (0.109) (0.077) (0.097) (0.065) (0.084) (0.049)
Yrs in power 0.003** 0.013** 0.013** 0.001 −0.001 0.009** −0.011** 0.004
(0.001) (0.005) (0.005) (0.006) (0.006) (0.004) (0.004) (0.003)
Constant 0.360*** 0.190*** 0.182** 0.237*** 0.586*** 0.169*** 0.585*** 0.148**
(0.017) (0.063) (0.065) (0.063) (0.091) (0.044) (0.077) (0.054)
Observations 69 69 69 69 69 69 69 69

R-squared 0.063 0.101 0.095 0.011 0.000 0.071 0.056 0.049
Notes: Robust standard errors clustered by country in parentheses. This sample includes only coup leaders. The dependent variable is the percentage of years during which
the leader had a constitution in place. For the independent variable, “high rank” refers to coup leaders with a “general” ranking. OLS used. Significance levels: *** p < 0.01,
** p < 0.05, * p < 0.10
Appendix Table 5.8 High-ranking coup leaders only
(1) (2) (3) (4) (5) (6) (7) (8)

High-rank coup leader −0.087*** −0.355*** −0.288*** −0.150* −0.165 −0.183** −0.088 −0.127**
(0.024) (0.104) (0.098) (0.075) (0.104) (0.084) (0.077) (0.057)
Military (non−coup) −0.004 −0.119 −0.053 0.061 0.104 0.126* −0.067 0.001
(0.024) (0.086) (0.084) (0.105) (0.092) (0.071) (0.089) (0.064)
Party −0.003 0.108 0.011 −0.041 −0.135 −0.153 −0.144 −0.145
(0.036) (0.137) (0.164) (0.137) (0.115) (0.094) (0.150) (0.090)
Years in power 0.002 0.011*** 0.009* −0.004 −0.004 0.005 −0.003 0.006***

(0.001) (0.004) (0.004) (0.005) (0.004) (0.003) (0.004) (0.002)
GDP per capita 0.025*** 0.090*** 0.077*** 0.075*** 0.046*** 0.047*** 0.004 −0.003
(0.005) (0.019) (0.022) (0.019) (0.010) (0.013) (0.018) (0.014)
Oil production −0.020* −0.100** −0.087** −0.046 −0.006 −0.006 0.026 0.018
(0.010) (0.039) (0.036) (0.030) (0.018) (0.017) (0.028) (0.020)
Population 0.015 0.024 0.028 0.049 0.025 0.033 0.106** 0.076**
(0.011) (0.045) (0.045) (0.034) (0.033) (0.027) (0.049) (0.035)
(continued)
Ethnic fractionalization 0.023 0.103 0.084 0.018 0.035 0.081 −0.049 0.038
(0.054) (0.221) (0.245) (0.166) (0.141) (0.114) (0.207) (0.133)
Constant 0.259*** 0.026 −0.019 −0.130 0.470* −0.011 −0.203 −0.356
(0.090) (0.365) (0.365) (0.282) (0.277) (0.244) (0.429) (0.296)
Observations 158 158 158 158 158 158 158 158
R-squared 0.169 0.192 0.126 0.105 0.102 0.167 0.106 0.132
Notes: Robust standard errors clustered by country in parentheses. This sample includes the entire time period. “High-rank coup leader” is a dummy variable that takes
a value of 1 only if the coup leader held the rank of general at the time of the coup; all other observations (including low-ranking coup leaders) are coded as a 0 for this variable.
All models use OLS. The dependent variable for model 1 is a continuous score that was created by the IRT model. The dependent variable for models 2–8 is the disaggregated

dimensions of institutionalization. All the dependent variables are the percentage of years in which the leader institutionalized. Significance levels: *** p < 0.01, ** p < 0.05, *
p < 0.10
Appendix Table 5.9 Rebel groups
(1) (2) (3) (4) (5) (6) (7) (8)

Rebel group 0.066* 0.222* 0.153 0.125 −0.051 −0.005 0.343*** 0.352***
(0.034) (0.114) (0.098) (0.132) (0.089) (0.094) (0.066) (0.067)
Military −0.050*** −0.221*** −0.146** −0.082 −0.102 −0.072 −0.130* −0.070
(0.016) (0.068) (0.063) (0.062) (0.072) (0.063) (0.066) (0.049)
Party −0.013 0.075 −0.008 −0.060 −0.117 −0.140 −0.217 −0.212***
(0.038) (0.136) (0.172) (0.153) (0.132) (0.106) (0.135) (0.077)
Years in power 0.001 0.010** 0.007 −0.005 −0.006 0.003 −0.003 0.005*

(0.001) (0.004) (0.004) (0.004) (0.004) (0.003) (0.004) (0.003)
GDP per capita 0.020*** 0.074*** 0.067*** 0.066** 0.048*** 0.047*** −0.022 −0.027
(0.007) (0.021) (0.020) (0.026) (0.010) (0.014) (0.017) (0.023)
Oil production −0.015 −0.085** −0.075** −0.033 0.002 0.005 0.044 0.038
(0.011) (0.040) (0.037) (0.034) (0.018) (0.022) (0.027) (0.024)
Population 0.015 0.023 0.027 0.053 0.037 0.042 0.103** 0.070**
(0.011) (0.044) (0.043) (0.033) (0.034) (0.030) (0.045) (0.032)
(continued)
Ethnic fractionalization −0.010 −0.007 −0.001 −0.059 −0.060 −0.009 −0.111 −0.020
(0.057) (0.248) (0.261) (0.171) (0.130) (0.101) (0.192) (0.124)
Constant 0.294*** 0.162 0.081 −0.072 0.479 0.014 −0.091 −0.235
(0.090) (0.363) (0.363) (0.282) (0.290) (0.253) (0.397) (0.282)
Observations 158 158 158 158 158 158 158 158
R-squared 0.189 0.209 0.123 0.105 0.087 0.120 0.183 0.226
Notes: Robust standard errors clustered by country in parentheses. This sample includes the entire time period. All models use OLS. The dependent variable for model 1 is
a continuous score that was created by the IRT model. The dependent variable for models 2–8 is the disaggregated dimensions of institutionalization. All the dependent

Appendix Table 5.10 Rebel groups (Cold War sample)
(1) (2) (3) (4) (5) (6) (7) (8)

Rebel group 0.036 0.305** −0.021 −0.133 −0.202 −0.109 0.333** 0.349**
(0.032) (0.134) (0.234) (0.141) (0.138) (0.102) (0.158) (0.145)
Military −0.067*** −0.290*** −0.203*** −0.109 −0.204** −0.145* −0.218** −0.160***
(0.016) (0.069) (0.070) (0.067) (0.090) (0.077) (0.090) (0.051)
Party 0.071** 0.336** 0.216 0.194 0.151 0.064 −0.167 −0.096
(0.032) (0.131) (0.180) (0.148) (0.150) (0.125) (0.114) (0.091)
Years in power 0.001 0.010 0.006 −0.006 −0.014** −0.003 0.001 0.008**
(0.001) (0.006) (0.005) (0.004) (0.007) (0.005) (0.005) (0.003)

GDP per capita 0.054*** 0.210*** 0.107 0.167** 0.095* 0.081 0.005 −0.032
(0.014) (0.046) (0.087) (0.068) (0.053) (0.052) (0.070) (0.035)
Oil production −0.047*** −0.192*** −0.102 −0.162** −0.028 −0.005 0.011 0.039
(0.012) (0.035) (0.080) (0.065) (0.053) (0.054) (0.064) (0.031)
Population 0.018* 0.075* 0.050 0.020 0.020 0.037 0.130*** 0.094***
(0.009) (0.041) (0.035) (0.030) (0.045) (0.034) (0.041) (0.028)
(continued)
Ethnic fractionalization −0.017 −0.031 0.031 −0.070 0.021 −0.060 −0.047 −0.039
(0.037) (0.223) (0.234) (0.159) (0.232) (0.164) (0.269) (0.142)
Constant 0.188** −0.568 −0.337 −0.073 0.458 0.040 −0.501 −0.489*
(0.072) (0.352) (0.309) (0.265) (0.417) (0.311) (0.393) (0.249)
Observations 107 107 107 107 107 107 107 107
R-squared 0.408 0.411 0.168 0.175 0.128 0.119 0.198 0.267

Appendix Table 5.11 Rebel groups (post-1991 sample)
(1) (2) (3) (4) (5) (6) (7) (8)

Rebel group 0.038 0.108 0.187* 0.082 −0.027 −0.004 0.338*** 0.284***
(0.026) (0.098) (0.101) (0.149) (0.092) (0.112) (0.096) (0.072)
Military −0.049* −0.219** −0.168* −0.122 −0.002 −0.022 −0.000 −0.032
(0.025) (0.095) (0.099) (0.123) (0.051) (0.058) (0.105) (0.080)
Party −0.067 −0.023 −0.005 −0.295 −0.273 −0.253** −0.272 −0.384***
(0.047) (0.185) (0.182) (0.204) (0.195) (0.111) (0.201) (0.104)
Years in power 0.003 0.016* 0.014 −0.003 −0.000 0.012** −0.008 0.009

(0.002) (0.009) (0.009) (0.008) (0.006) (0.005) (0.007) (0.006)
GDP per capita 0.009* 0.033** 0.017 0.033 0.016 0.030*** −0.044*** −0.031*
(0.005) (0.016) (0.018) (0.024) (0.014) (0.011) (0.011) (0.018)
Oil production −0.016*** −0.099*** −0.083*** −0.021 −0.025* −0.024** 0.059*** 0.045***
(0.005) (0.023) (0.021) (0.021) (0.014) (0.010) (0.016) (0.014)
Population −0.006 −0.074 −0.074 0.026 −0.035 0.003 0.014 0.023
(0.014) (0.058) (0.057) (0.052) (0.039) (0.037) (0.045) (0.034)
(continued)
Ethnic fractionalization 0.041 0.090 0.083 0.121 0.067 0.129 0.009 0.162
(0.078) (0.271) (0.272) (0.328) (0.125) (0.099) (0.162) (0.147)
Constant 0.532*** 1.262** 1.241** 0.347 1.184*** 0.332 0.751* 0.120
(0.114) (0.478) (0.480) (0.447) (0.325) (0.306) (0.388) (0.291)
Observations 80 80 80 80 80 80 80 80
R-squared 0.238 0.278 0.216 0.114 0.136 0.218 0.147 0.294

Appendix Table 5.12 Were autocratic regimes less institutionalized during the Cold War?
(1) (2) (3) (4) (5) (6) (7) (8)

Cold War −0.112*** −1.888*** −2.721*** −3.684*** −2.856*** −1.067*** −1.873*** −0.844***
(0.014) (0.536) (0.539) (0.571) (0.508) (0.245) (0.411) (0.233)
Constant 0.646*** 2.581*** 3.415*** 4.378*** 4.107*** 3.631*** 4.512*** 1.537***
(0.014) (0.536) (0.539) (0.571) (0.479) (0.245) (0.411) (0.233)
Observations 1,751 1,198 1,280 1,517 1,468 1,664 1,477 1,660
Country FE Yes Yes Yes Yes Yes Yes Yes Yes

R2 0.538 0.247 0.327 0.441 0.360 0.206 0.277 0.181
Notes: Robust standard errors clustered by country in parentheses. For models 2–8, pseudo R-square statistics are reported. I retained the panel structure of the data for these
regressions. Model 1 uses OLS and models 2–8 are logit models. The dependent variable for models 2–8 is a dummy that takes a value of 1 if that dimension of
institutionalization was in place for that year. The dependent variable for model 1 is a continuous score that was created by the IRT model. *** p < 0.01, ** p < 0.05, *
p < 0.10
Appendix Table 5.13 Differences-in-differences test
(1) (2) (3) (4) (5) (6) (7) (8)

Post-Cold War 0.121*** 2.640*** 4.757*** 3.913*** 3.457*** 1.302*** 2.301*** 0.855***
(0.020) (0.893) (0.994) (0.810) (0.634) (0.296) (0.594) (0.277)
Oil −0.013 −0.439 −0.286 0.367 1.205*** 1.189*** 0.077 0.601**
(0.013) (0.305) (0.240) (0.990) (0.129) (0.273) (0.228) (0.266)
Oil*post-Cold War −0.027 −1.848* −3.57*** −0.512 −1.578 −0.643 −1.038 −0.029
Constant 0.390*** −0.126 −0.44*** −3.596*** 0.134* −1.753*** 0.148 −1.619***
(0.011) (0.230) (0.064) (0.747) (0.074) (0.212) (0.160) (0.184)

Country FE Yes Yes Yes Yes Yes Yes Yes Yes
Observations 1,751 1,751 1,751 1,751 1,751 1,708 1,751 1,745
Notes: Robust standard errors clustered by country in parentheses. Model 1 uses OLS and models 2–8 are logit models. The dependent variable for models 2–8 is a dummy
that takes a value of 1 if that dimension of institutionalization was in place for that year. The dependent variable for model 1 is a continuous score that was created by the
IRT model. *** p < 0.01, ** p < 0.05, * p < 0.10
6
What Are the Consequences of Institutionalization
This book has so far been primarily been concerned with explaining the
emergence of institutionalized autocratic rule. We now return to the first
central puzzle presented at the very onset of Chapter 1: What explains
differences in authoritarian regime stability? Why is it that some dictator-
ships are subject to constant coup attempts and regime change, while
others withstand multiple peaceful leadership transitions and remain
stable over long periods of time?
The next two chapters will examine the effects of institutionalization on
regime durability. Do institutionalized regimes actually perform better on
a number of key outcomes, such as leader longevity, the prevention of coups,
and the promotion of peaceful leadership succession? This chapter will focus
on estimating the effects of regime institutionalization on key leader out-
comes: the length of leader tenure and susceptibility to coups attempts. The
next chapter will focus on estimating the effects of institutionalization on key
regime outcomes: leadership succession and regime survival.
This chapter will show that leaders who institutionalize their regimes
tend to remain in office for longer periods of time and face fewer coup
attempts. The next chapter will demonstrate that institutionalized regimes
are significantly more likely to undergo peaceful leadership transitions.
Together these two chapters demonstrate that executive constraints are
not simply empty parchment rules. Institutionalization does indeed have
consequences for regime stability.
However, estimating the effects of endogenously created institutions is not
as straightforward as it may seem. In determining the effects of institutions,
these chapters will take into consideration the subject of the first half of this
book – the fact that leaders choose whether to institutionalize based on
underlying threats to their rule. Since constitutional rules and cabinet
appointments are not randomly assigned across regimes, we must consider
the types of leaders who choose to create these institutions when examining
whether such rules have an independent effect on regime durability.
186
6.1 Are Rules Followed in Dictatorships?
Existing work on the empirical determinants of regime stability has

largely ignored the empirical challenge of estimating the effect of endo-
genous institutions, even though much of the scholarship does indeed
acknowledge theoretically that dictators choose whether or not to allow
for institutional constraints. This chapter argues that to determine
whether institutions play a stabilizing role, we cannot simply regress
regime outcomes on levels of institutionalization.
Recall that the formal model in Chapter 2 demonstrated how leaders
who enter office already strong can remain in power without institu-
tionalizing the regime. Such leaders will therefore be associated with
stable regimes and no institutionalization. Weak leaders, on the other
hand, must institutionalize in order to maintain their hold on office.
Therefore, if we focus on the subset of leaders who enter power weak,
we should observe a positive effect of institutionalization on regime
stability. Yet, without taking into account leader type, simply regres-
sing regime outcomes on institutionalization would result in biased
estimates because strong leaders do not need to institutionalize to stay
in power.
In this chapter I demonstrate that when we condition on initial leader
strength, institutionalization does indeed have a stabilizing effect for weak
leaders but little or no effect for strong leaders. This approach helps to
make sense of the existing divergent empirical findings on the question of
whether institutions promote regime stability. It heeds Pepinsky’s (2014)
call to deal with the “nagging problem” facing the literature on author-
itarian institutions: If “institutions under authoritarian rule are vulnerable
to manipulation because political actors believe that institutional manip-
ulation can shape political outcomes in their favor, then it is also true that
factors that explain the origins of (and changes in) [these institutions] also
directly affect those political outcomes” (631).
6. 1 are constitutional rules actually

followed i n dictatorships?
Before we discuss the effects of institutionalization on regime outcomes, it
is important to first address an important descriptive question: once
established, are constitutional rules followed in dictatorships? If we are
to believe that formal rules promote regime stability, then we should
observe leaders and elites actually following the letter of the law. To
answer this question, I focus on the constitutional rules in my dataset:
leadership succession procedures and term limits. I find that constitutional
leadership succession procedures are generally adhered to in my sample,
while term limits are violated much more frequently.
187
6 Consequences on Autocratic Stability
To examine whether constitutional leadership succession rules were

followed, I identified all instances of peaceful leader exit, whether by
death or voluntarily retirement. For all instances of peaceful leader exit,
I then identify whether a constitutional succession rule existed at the
time of leader exit. The procedure results in thirty-four cases that allow
us to observe whether the succession policy was followed at the time of
leader exit. Cases of leader departure but no succession rule were
excluded because the question of whether succession rules would have
been followed is unobserved in such instances. Cases of violent leader
deposal were also excluded because coups result in new regimes rather
than the continuation of the existing regime via succession. I also exclude
cases of democratization after the death of the leader: in these instances
the next leader was elected via elections, rendering the succession policy
irrelevant.
For the majority of the observations, leadership succession occurs
according to constitutional rules. Out of thirty-four observations, 79 per-
cent consist of peaceful leadership transitions where the constitutional
policy was followed exactly – in particular, the designated successor
became the interim president following the death or departure of the
leader. In 21 percent of these cases, a peaceful leadership transition did
occur, but the interim president was not the designated successor who was
specified in the constitution. Summary statistics are listed in Table 6.1.
To determine whether term limits were followed, I identified all leaders
who had term limits in the constitution and remained in power long
enough to observe whether they would adhere to those limits. This
resulted in forty-five observable cases. Out of these forty-five cases, leaders
adhered to term limits 29 percent of the time. To summarize, while
constitutional leadership succession policies were indeed followed in
most observable cases, leaders seem to violate term limits much more
frequently. The next chapter will provide a theory of why term limits
seem to be easier to violate compared to constitutional succession policies.
Since succession rules are generally followed but term limits are fre-
quently ignored, I exclude term limits from my analysis of the effects of
constitutional rules on regime stability for the remainder of this chapter.
Table 6.1 Summary statistics
Policy Observable cases Rule followed As percent
Succession rules 34 27 79%

188
6.2 Do Institutions Promote Regime Stability?
6 . 2 e m p i r i cal d i s ag re em en t : do i ns t i t ut i ons
promote regime s tabili ty?
Despite the general consensus that institutionalized forms of dictatorship
tend to be more stable, the empirical record of the relationship between
institutions and regime stability has been surprisingly mixed. In a seminal
study, Geddes (1999a) finds that single-party regimes remain in power for
longer periods of time compared to personalist regimes. Military regimes,
which tend to have weak or no civilian institutions, are less durable
compared with personalist regimes. This study catalyzed the idea that
institutionalized forms of authoritarianism tend to be most durable.
Geddes’ empirical analysis was replicated by Brownlee (2007), who
found similar effects. Since then, many scholars have made the theoretical
argument that institutions, such as parties and legislatures, can prolong
the lifespan of authoritarian regimes (Boix and Svolik 2013; Gandhi 2008;
Geddes, Wright, and Frantz 2018; Gandhi and Przeworski 2007;
Magaloni 2008; Svolik 2009).
However, a number of subsequent studies have since challenged this
initial finding. Smith (2005), for instance, shows that the significant effect
of single-party regimes is driven primarily by two outliers: Mexico and the
Soviet Union. He replicates Geddes’ original analysis, producing similar
results, but then demonstrates that when the two longest-living regimes
are dropped from the analysis, single-party regimes are no longer signifi-
cantly more durable compared with personalist regimes.
A small number of studies have also pointed out that autocratic leaders
endogenously respond to threats with “appropriate” degrees of institutiona-
lization, therefore we should not expect to see a relationship between institu-
tions and stability. Gandhi (2008), for instance, finds that leaders with ruling
parties do not tend to survive in office longer compared with those without
parties, “suggesting that most authoritarian incumbents are able to perceive
with some accuracy the threats germinating within society and to respond
with the appropriate degree of institutionalization” (176–177). According to
this study, leaders institutionalize optimally, and therefore there should be no
observable effect between institutions and stability.
Boix and Svolik (2013) also present a study in which institutional choice
is endogenous. They argue that power-sharing institutions, such as parties
or legislatures, can be established only when backed by a credible threat of
elite rebellion. However, despite the fact that their theory endogenizes the
creation of institutions, two observable implications of their study are that
regimes with institutions should be longer lived and experience fewer
coups. Using data on the existence of parties and legislatures in autocra-
cies, they find empirical support for their arguments.
189
To sum up, the existing literature on the relationship between auto-

cratic institutions and regime stability has produced a confusing array of
theoretical predictions and empirical findings. Two questions remain
unresolved. First, should we expect, theoretically, to see an effect
of endogenously created institutions on autocratic regime stability?
Second, what does the data show us empirically?
6.3 p roblems with e xisting approaches

How do we make sense of these contrasting theories and divergent
empirical findings? I argue that there are two possible reasons why exist-
ing empirical studies of the relationship between autocratic institutions
and regime durability may be biased: (1) measurement problems and (2)
theoretical problems.
First, existing empirical findings may be subject to a number of mea-
surement problems due to the difficulties of obtaining fine-grained data
that accurately reflects the quality of authoritarian institutions. As
Chapter 4 demonstrated, the data on autocratic regime types (Geddes,
Wright, and Frantz 2014) has a number of inconsistencies when compared
with this book’s dataset on executive constraints. In particular, many
single-party regimes, which are supposed to be the most institutionalized
type of autocracy, appear to lack many dimensions of formal and informal
constraints on presidential power. Many regimes that are coded as mili-
tary and personalist, which are supposed to be less institutionalized, do
actually appear to have many executive constraints in place.
On the other hand, empirical studies (Boix and Svolik 2013; Gandhi
2008) that use the existence of institutions – such as the presence of
a single, multiple, or no political parties – may also be subject to measure-
ment problems, due to the fact that such data often does not capture
detailed information about the quality of the institutions. Therefore it is
unclear whether the lack of an effect of the presence of institutions on
regime durability is due to a true lack of an effect or whether the null
findings are due to measures that do not capture institutional quality.
Beyond measurement problems, I also argue that existing empirical
studies on the relationship between autocratic institutions and regime
durability are also subject to biased theoretical modeling. In particular,
while many existing studies acknowledge the fact that leaders choose to
create or allow institutions in order to promote regime stability, this
endogenous institutional choice is not taken into account when estimating
the effect of institutions on outcomes, such as regime stability or long-
evity. As Pepinsky (2014) notes, “factors that explain the origins of (and
changes in) [institutions] directly affect those political outcomes” (631).
190
6.3 Problems with Existing Approaches
I argue that we need to differentiate between initially strong and initially

weak leaders when considering the effects of institutionalization on regime
durability, since these different types of leaders make different institutional
decisions. Since leaders choose to institutionalize based on existing threats to
their rule, the creation of executive constraints should only help initially
weak leaders stay in power longer than they otherwise would have.
Initially strong leaders who have already consolidated power by the time
they take office do not need to rely on power-sharing institutions to remain in
office. Their rule is secure with or without regime institutionalization. We
therefore should generally not observe a relationship between party institu-
tionalization and regime durability for strong leaders.1
For leaders who are weak, however, building an institutionalized regime
should promote regime stability. Leaders who are highly vulnerable to being
deposed by their fellow elites require institutional mechanisms in order to
make credible commitments to share power and resources. Institutionalized
regimes should therefore lengthen the time horizons of initially vulnerable
leaders. Initially strong leaders, on the other hand, have no incentive to
institutionalize their regimes to begin with. Figure 6.1 summarizes my
argument.
Figure 6.1 illustrates why simply regressing institutionalization on
regime stability would produce biased estimates of the effect of institu-
tions on stability. For initially weak leaders, those who institutionalize
should experience more stable regimes. For initially strong leaders,
however, there should be little or no effect of institutionalization on
regime durability because strong leaders should be able to remain in
power regardless of whether they institutionalize. A regression model
Strong No
leader Stable regime
institutionalization
No Unstable
Institutionalization regime
Weak
leader
Institutionalization Stable regime
Figure 6.1 What is the effect of institutionalization on regime stability?
1
It is important to note that it is possible for strong leaders to institutionalize, even
though this would be out-of-equilibrium behavior according to the model. I would
expect institutionalized regimes, even if the leader was initially strong, to be more
stable due to the fact that the creation of constitutional rules and cabinet appoint-
ments creates mechanisms for credible power-sharing and patronage distribution.
191
that regresses institutionalization on regime stability without taking into

Hypothesis 1: Institutionalization has a positive effect on regime
stability, conditional on leader weakness.
6.4 e stimating t he effects o f e ndogenous

How should we estimate the effects of institutionalization on regime
stability without ignoring the origins of these institutions? One approach
is to incorporate an interaction term between regime institutionalization
and leader strength which will account for differential effects of institu-
tionalization based on whether the leader was initially strong or weak.
6.4.1 Data
The outcome variable of interest in this chapter is autocratic stability,
particularly with respect to leader outcomes.2 Leader stability will be
measured via two dependent variables: the length of a leader’s tenure
and the number of coup attempts a leader faces during her tenure.
“Years in power” is a variable that counts the number of years a leader
remained in power. “Coup attempts” records the percent of the number of
office years for which a leader faced coup attempts. This variable is
presented as a percent, rather than a count of the number of coup attempts
a leader faced in order to take into account the fact that various leaders
remained in office for different lengths of time.
There are two primary independent variables of interest: leader
strength and regime institutionalization. Leader strength will be proxied
using two measurement strategies from Chapter 5: founders versus suc-
cessors and coup leaders versus noncoup leaders. “Founder” is a dummy
variable that takes a value of 1 if the leader was the first leader of the
country after independence and 0 otherwise. As Chapter 5 argued, found-
ing leaders tended to enjoy special legitimacy and mass support, and this
was especially true for these leaders who rose to power in the immediate
aftermath of independence. Founding presidents can therefore be inter-
preted as stronger leaders compared with subsequent presidents. “Coup
2
The next chapter will examine the effect of institutionalization on leadership succes-
sion and regime survival.
192
6.4 Estimating Effects of Endogenous Institutions
leader” is a dummy variable that takes a value of 1 if the leader came to

power through a coup d’etat and 0 otherwise. As Chapter 5 argued,
leaders who successfully take power through a coup demonstrate that
they have control over the coercive apparatus that allowed them to launch
a successful coup in the first place. Since one of the most precarious
challenges to any civilian leader is the threat of military takeover, when
the leader is the head of the military this threat is diminished. Coup leaders
can therefore be interpreted as strong leaders, compared with leaders who
did not come to power via a coup.3
The second primary independent variable of interest is level of regime
institutionalization. I use my indicators of regime institutionalization,
measured separately. For this chapter, I use the percent of years for
which the leader had a succession policy in place (“successor”), stable
VP/PM appointment (“VP/PM same”), and stable defense appointment
(“Defense same”).4 Each of these measures take a value from 0 to 1, with
high values indicating a higher level of regime institutionalization.
6.4.2 Summary Statistics

Before turning to the empirical strategy, I first present summary statistics of
the leader outcome variables: length of leader tenure and frequency of coup
attempts. The average leader in my sample was in power for 8.48 years, with
the shortest tenure being one year and the longest tenure being forty-one
years. Appendix Table 6.1 presents summary statistics of this variable. On
average, founding leaders remain in power for 13 years, while subsequent
leaders remain in power for 7.4 years. The average coup leader remains in
office for 8.2 years and the average noncoup leader remains in power for 8.6
years. This finding runs contrary to existing scholarship portraying coup
leaders as especially weak or susceptible to falling into coup traps, yet is
3
The reader may recall that in Chapter 5 I presented a third strategy for measuring
leader strength by looking at the subset of founding leaders, and comparing strong
nationalist leaders against those who were close allies of the colonial state. Since this
strategy requires looking within the subset of independence regimes, the number of
observations available for this strategy are much more limited. I therefore exclude
this strategy from the empirical tests in this chapter.
4
Since the strict and weak version of the successor rule are highly correlated, I exclude
the “strict successor” variable from the analysis in this chapter. Since “VP/PM
appoint” and “VP/PM same” as well as “defense appoint” and “defense same” are
also highly correlated, I also exclude the “VP/PM appoint” and “defense appoint”
variables from the analysis in this chapter. Since term limits were generally not
implemented until after the 1990s and are not frequently adhered to, I exclude it
from the analysis in this chapter.
193
consistent with my argument that leaders endogenously institutionalize

based on threat perceptions.
Turning to the other dependent variable, the average leader faces
a (successful or unsuccessful) coup attempt for 19 percent of the years
she is in power. Appendix Table 6.2 presents summary statistics of this
variable. In other words, if a leader were in power for ten years, she would
face a coup attempt in roughly two of the ten years.
There are significant differences in the percentage of years for which
strong leaders face coups, relative to weak leaders. Founders face coup
attempts for about 5 percent of the years they are in power, compared with
subsequent leaders who face coup attempts for about 22 percent of the
years they are in power. Coup leaders face coup attempts about 35 percent
of the of the years they are in power, compared with noncoup leaders who
face coup attempts for about 11 percent of the years they are in power.
It is important to remind the reader that the analysis conducted in this
chapter does not focus on these baseline differences in the length of leader
tenure or vulnerability to coups. Instead, what we are interested in is
whether institutionalization affects these outcomes within the subgroup
of weak leaders.
6.4.3 Empirical Strategy

Let Y represent the outcome of interest – leader tenure and percent of
years with coup attempts. Let x1 represent a dummy variable that takes
a value of 1 if the leader was initially strong, and 0 otherwise. Let x2
represent the degree of institutionalization, which is measured as
the percent of years for which a particular dimension of institutionaliza-
tion was in place during a leader’s tenure. x2 takes a value between 0 and
1, and higher values indicate more institutionalized regimes. Let
W represent a vector of control variables, including mean GDP per capita,
oil production, population size, and ethnic fractionalization. ε represents
the error term. Since the outcome variables and leader strength variable of
interest are cross-sectional, our unit of analysis will be the leader.
Y ¼ α þ β 1 x1 þ β 2 x2 þ β 3 x1 x2 þ W þ ε
Recall that the hypothesis we’re interested in testing is that institutionali-
zation has a positive effect on regime stability, conditional on leader
weakness. This means that we should observe better regime outcomes
for weak leaders but no relationship between institutionalization and
regime outcomes for strong leaders.
How should we interpret the regression coefficients? We break down
our analysis into two sets of cases: those with a strong leader and those
194
6.5 Empirical Analysis and Findings
with a weak leader. For x1 =0, the subset of weak leaders (recall that x1 is
a dummy variable representing leader strength or weakness), the equation
reduces to the following:
Y ¼ α þ β2 x2 þ ε
I hypothesize that when the leader is weak, institutionalization should
have a positive and significant effect on leader outcomes. Since x2 denotes
levels of regime institutionalization, we should expect β2 to be signifi-
cantly different than zero.5
For x1 =1, the subset of strong leaders, the equation reduces to the
following:
Y ¼ α þ β1 þ ðβ2 þ β3 Þx2 þ ε
I hypothesize that when the leader is strong, there should be no effect of
institutionalization on leader stability. Since x2 denotes levels of regime
institutionalization, then the coefficients ðβ2 þ β3 Þ should not be signifi-
cantly different from 0.
To summarize, if my theory is correct, then we should expect β2 to be
significantly different from 0 and ðβ2 þ β3 Þ not to be significantly different
from 0.
6 . 5 e m p i r i c a l ana l y s i s a n d fin d i n g s
I use two different proxies of leader strength to condition the differential
effects of institutionalization on length of leader tenure and vulnerability to
coups: (1) founding leaders versus subsequent leaders and (2) coup leaders
versus noncoup leaders. I will present results using both of these strategies
separately.
6.5.1 What Is the Effect of Institutionalization on the Length

of Leader Tenure?
As hypothesized in Section 6.3, we should expect institutionalization to
lengthen the time in office of initially weak leaders but have no effect on
the tenure of initially strong leaders. Therefore we should expect β2 > 0
and ðβ2 þ β3 Þ to not be significantly different from 0. To estimate these
5
Whether β2 should be greater than or less than 0 depends on the outcome we are
examining. A lower number of coup attempts denotes a more stable regime, so for
that analysis, we should expect β2 < 0: Longer leader tenure denotes a more stable
regime, so for that analysis, we should expect β2 > 0.
195
Figure 6.2 What is the effect of institutionalization on leader tenure (founders)?

Note: 90 percent confidence intervals reported. Proxy for leader strength is founders versus
subsequent leaders. The dependent variable is the number of years for which a leader was in
power. Our hypothesis predicts a positive relationship between institutionalization and
leader tenure for weak leaders.
effects, we include an indicator of leader strength and an interaction term

between leader strength and institutionalization.
First I use founding leaders as a proxy for initially strong leaders. The
results of these regressions are illustrated in Figure 6.2, and the full
regression results are reported in Appendix Table 6.3. When weak leaders
(nonfounders) institutionalize, they tend to remain in office for longer
periods, and this result is statistically significant for all three dimensions of
institutionalization: succession procedures, having a stable VP/PM, and
having a stable defense minister.6
The results also show little or no effect of institutionalization on leader
tenure for strong leaders (founders).7 Designating a stable VP/PM and
defense minister does not appear to have a statistically significant effect on
leader tenure. Having a constitutional succession policy does have
a statistically significant effect of lengthening leaders’ time in office, even
when the leader is initially strong. This could be due to some measurement
6
Recall that (ß2) is the relevant coefficient for weak leaders. These rows are
highlighted in Appendix Table 6.3.
7
Recall that the effect of institutionalization for strong leaders is denoted by ß2 + ß3.
This row is highlighted in Appendix Table 6.3.
196
Figure 6.3 What is the effect of institutionalization on leader tenure (coup leaders)?
Note: 90 percent confidence intervals reported. Proxy for leader strength is coup leaders
versus noncoup leaders. The dependent variable is the number of years for which a leader
was in power. Our hypothesis predicts a positive relationship between institutionalization
and leader tenure for weak leaders.
error for the leader strength variable. It could be the case that founding
leaders does not serve as a perfect proxy for initial leader strength. On other
hand, it is also possible that having a leadership succession policy in place
does indeed lengthen the tenure of all leaders, even those who enter office
initially strong. Chapter 7 provides an extensive discussion of leadership
succession.
I also analyze the relationship between institutionalization and leader
tenure using coup leaders as a proxy for initially strong leaders. The
results are illustrated in Figure 6.3 and the full regression results are
reported in Appendix Table 6.4. Once again, I find that when initially
weak leaders (noncoup leaders) institutionalize, they tend to remain in
office for longer periods of time, and this finding is statistically significant
along cabinet stability dimensions.
For initially strong leaders (coup leaders), institutionalization – particu-
larly the stable appointment of VP/PMs and the creation of succession
policies – does appear to have effects on leader tenure. It is possible that
coup leaders, while having consolidated authority over the coercive appara-
tus of the regime, have vulnerability along other nonmilitary dimensions, and
therefore benefit from the stabilizing effect of institutionalization along such
dimensions. Coup leaders who institutionalize along civilian dimensions may
197
experience more stable rule, even if they took power having already con-
solidated power along coercive dimensions.
6.5.2 What Is the Effect of Institutionalization on the Frequency

of Coup Attempts?
Does institutionalization prevent coup attempts? First I use founding
presidents as a proxy for strong leaders. The results from these regressions
are illustrated in Figure 6.4, and the full regression results are reported in
Appendix Table 6.5. Initially weak leaders (non-founding presidents) who
institutionalize are less likely to face coup attempts, and the results are
statistically significant across all dimensions of institutionalization. There
does not appear to be an effect of institutionalization on coup threats for
initially strong leaders (founding presidents) and this is true for all dimen-
sions of institutionalization.
Finally, I also analyze the effect of institutionalization on coup attempts
using postcoup leaders as a proxy for strong leaders. The results are illu-
strated in Figure 6.5, and the full regression results are reported in Appendix
Table 6.6. Initially weak leaders (noncoup leaders) who institutionalize are
significantly less likely to face coup attempts, and this is true for all
Figure 6.4 What is the effect of institutionalization on the frequency of coup attempts?
subsequence leaders. The dependent variable is the percentage of years for which a leader
faces a coup attempt. Our hypothesis predicts a negative relationship between
institutionalization and coup attempts for weak leaders.
198
6.6 Conclusion
Figure 6.5 What is the effect of institutionalization on coup vulnerability?

versus noncoup leaders. The dependent variable is the percentage of years for which a leader
dimensions of institutionalization. There does not appear to be a significant

effect of regime institutionalization on coup threats for initially strong leaders
(coup leaders). The one exception to this general finding is that initially
strong leaders who have constitutional leadership succession procedures
appear to be significantly less likely to face coup attempts. This may be due
to the fact that, as discussed previously, while coup leaders enter power
having already consolidated coercive authority, they may be more vulnerable
along other civilian dimensions. Therefore implementing succession policies
may lessen coup risk, even for leaders who took power via a coup.
6.6 conclusion
This chapter examined the effect of institutionalization on autocratic sta-
bility, particularly the length of leader tenure and frequency of coup threats.
I argued that in estimating the effect of institutionalization on these out-
comes we need to take into account the fact that initially strong and initially
weak leaders make different institutionalization decisions, depending on
threats to their rule. Weak leaders institutionalize in order to alleviate
threats to their rule, and therefore we should expect to see an effect of
institutionalization on stability for initially weak leaders. Initially strong
leaders are able to remain in power whether they institutionalize or not, so
199
we should expect to see little or no effect of institutionalization on stability

for strong leaders. Using founding presidents and postcoup leaders as my
proxies for leader strength, I demonstrate that these patterns are indeed
born out in the data. The next chapter will consider the effect of institutio-
nalization on leadership succession.
Appendix
Appendix Table 6.1 Summary statistics: length of leader tenure
DV: length of leader tenure Mean SE N
Whole sample 8.483 8.224 207

Measure of leader strength: founder vs subsequent leaders
Founder (strong leader) 13.000 1.320 40
Nonfounder (weak leader) 7.401 0.606 167
Measure of leader strength: coup leader vs noncoup leaders
Coup leader (strong leader) 8.214 1.061 70
Noncoup leader (weak leader) 8.620 .674 137
Note: Dependent variable is a count variable of the number of years the leader was in power.
Appendix Table 6.2 Summary statistics: frequency of coups
Whole sample .190 .301 206

Founder (strong leader) .051 .013 40
Nonfounder (weak leader) .223 .025 166
Coup leader (strong leader) .349 .043 69
Noncoup leader (weak leader) .109 .019 137
Note: Dependent variable is the percentage of years in which a leader faced a coup attempt
while in power.
200
6.6 Conclusion
Appendix Table 6.3 What is the effect of institutionalization on leader tenure

(leader strength proxy: founder versus successor)?
DV: number of years leader is in power (1) (2) (3)
Founder 7.388*** 4.781** 4.886**

(2.323) (2.013) (2.263)
VP/PM same (ß2) 4.999***
(1.571)
Founder*VP/PM same (ß3) −4.000
(4.252)
Successor (ß2) 2.770*
(1.474)
Founder*successor (ß3) 2.525
(3.265)
Defense same (ß2) 4.620**
(1.945)
Founder*defense same (ß3) 1.631
(4.345)
GDP per capita −0.227 −0.242 −0.008
(0.358) (0.406) (0.348)
Oil production 1.658*** 1.870*** 1.493***
(0.508) (0.567) (0.468)
Population 0.287 0.500 0.086
(0.651) (0.744) (0.698)
Ethnic fractionalization −0.343 −1.181 −0.491
(1.751) (1.931) (1.904)
Constant 3.027 2.207 5.093
(5.570) (6.494) (5.839)
ß 2 + ß3 0.999 5.295* 6.251
(3.860) (2.924) (4.171)
Observations 194 194 194
R-squared 0.153 0.155 0.155
Note: Robust standard errors in parentheses clustered by country. The effect of

institutionalization for weak leaders is denoted by (ß2). The effect of institutionalization
for strong leaders is denoted by (ß2 + ß3). P-value and standard errors of (ß2 + ß3) calculated
using Stata’s “lincom” function. *** p < 0.01, ** p < 0.05, * p < 0.10
201

(leader strength proxy: coup leaders versus noncoup leaders)?
DV: years in power (1) (2) (3)
Coup leader −0.169 −0.189 0.875

(1.830) (1.413) (1.646)
VP/PM same (ß2) 3.808*
(2.227)
Founder*VP/PM same (ß3) 3.402
(4.470)
Successor (ß2) 2.700
(1.884)
(2.883)
(2.650)
Founder*defense same (ß3) −0.714
(4.445)
GDP per capita −0.119 −0.120 0.087
(0.356) (0.412) (0.346)
Oil production 1.452*** 1.648*** 1.259**
(0.535) (0.595) (0.503)
Population −0.038 0.251 −0.308
(0.686) (0.762) (0.716)
Ethnic fractionalization 0.553 0.122 0.298
(2.369) (2.468) (2.413)
Constant 6.504 4.311 8.213
(5.735) (6.504) (5.888)
ß 2 + ß3 7.21* 6.426** 5.483
(3.924) (2.565) (3.623)
R-squared 0.078 0.082 0.092

202
6.6 Conclusion
Appendix Table 6.5 What is the effect of institutionalization on coup

vulnerability (leader strength proxy: founder versus successors)?
DV: coup attempts (1) (2) (3)
Founder −0.222*** −0.227*** −0.201***

(0.045) (0.058) (0.058)
VP/PM same (ß2) −0.235***
(0.065)
Founder*VP/PM same (ß3) 0.263***
(0.065)
Successor (ß2) −0.155***
(0.051)
Founder*successor (ß3) 0.200***
(0.067)
Defense same (ß2) −0.232**
(0.098)
Founder*defense same (ß3) 0.261**
(0.121)
GDP per capita −0.017* −0.019** −0.029***
(0.009) (0.009) (0.006)
Oil production 0.006 −0.002 0.017
(0.011) (0.014) (0.012)
Population −0.019 −0.038** −0.016
(0.018) (0.018) (0.015)
Ethnic fractionalization 0.103 0.144** 0.107**
(0.068) (0.061) (0.051)
Years in power −0.011*** −0.011*** −0.011***
(0.002) (0.002) (0.002)
Constant 0.511*** 0.641*** 0.477***
(0.148) (0.146) (0.123)
ß2 + ß3 0.028 0.045 0.029
(.048) (.052) (.070)
R-squared 0.243 0.234 0.239

203

vulnerability (leader strength proxy: coup leaders versus noncoup leaders)?
Coup leader 0.039 0.100* 0.053

(0.038) (0.059) (0.059)
VP/PM same (ß2) −0.096**
(0.039)
Founder*VP/PM same (ß3) 0.158*
(0.082)
Successor (ß2) −0.071
(0.049)
Founder*successor (ß3) −0.079
(0.073)
(0.050)
(0.085)
GDP per capita −0.002 −0.004 −0.008
(0.008) (0.008) (0.008)
Oil production 0.011 0.004 0.013
(0.011) (0.011) (0.010)
Population −0.010 −0.023* −0.014
(0.013) (0.013) (0.013)
(0.062) (0.055) (0.059)
Years in power −0.008*** −0.007*** −0.008***
(0.002) (0.001) (0.002)
Constant 0.266** 0.361** 0.297**
(0.128) (0.134) (0.124)
ß 2 + ß3 .062 − .149*** .025
(.072) (.053) (.071)
R-squared 0.265 0.303 0.276

204
6
This book has so far been primarily been concerned with explaining the
emergence of institutionalized autocratic rule. We now return to the first
central puzzle presented at the very onset of Chapter 1: What explains
differences in authoritarian regime stability? Why is it that some dictator-
ships are subject to constant coup attempts and regime change, while
others withstand multiple peaceful leadership transitions and remain
stable over long periods of time?
The next two chapters will examine the effects of institutionalization on
regime durability. Do institutionalized regimes actually perform better on
a number of key outcomes, such as leader longevity, the prevention of coups,
and the promotion of peaceful leadership succession? This chapter will focus
on estimating the effects of regime institutionalization on key leader out-
comes: the length of leader tenure and susceptibility to coups attempts. The
next chapter will focus on estimating the effects of institutionalization on key
regime outcomes: leadership succession and regime survival.
This chapter will show that leaders who institutionalize their regimes
tend to remain in office for longer periods of time and face fewer coup
attempts. The next chapter will demonstrate that institutionalized regimes
are significantly more likely to undergo peaceful leadership transitions.
Together these two chapters demonstrate that executive constraints are
not simply empty parchment rules. Institutionalization does indeed have
consequences for regime stability.
However, estimating the effects of endogenously created institutions is not
as straightforward as it may seem. In determining the effects of institutions,
these chapters will take into consideration the subject of the first half of this
book – the fact that leaders choose whether to institutionalize based on
underlying threats to their rule. Since constitutional rules and cabinet
appointments are not randomly assigned across regimes, we must consider
the types of leaders who choose to create these institutions when examining
whether such rules have an independent effect on regime durability.
186
6.1 Are Rules Followed in Dictatorships?
Existing work on the empirical determinants of regime stability has

largely ignored the empirical challenge of estimating the effect of endo-
genous institutions, even though much of the scholarship does indeed
acknowledge theoretically that dictators choose whether or not to allow
for institutional constraints. This chapter argues that to determine
whether institutions play a stabilizing role, we cannot simply regress
regime outcomes on levels of institutionalization.
Recall that the formal model in Chapter 2 demonstrated how leaders
who enter office already strong can remain in power without institu-
tionalizing the regime. Such leaders will therefore be associated with
stable regimes and no institutionalization. Weak leaders, on the other
hand, must institutionalize in order to maintain their hold on office.
Therefore, if we focus on the subset of leaders who enter power weak,
we should observe a positive effect of institutionalization on regime
stability. Yet, without taking into account leader type, simply regres-
sing regime outcomes on institutionalization would result in biased
estimates because strong leaders do not need to institutionalize to stay
in power.
In this chapter I demonstrate that when we condition on initial leader
strength, institutionalization does indeed have a stabilizing effect for weak
leaders but little or no effect for strong leaders. This approach helps to
make sense of the existing divergent empirical findings on the question of
whether institutions promote regime stability. It heeds Pepinsky’s (2014)
call to deal with the “nagging problem” facing the literature on author-
itarian institutions: If “institutions under authoritarian rule are vulnerable
to manipulation because political actors believe that institutional manip-
ulation can shape political outcomes in their favor, then it is also true that
factors that explain the origins of (and changes in) [these institutions] also
directly affect those political outcomes” (631).
6. 1 are constitutional rules actually

followed i n dictatorships?
Before we discuss the effects of institutionalization on regime outcomes, it
is important to first address an important descriptive question: once
established, are constitutional rules followed in dictatorships? If we are
to believe that formal rules promote regime stability, then we should
observe leaders and elites actually following the letter of the law. To
answer this question, I focus on the constitutional rules in my dataset:
leadership succession procedures and term limits. I find that constitutional
leadership succession procedures are generally adhered to in my sample,
while term limits are violated much more frequently.
187
To examine whether constitutional leadership succession rules were

followed, I identified all instances of peaceful leader exit, whether by
death or voluntarily retirement. For all instances of peaceful leader exit,
I then identify whether a constitutional succession rule existed at the
time of leader exit. The procedure results in thirty-four cases that allow
us to observe whether the succession policy was followed at the time of
leader exit. Cases of leader departure but no succession rule were
excluded because the question of whether succession rules would have
been followed is unobserved in such instances. Cases of violent leader
deposal were also excluded because coups result in new regimes rather
than the continuation of the existing regime via succession. I also exclude
cases of democratization after the death of the leader: in these instances
the next leader was elected via elections, rendering the succession policy
irrelevant.
For the majority of the observations, leadership succession occurs
according to constitutional rules. Out of thirty-four observations, 79 per-
cent consist of peaceful leadership transitions where the constitutional
policy was followed exactly – in particular, the designated successor
became the interim president following the death or departure of the
leader. In 21 percent of these cases, a peaceful leadership transition did
occur, but the interim president was not the designated successor who was
specified in the constitution. Summary statistics are listed in Table 6.1.
To determine whether term limits were followed, I identified all leaders
who had term limits in the constitution and remained in power long
enough to observe whether they would adhere to those limits. This
resulted in forty-five observable cases. Out of these forty-five cases, leaders
adhered to term limits 29 percent of the time. To summarize, while
constitutional leadership succession policies were indeed followed in
most observable cases, leaders seem to violate term limits much more
frequently. The next chapter will provide a theory of why term limits
seem to be easier to violate compared to constitutional succession policies.
Since succession rules are generally followed but term limits are fre-
quently ignored, I exclude term limits from my analysis of the effects of
constitutional rules on regime stability for the remainder of this chapter.
Table 6.1 Summary statistics
Policy Observable cases Rule followed As percent
Succession rules 34 27 79%

188
6.2 Do Institutions Promote Regime Stability?
6 . 2 e m p i r i cal d i s ag re em en t : do i ns t i t ut i ons
promote regime s tabili ty?
Despite the general consensus that institutionalized forms of dictatorship
tend to be more stable, the empirical record of the relationship between
institutions and regime stability has been surprisingly mixed. In a seminal
study, Geddes (1999a) finds that single-party regimes remain in power for
longer periods of time compared to personalist regimes. Military regimes,
which tend to have weak or no civilian institutions, are less durable
compared with personalist regimes. This study catalyzed the idea that
institutionalized forms of authoritarianism tend to be most durable.
Geddes’ empirical analysis was replicated by Brownlee (2007), who
found similar effects. Since then, many scholars have made the theoretical
argument that institutions, such as parties and legislatures, can prolong
the lifespan of authoritarian regimes (Boix and Svolik 2013; Gandhi 2008;
Geddes, Wright, and Frantz 2018; Gandhi and Przeworski 2007;
Magaloni 2008; Svolik 2009).
However, a number of subsequent studies have since challenged this
initial finding. Smith (2005), for instance, shows that the significant effect
of single-party regimes is driven primarily by two outliers: Mexico and the
Soviet Union. He replicates Geddes’ original analysis, producing similar
results, but then demonstrates that when the two longest-living regimes
are dropped from the analysis, single-party regimes are no longer signifi-
cantly more durable compared with personalist regimes.
A small number of studies have also pointed out that autocratic leaders
endogenously respond to threats with “appropriate” degrees of institutiona-
lization, therefore we should not expect to see a relationship between institu-
tions and stability. Gandhi (2008), for instance, finds that leaders with ruling
parties do not tend to survive in office longer compared with those without
parties, “suggesting that most authoritarian incumbents are able to perceive
with some accuracy the threats germinating within society and to respond
with the appropriate degree of institutionalization” (176–177). According to
this study, leaders institutionalize optimally, and therefore there should be no
observable effect between institutions and stability.
Boix and Svolik (2013) also present a study in which institutional choice
is endogenous. They argue that power-sharing institutions, such as parties
or legislatures, can be established only when backed by a credible threat of
elite rebellion. However, despite the fact that their theory endogenizes the
creation of institutions, two observable implications of their study are that
regimes with institutions should be longer lived and experience fewer
coups. Using data on the existence of parties and legislatures in autocra-
cies, they find empirical support for their arguments.
189
To sum up, the existing literature on the relationship between auto-

cratic institutions and regime stability has produced a confusing array of
theoretical predictions and empirical findings. Two questions remain
unresolved. First, should we expect, theoretically, to see an effect
of endogenously created institutions on autocratic regime stability?
Second, what does the data show us empirically?
6.3 p roblems with e xisting approaches

How do we make sense of these contrasting theories and divergent
empirical findings? I argue that there are two possible reasons why exist-
ing empirical studies of the relationship between autocratic institutions
and regime durability may be biased: (1) measurement problems and (2)
theoretical problems.
First, existing empirical findings may be subject to a number of mea-
surement problems due to the difficulties of obtaining fine-grained data
that accurately reflects the quality of authoritarian institutions. As
Chapter 4 demonstrated, the data on autocratic regime types (Geddes,
Wright, and Frantz 2014) has a number of inconsistencies when compared
with this book’s dataset on executive constraints. In particular, many
single-party regimes, which are supposed to be the most institutionalized
type of autocracy, appear to lack many dimensions of formal and informal
constraints on presidential power. Many regimes that are coded as mili-
tary and personalist, which are supposed to be less institutionalized, do
actually appear to have many executive constraints in place.
On the other hand, empirical studies (Boix and Svolik 2013; Gandhi
2008) that use the existence of institutions – such as the presence of
a single, multiple, or no political parties – may also be subject to measure-
ment problems, due to the fact that such data often does not capture
detailed information about the quality of the institutions. Therefore it is
unclear whether the lack of an effect of the presence of institutions on
regime durability is due to a true lack of an effect or whether the null
findings are due to measures that do not capture institutional quality.
Beyond measurement problems, I also argue that existing empirical
studies on the relationship between autocratic institutions and regime
durability are also subject to biased theoretical modeling. In particular,
while many existing studies acknowledge the fact that leaders choose to
create or allow institutions in order to promote regime stability, this
endogenous institutional choice is not taken into account when estimating
the effect of institutions on outcomes, such as regime stability or long-
evity. As Pepinsky (2014) notes, “factors that explain the origins of (and
changes in) [institutions] directly affect those political outcomes” (631).
190
6.3 Problems with Existing Approaches
I argue that we need to differentiate between initially strong and initially

weak leaders when considering the effects of institutionalization on regime
durability, since these different types of leaders make different institutional
decisions. Since leaders choose to institutionalize based on existing threats to
their rule, the creation of executive constraints should only help initially
weak leaders stay in power longer than they otherwise would have.
Initially strong leaders who have already consolidated power by the time
they take office do not need to rely on power-sharing institutions to remain in
office. Their rule is secure with or without regime institutionalization. We
therefore should generally not observe a relationship between party institu-
tionalization and regime durability for strong leaders.1
For leaders who are weak, however, building an institutionalized regime
should promote regime stability. Leaders who are highly vulnerable to being
deposed by their fellow elites require institutional mechanisms in order to
make credible commitments to share power and resources. Institutionalized
regimes should therefore lengthen the time horizons of initially vulnerable
leaders. Initially strong leaders, on the other hand, have no incentive to
institutionalize their regimes to begin with. Figure 6.1 summarizes my
argument.
Figure 6.1 illustrates why simply regressing institutionalization on
regime stability would produce biased estimates of the effect of institu-
tions on stability. For initially weak leaders, those who institutionalize
should experience more stable regimes. For initially strong leaders,
however, there should be little or no effect of institutionalization on
regime durability because strong leaders should be able to remain in
power regardless of whether they institutionalize. A regression model
Strong No
leader Stable regime
institutionalization
No Unstable
Institutionalization regime
Weak
leader
Institutionalization Stable regime
Figure 6.1 What is the effect of institutionalization on regime stability?
1
It is important to note that it is possible for strong leaders to institutionalize, even
though this would be out-of-equilibrium behavior according to the model. I would
expect institutionalized regimes, even if the leader was initially strong, to be more
stable due to the fact that the creation of constitutional rules and cabinet appoint-
ments creates mechanisms for credible power-sharing and patronage distribution.
191
that regresses institutionalization on regime stability without taking into

Hypothesis 1: Institutionalization has a positive effect on regime
stability, conditional on leader weakness.
6.4 e stimating t he effects o f e ndogenous

How should we estimate the effects of institutionalization on regime
stability without ignoring the origins of these institutions? One approach
is to incorporate an interaction term between regime institutionalization
and leader strength which will account for differential effects of institu-
tionalization based on whether the leader was initially strong or weak.
6.4.1 Data
The outcome variable of interest in this chapter is autocratic stability,
particularly with respect to leader outcomes.2 Leader stability will be
measured via two dependent variables: the length of a leader’s tenure
and the number of coup attempts a leader faces during her tenure.
“Years in power” is a variable that counts the number of years a leader
remained in power. “Coup attempts” records the percent of the number of
office years for which a leader faced coup attempts. This variable is
presented as a percent, rather than a count of the number of coup attempts
a leader faced in order to take into account the fact that various leaders
remained in office for different lengths of time.
There are two primary independent variables of interest: leader
strength and regime institutionalization. Leader strength will be proxied
using two measurement strategies from Chapter 5: founders versus suc-
cessors and coup leaders versus noncoup leaders. “Founder” is a dummy
variable that takes a value of 1 if the leader was the first leader of the
country after independence and 0 otherwise. As Chapter 5 argued, found-
ing leaders tended to enjoy special legitimacy and mass support, and this
was especially true for these leaders who rose to power in the immediate
aftermath of independence. Founding presidents can therefore be inter-
preted as stronger leaders compared with subsequent presidents. “Coup
2
The next chapter will examine the effect of institutionalization on leadership succes-
sion and regime survival.
192
6.4 Estimating Effects of Endogenous Institutions
leader” is a dummy variable that takes a value of 1 if the leader came to

power through a coup d’etat and 0 otherwise. As Chapter 5 argued,
leaders who successfully take power through a coup demonstrate that
they have control over the coercive apparatus that allowed them to launch
a successful coup in the first place. Since one of the most precarious
challenges to any civilian leader is the threat of military takeover, when
the leader is the head of the military this threat is diminished. Coup leaders
can therefore be interpreted as strong leaders, compared with leaders who
did not come to power via a coup.3
The second primary independent variable of interest is level of regime
institutionalization. I use my indicators of regime institutionalization,
measured separately. For this chapter, I use the percent of years for
which the leader had a succession policy in place (“successor”), stable
VP/PM appointment (“VP/PM same”), and stable defense appointment
(“Defense same”).4 Each of these measures take a value from 0 to 1, with
high values indicating a higher level of regime institutionalization.
6.4.2 Summary Statistics

Before turning to the empirical strategy, I first present summary statistics of
the leader outcome variables: length of leader tenure and frequency of coup
attempts. The average leader in my sample was in power for 8.48 years, with
the shortest tenure being one year and the longest tenure being forty-one
years. Appendix Table 6.1 presents summary statistics of this variable. On
average, founding leaders remain in power for 13 years, while subsequent
leaders remain in power for 7.4 years. The average coup leader remains in
office for 8.2 years and the average noncoup leader remains in power for 8.6
years. This finding runs contrary to existing scholarship portraying coup
leaders as especially weak or susceptible to falling into coup traps, yet is
3
The reader may recall that in Chapter 5 I presented a third strategy for measuring
leader strength by looking at the subset of founding leaders, and comparing strong
nationalist leaders against those who were close allies of the colonial state. Since this
strategy requires looking within the subset of independence regimes, the number of
observations available for this strategy are much more limited. I therefore exclude
this strategy from the empirical tests in this chapter.
4
Since the strict and weak version of the successor rule are highly correlated, I exclude
the “strict successor” variable from the analysis in this chapter. Since “VP/PM
appoint” and “VP/PM same” as well as “defense appoint” and “defense same” are
also highly correlated, I also exclude the “VP/PM appoint” and “defense appoint”
variables from the analysis in this chapter. Since term limits were generally not
implemented until after the 1990s and are not frequently adhered to, I exclude it
from the analysis in this chapter.
193
consistent with my argument that leaders endogenously institutionalize

based on threat perceptions.
Turning to the other dependent variable, the average leader faces
a (successful or unsuccessful) coup attempt for 19 percent of the years
she is in power. Appendix Table 6.2 presents summary statistics of this
variable. In other words, if a leader were in power for ten years, she would
face a coup attempt in roughly two of the ten years.
There are significant differences in the percentage of years for which
strong leaders face coups, relative to weak leaders. Founders face coup
attempts for about 5 percent of the years they are in power, compared with
subsequent leaders who face coup attempts for about 22 percent of the
years they are in power. Coup leaders face coup attempts about 35 percent
of the of the years they are in power, compared with noncoup leaders who
face coup attempts for about 11 percent of the years they are in power.
It is important to remind the reader that the analysis conducted in this
chapter does not focus on these baseline differences in the length of leader
tenure or vulnerability to coups. Instead, what we are interested in is
whether institutionalization affects these outcomes within the subgroup
of weak leaders.
6.4.3 Empirical Strategy

Let Y represent the outcome of interest – leader tenure and percent of
years with coup attempts. Let x1 represent a dummy variable that takes
a value of 1 if the leader was initially strong, and 0 otherwise. Let x2
represent the degree of institutionalization, which is measured as
the percent of years for which a particular dimension of institutionaliza-
tion was in place during a leader’s tenure. x2 takes a value between 0 and
1, and higher values indicate more institutionalized regimes. Let
W represent a vector of control variables, including mean GDP per capita,
oil production, population size, and ethnic fractionalization. ε represents
the error term. Since the outcome variables and leader strength variable of
interest are cross-sectional, our unit of analysis will be the leader.
Y ¼ α þ β 1 x1 þ β 2 x2 þ β 3 x1 x2 þ W þ ε
Recall that the hypothesis we’re interested in testing is that institutionali-
zation has a positive effect on regime stability, conditional on leader
weakness. This means that we should observe better regime outcomes
for weak leaders but no relationship between institutionalization and
regime outcomes for strong leaders.
How should we interpret the regression coefficients? We break down
our analysis into two sets of cases: those with a strong leader and those
194
with a weak leader. For x1 =0, the subset of weak leaders (recall that x1 is
a dummy variable representing leader strength or weakness), the equation
reduces to the following:
Y ¼ α þ β2 x2 þ ε
I hypothesize that when the leader is weak, institutionalization should
have a positive and significant effect on leader outcomes. Since x2 denotes
levels of regime institutionalization, we should expect β2 to be signifi-
cantly different than zero.5
For x1 =1, the subset of strong leaders, the equation reduces to the
following:
Y ¼ α þ β1 þ ðβ2 þ β3 Þx2 þ ε
I hypothesize that when the leader is strong, there should be no effect of
institutionalization on leader stability. Since x2 denotes levels of regime
institutionalization, then the coefficients ðβ2 þ β3 Þ should not be signifi-
cantly different from 0.
To summarize, if my theory is correct, then we should expect β2 to be
significantly different from 0 and ðβ2 þ β3 Þ not to be significantly different
from 0.
6 . 5 e m p i r i c a l ana l y s i s a n d fin d i n g s
I use two different proxies of leader strength to condition the differential
effects of institutionalization on length of leader tenure and vulnerability to
coups: (1) founding leaders versus subsequent leaders and (2) coup leaders
versus noncoup leaders. I will present results using both of these strategies
separately.
6.5.1 What Is the Effect of Institutionalization on the Length

of Leader Tenure?
As hypothesized in Section 6.3, we should expect institutionalization to
lengthen the time in office of initially weak leaders but have no effect on
the tenure of initially strong leaders. Therefore we should expect β2 > 0
and ðβ2 þ β3 Þ to not be significantly different from 0. To estimate these
5
Whether β2 should be greater than or less than 0 depends on the outcome we are
examining. A lower number of coup attempts denotes a more stable regime, so for
that analysis, we should expect β2 < 0: Longer leader tenure denotes a more stable
regime, so for that analysis, we should expect β2 > 0.
195
Figure 6.2 What is the effect of institutionalization on leader tenure (founders)?

subsequent leaders. The dependent variable is the number of years for which a leader was in
power. Our hypothesis predicts a positive relationship between institutionalization and
leader tenure for weak leaders.
effects, we include an indicator of leader strength and an interaction term

between leader strength and institutionalization.
First I use founding leaders as a proxy for initially strong leaders. The
results of these regressions are illustrated in Figure 6.2, and the full
regression results are reported in Appendix Table 6.3. When weak leaders
(nonfounders) institutionalize, they tend to remain in office for longer
periods, and this result is statistically significant for all three dimensions of
institutionalization: succession procedures, having a stable VP/PM, and
having a stable defense minister.6
The results also show little or no effect of institutionalization on leader
tenure for strong leaders (founders).7 Designating a stable VP/PM and
defense minister does not appear to have a statistically significant effect on
leader tenure. Having a constitutional succession policy does have
a statistically significant effect of lengthening leaders’ time in office, even
when the leader is initially strong. This could be due to some measurement
6
Recall that (ß2) is the relevant coefficient for weak leaders. These rows are
highlighted in Appendix Table 6.3.
7
Recall that the effect of institutionalization for strong leaders is denoted by ß2 + ß3.
This row is highlighted in Appendix Table 6.3.
196
Figure 6.3 What is the effect of institutionalization on leader tenure (coup leaders)?
versus noncoup leaders. The dependent variable is the number of years for which a leader
was in power. Our hypothesis predicts a positive relationship between institutionalization
and leader tenure for weak leaders.
error for the leader strength variable. It could be the case that founding
leaders does not serve as a perfect proxy for initial leader strength. On other
hand, it is also possible that having a leadership succession policy in place
does indeed lengthen the tenure of all leaders, even those who enter office
initially strong. Chapter 7 provides an extensive discussion of leadership
succession.
I also analyze the relationship between institutionalization and leader
tenure using coup leaders as a proxy for initially strong leaders. The
results are illustrated in Figure 6.3 and the full regression results are
reported in Appendix Table 6.4. Once again, I find that when initially
weak leaders (noncoup leaders) institutionalize, they tend to remain in
office for longer periods of time, and this finding is statistically significant
along cabinet stability dimensions.
For initially strong leaders (coup leaders), institutionalization – particu-
larly the stable appointment of VP/PMs and the creation of succession
policies – does appear to have effects on leader tenure. It is possible that
coup leaders, while having consolidated authority over the coercive appara-
tus of the regime, have vulnerability along other nonmilitary dimensions, and
therefore benefit from the stabilizing effect of institutionalization along such
dimensions. Coup leaders who institutionalize along civilian dimensions may
197
experience more stable rule, even if they took power having already con-
solidated power along coercive dimensions.
6.5.2 What Is the Effect of Institutionalization on the Frequency

of Coup Attempts?
Does institutionalization prevent coup attempts? First I use founding
presidents as a proxy for strong leaders. The results from these regressions
are illustrated in Figure 6.4, and the full regression results are reported in
Appendix Table 6.5. Initially weak leaders (non-founding presidents) who
institutionalize are less likely to face coup attempts, and the results are
statistically significant across all dimensions of institutionalization. There
does not appear to be an effect of institutionalization on coup threats for
initially strong leaders (founding presidents) and this is true for all dimen-
sions of institutionalization.
Finally, I also analyze the effect of institutionalization on coup attempts
using postcoup leaders as a proxy for strong leaders. The results are illu-
strated in Figure 6.5, and the full regression results are reported in Appendix
Table 6.6. Initially weak leaders (noncoup leaders) who institutionalize are
significantly less likely to face coup attempts, and this is true for all
Figure 6.4 What is the effect of institutionalization on the frequency of coup attempts?
subsequence leaders. The dependent variable is the percentage of years for which a leader
198
6.6 Conclusion
Figure 6.5 What is the effect of institutionalization on coup vulnerability?

versus noncoup leaders. The dependent variable is the percentage of years for which a leader
dimensions of institutionalization. There does not appear to be a significant

effect of regime institutionalization on coup threats for initially strong leaders
(coup leaders). The one exception to this general finding is that initially
strong leaders who have constitutional leadership succession procedures
appear to be significantly less likely to face coup attempts. This may be due
to the fact that, as discussed previously, while coup leaders enter power
having already consolidated coercive authority, they may be more vulnerable
along other civilian dimensions. Therefore implementing succession policies
may lessen coup risk, even for leaders who took power via a coup.
6.6 conclusion
This chapter examined the effect of institutionalization on autocratic sta-
bility, particularly the length of leader tenure and frequency of coup threats.
I argued that in estimating the effect of institutionalization on these out-
comes we need to take into account the fact that initially strong and initially
weak leaders make different institutionalization decisions, depending on
threats to their rule. Weak leaders institutionalize in order to alleviate
threats to their rule, and therefore we should expect to see an effect of
institutionalization on stability for initially weak leaders. Initially strong
leaders are able to remain in power whether they institutionalize or not, so
199
we should expect to see little or no effect of institutionalization on stability

for strong leaders. Using founding presidents and postcoup leaders as my
proxies for leader strength, I demonstrate that these patterns are indeed
born out in the data. The next chapter will consider the effect of institutio-
nalization on leadership succession.
Appendix
Appendix Table 6.1 Summary statistics: length of leader tenure
Whole sample 8.483 8.224 207

Founder (strong leader) 13.000 1.320 40
Nonfounder (weak leader) 7.401 0.606 167
Coup leader (strong leader) 8.214 1.061 70
Noncoup leader (weak leader) 8.620 .674 137
Note: Dependent variable is a count variable of the number of years the leader was in power.
Appendix Table 6.2 Summary statistics: frequency of coups
Whole sample .190 .301 206

Founder (strong leader) .051 .013 40
Nonfounder (weak leader) .223 .025 166
Coup leader (strong leader) .349 .043 69
Noncoup leader (weak leader) .109 .019 137
Note: Dependent variable is the percentage of years in which a leader faced a coup attempt
while in power.
200
6.6 Conclusion

(leader strength proxy: founder versus successor)?
DV: number of years leader is in power (1) (2) (3)
Founder 7.388*** 4.781** 4.886**

(2.323) (2.013) (2.263)
VP/PM same (ß2) 4.999***
(1.571)
Founder*VP/PM same (ß3) −4.000
(4.252)
Successor (ß2) 2.770*
(1.474)
(3.265)
(1.945)
(4.345)
GDP per capita −0.227 −0.242 −0.008
(0.358) (0.406) (0.348)
Oil production 1.658*** 1.870*** 1.493***
(0.508) (0.567) (0.468)
Population 0.287 0.500 0.086
(0.651) (0.744) (0.698)
Ethnic fractionalization −0.343 −1.181 −0.491
(1.751) (1.931) (1.904)
Constant 3.027 2.207 5.093
(5.570) (6.494) (5.839)
ß 2 + ß3 0.999 5.295* 6.251
(3.860) (2.924) (4.171)
R-squared 0.153 0.155 0.155

201

(leader strength proxy: coup leaders versus noncoup leaders)?
DV: years in power (1) (2) (3)
Coup leader −0.169 −0.189 0.875

(1.830) (1.413) (1.646)
VP/PM same (ß2) 3.808*
(2.227)
Founder*VP/PM same (ß3) 3.402
(4.470)
Successor (ß2) 2.700
(1.884)
(2.883)
(2.650)
Founder*defense same (ß3) −0.714
(4.445)
GDP per capita −0.119 −0.120 0.087
(0.356) (0.412) (0.346)
Oil production 1.452*** 1.648*** 1.259**
(0.535) (0.595) (0.503)
Population −0.038 0.251 −0.308
(0.686) (0.762) (0.716)
(2.369) (2.468) (2.413)
Constant 6.504 4.311 8.213
(5.735) (6.504) (5.888)
ß 2 + ß3 7.21* 6.426** 5.483
(3.924) (2.565) (3.623)
R-squared 0.078 0.082 0.092

202
6.6 Conclusion

vulnerability (leader strength proxy: founder versus successors)?
Founder −0.222*** −0.227*** −0.201***

(0.045) (0.058) (0.058)
VP/PM same (ß2) −0.235***
(0.065)
Founder*VP/PM same (ß3) 0.263***
(0.065)
Successor (ß2) −0.155***
(0.051)
Founder*successor (ß3) 0.200***
(0.067)
(0.098)
Founder*defense same (ß3) 0.261**
(0.121)
GDP per capita −0.017* −0.019** −0.029***
(0.009) (0.009) (0.006)
Oil production 0.006 −0.002 0.017
(0.011) (0.014) (0.012)
Population −0.019 −0.038** −0.016
(0.018) (0.018) (0.015)
Ethnic fractionalization 0.103 0.144** 0.107**
(0.068) (0.061) (0.051)
Years in power −0.011*** −0.011*** −0.011***
(0.002) (0.002) (0.002)
Constant 0.511*** 0.641*** 0.477***
(0.148) (0.146) (0.123)
ß2 + ß3 0.028 0.045 0.029
(.048) (.052) (.070)
R-squared 0.243 0.234 0.239

203

vulnerability (leader strength proxy: coup leaders versus noncoup leaders)?
Coup leader 0.039 0.100* 0.053

(0.038) (0.059) (0.059)
VP/PM same (ß2) −0.096**
(0.039)
Founder*VP/PM same (ß3) 0.158*
(0.082)
Successor (ß2) −0.071
(0.049)
Founder*successor (ß3) −0.079
(0.073)
(0.050)
(0.085)
GDP per capita −0.002 −0.004 −0.008
(0.008) (0.008) (0.008)
Oil production 0.011 0.004 0.013
(0.011) (0.011) (0.010)
Population −0.010 −0.023* −0.014
(0.013) (0.013) (0.013)
(0.062) (0.055) (0.059)
Years in power −0.008*** −0.007*** −0.008***
(0.002) (0.001) (0.002)
Constant 0.266** 0.361** 0.297**
(0.128) (0.134) (0.124)
ß 2 + ß3 .062 − .149*** .025
(.072) (.053) (.071)
R-squared 0.265 0.303 0.276

204
7
In this chapter, we now turn to examine the effects of institutionalization

on key regime outcomes. In particular, we focus on whether highly insti-
tutionalized regimes are more likely to undergo peaceful leadership tran-
sitions. One of the fundamental challenges of continued authoritarian rule
is the transfer of power from one leader to another. Yet dictatorships
inherently lack mechanisms of electoral competition and leadership suc-
cession. Incumbents often remain in office for long periods of time, mak-
ing it difficult to establish norms of leader turnover. Elections – even when
they exist – are largely meant to sustain, rather than disrupt, the incum-
bent’s regime. As a result, the process of peacefully transferring power
from one leader to another is often quite precarious in dictatorships.
How can autocratic regimes overcome this fundamental challenge of
leadership succession? In this chapter I argue that constitutional succes-
sion rules and the appointment of a de facto successor play a critical role in
regulating the transfer of power. The institutionalization of a clear line of
succession provides certainty in the chain of command. Designated suc-
cessors have an incentive to uphold the existing regime, since preserving
the status quo is the best guarantee that they will become the next incum-
bent. Leaders who plan for succession are therefore less likely to be
deposed in a coup. Moreover, regimes with succession procedures are
significantly more likely to undergo a peaceful transition of power because
succession planning transforms conflict over succession from a single-shot
game after the leader’s death into a dynamic process in which elites
compete for power peacefully within the political arena.
However, I argue that only certain types of rules are effective in pro-
moting peaceful leadership succession. Constitutional amendments that
transfer of power. Such policies establish certainty and predictability
about the actors who are expected to come into office, providing these
elites with a stake in maintaining the existing regime in order to reap
205
7 Consequences on Leadership Succession
accountable.
Importantly, the argument made in this chapter departs from the
argument made in Chapter 6 in the following way. In this chapter,
I argue that when estimating the effect of institutionalization on leader-
ship succession outcomes, we do not need to condition on the initial
strength of the leader. This is because peaceful succession outcomes are
an unintended consequence of regime institutionalization. As Chapter 6
demonstrated, self-interested leaders institutionalize their regimes in
order to insulate themselves against coup attempts. At the same time,
these measures – namely the creation of constitutional succession rules
and the appointment of a de facto successor – create certainty surrounding
the line of succession, therefore paving the way for a peaceful leadership
transition. While leaders do not necessarily care about the outcome of the
regime after their death, the measures they take to stabilize their rule while
in power nonetheless create conditions for the peaceful transfer of power.
This chapter makes two important contributions to theories of leader-
ship succession in dictatorships. First, it is one of the few studies to test
mechanisms of peaceful leadership change in autocracies. Though the
politics of succession is considered to be one of the central challenges of
autocratic rule, the mechanisms that facilitate peaceful leadership transi-
tions are not well understood for modern dictatorships. A limited number
of studies on this topic have focused on the role of hereditary succession
and ruling parties.
Early studies of leadership succession by Tullock (1987) focused on
importance of hereditary succession and primogeniture, and his the-
ories have since been extended and tested on pretwentieth century
European monarchies (Kurrild-Klitgaard 2000; Kokkonen and
Sundell 2014). However, the vast majority of authoritarian regimes
today are not monarchies. In fact, only 7 percent of all autocratic
regimes from 1946 to 2010 are coded as monarchies, and out of those
monarchies only two remain in power today (Geddes, Wright, and
Franz 2014).1 Furthermore, I show in this chapter that hereditary
succession is incredibly rare in modern dictatorships. Within my
1
Out of 280 autocratic regimes from 1946 to 2010, nineteen are coded as monarchies.
206
7.1 Do Dictators Care About Leadership Succession?
dataset only 5 percent of leadership transitions (whether peaceful or

not) were between family members.
A second group of studies suggests that ruling parties play an important
role in facilitating leadership succession. Scholars posit that an essential
role of ruling parties is that they help to mitigate and manage elite conflict,
providing an arena for politicians to compete for power and resources
(Brownlee 2007; Geddes 1999b; Magaloni 2008). Parties can also provide
a mechanism for elite power sharing (Boix and Svolik 2013; Svolik 2012).
Through these mechanisms, ruling parties can help facilitate peaceful
leadership transitions. However, in this chapter, I provide empirical evi-
dence that the presence of a ruling party is never a strong predictor of
peaceful leadership succession.
A second important contribution of this chapter is that it emphasizes
the fact that peaceful transitions of power have indeed occurred within
sub-Saharan African, even during the most authoritarian decades. Past
scholarship on leadership succession in sub-Saharan Africa has generally
held a pessimistic view about the prospect of peaceful or constitutional
transitions. Sylla and Goldhammer (1982), for instance, refer to the
challenge of succession as the “Gordian Knot of African Politics.”
Jackson and Rosberg (1982) predicted that “power politics rather
than institutionalized conventions and procedures may have to resolve
the succession issue” (69). By contrast, this chapter demonstrates that
not all leadership transitions that take place within autocracies in sub-
Saharan Africa are unregulated or driven solely by violence. Instead,
I show that institutional mechanisms can play an important role in
promoting peaceful successions. Regimes that are institutionalized are
significantly more likely to experience continuity across leadership
transitions.
7.1 do dictators care about leadership

s u c c e s si o n ?
As mentioned in the introduction of this chapter, dictators institutionalize
their regimes not necessarily because they care about succession politics
but in order to stabilize their own rule. After all, readers may have been
wondering why a dictator would care about leadership succession if the
transfer of power occurs after the leader has already died or left power.
However, it turns out that the problem of how to extend executive
authority to another elite is a concern not only to others in the regime,
but also to the dictator herself. As Herz (1957) so aptly summarizes: “It
also colors, in anticipation, so to speak, the entire situation during the
dictator’s life and rule. To the dictator it poses a problem and constitutes
207
a danger. To his aids it is a temptation. To the bystander, within and

without, it is fascination” (20).
Scholars originally believed that dictators were quite wary of naming
formal successors due to the “crown prince problem.” Herz famously
argued that by grooming a successor, incumbents create their own worst
enemy. The mere designation of an alternative center of power makes
leaders vulnerable to being deposed prematurely by their own chosen
successor. According to Herz, by naming a successor, the ruler voluntarily
shifts the center of power away from herself, putting herself at risk of
being deposed by her own appointee.
Yet not planning for succession is also equally dangerous for the regime
as well as the dictator herself. When autocratic leaders die without desig-
nating a successor a coordination problem emerges among regime elites
because it is not clear who the de facto successor should be (Kokkonen and
Sundell 2014). Such power vacuums in the aftermath of the death of the
leader often invite coups (Frantz and Stein 2017) or even civil wars
(Kokkonen and Sundell 2020) as elites vie for the incumbency.
In addition, leaders may actually endanger themselves by not planning
for succession. In the absence of an institutionalized succession order,
elites may be incentivized to preemptively try to take power through
coercion in anticipation of eventual conflict over succession. Indeed,
many dictators seem to fall prey to this dilemma, as the majority of
autocratic leaders from 1946 to 2008 have been deposed via nonconstitu-
tional means (Svolik 2012). In fact, leaders themselves often support the
institutionalization of succession policies because it buffers them from
coup attempts (Frantz and Stein 2017). In sum, succession planning
promotes stability during the incumbent’s tenure as well as for the regime
as a whole during the critical moments of leadership transition.
This argument builds on recent literature highlighting the importance
of constitutions in supporting autocratic rule (Albertus and Menaldo
2012; Ginsburg and Simpser 2014). Constitutional rules matter greatly
because they are publicly observable, therefore elites can condition their
behavior and credibly threaten to punish autocratic leaders who violate
existing rules. Autocratic constitutions are most helpful when they can
serve as “focal points” for elites by reducing ambiguity around a clear set
of rules and enforcement behavior (Albertus and Menaldo 2012; Hadfield
and Weingast 2013).2 In fact, Ginsburg and Simpser (2014) argue that
2
This argument is also related to Fearon’s (2011) concept of “self-enforcing democ-
racy” whereby elections provide citizens with a commonly understood set of rules
and procedures that allow them to credibly protest if the leader violates election
principles.
208
7.2 How Succession Planning Promotes Regime Stability
constitutional documents may even be especially helpful in autocracies

because the creation of clear and observable rules eases coordination
problems and creates a self-enforcing system – one that can operate in
the absence of a strong judicial system because deviations may be pun-
ished using force. In this way, formal institutions can also serve to “tie the
leader’s hands” and help make autocratic commitments credible, precisely
because creating constitutional rules makes leaders vulnerable to being
deposed if they violate the terms (North and Weingast 1989; Myerson
2008).
In this sense, the “crown prince problem” that Herz highlights actually
helps keep leaders in power. Because the process of designating
a successor and formalizing the line of succession shifts the distribution
of power in favor of elites, they are able to credibly sanction the leader in
the event that the leader reneges on distributive promises. As the model in
Chapter 2 emphasized, autocratic leaders can enhance the stability of their
own tenure by empowering other elites. This dynamic seems to play out
empirically as well. As the results from Chapter 6 demonstrate, leaders
who create constitutional succession rules and appoint a vice president or
prime minister do indeed face fewer coup attempts.
7.2 how does s uccessio n planning promote

regime stability?
Beyond preventing elites from preemptively launching coups in hopes of
capturing power, how does institutionalized succession planning promote
regime stability? I argue that regimes with constitutional amendments
specifying the process of leadership turnover are more likely to undergo
peaceful transitions compared with those without such formal rules.
In particular, three main mechanisms drive the relationship between
formal succession planning and peaceful transfers of power. First, consti-
tutional succession rules provide certainty and clarity surrounding the
chain of command. As a result, designated successors have an incentive
to protect the incumbent and existing regime, since preserving the status
quo is the best guarantee that they will become the next incumbent. Once
named, successors can also start to build their own alliances in anticipa-
tion of the leadership transition. Therefore, a clear line of succession
protects the regime from other potential elite challengers.
The transition in Kenya from President Jomo Kenyatta to Vice
President Daniel arap Moi illustrates this mechanism well. Kenya’s
first presidential transition occurred peacefully in 1978 upon the
death of Jomo Kenyatta, the country’s first president after indepen-
dence, although the struggle over the succession process occurred in
209
the final years of Kenyatta’s presidency. Kenyatta, a highly influential

figure in the country’s struggle for self-determination, took office upon
independence in 1964. In 1967, he named Daniel arap Moi as his vice
president, a position that Moi served for eleven years through the
transition. Notably, Moi was not a member of Kenyatta’s Kikuyu
ethnic group – a power-sharing mechanism by design. Cabinet positions
were often utilized as implicit contracts to maintain support across
various ethnic groups and factions, and Kenyatta always appointed
non-Kikuyus as vice president (Arriola 2009; Karimi and Ochieng
1980; Widner 1992).
The Kenyan constitution, which has been in effect since independence
in 1963, includes specific instructions for succession. Chapter II, Part I,
Section 6 reads:

resignation of the President . . . an election of a President shall be held
Members of the Kikuyu faction within the government did not want
Moi to ascend to the presidency because they feared that they would lose
valuable resources and influential positions that they enjoyed under the
Kenyatta presidency. Furthermore, they feared that the constitutional
succession rules solidified Moi’s role as successor to the presidency. In
the case of Kenyatta’s death, Moi would quickly use his role as interim
president to consolidate power within ninety days to confirm his new role
(Karimi and Ochieng 1980, 15).
In September of 1976, the Kikuyu faction, now calling themselves the
“Change the Constitution” group, proposed that the constitutional suc-
cession rule be changed so that the speaker of the National Assembly (a
position within the faction’s control) would be the interim president,
rather than the vice president. They called a meeting to discuss these
constitutional changes, and twenty MPs attended. Moi and his supporters
immediately responded by obtaining signatures from ninety-eight MPs
(including ten cabinet ministers) who opposed the proposed amendment
change. Importantly, this document showed that the Change the
Constitution group did not have a parliamentary majority needed to
210
pass a constitutional change. (Karimi and Ochieng 1980, 19–25;

Tamarkin 1979, 24; Widner 1992, 115–116).
The attorney general, who was a Moi supporter, further issued
a warning: “It is a criminal offence for any person to encompass, imagine,
devise, or intend the death or deposition of the President.” As a final nail in
the coffin, Kenyatta himself made a statement supporting the attorney
general’s earlier warning, despite the fact that many members of the
Change the Constitution group were from Kenyatta’s own family
(Tamarkin 1979, 24; Widner 1992, 166–117).3
Of course, the constitutional amendment was likely not the only
factor that led to the ascension of Moi to the presidency. Kenyatta’s
influence played a role in supporting Moi’s claims to the office, and the
Change the Constitution group lacked the support of a majority in the
parliament. However, the presence of formal succession rules acted as
a critical barrier against attempts by other elites to dethrone the
appointed successor. As Tamarkin (1979) effectively summarizes: “The
anti-Moi group suffered a severe setback in its first attempt to assert
itself. If they thought of pursuing their struggle they would have to do it
against the backdrop of a grim political reality and with a debilitating
constitutional constraint” (24).
A second benefit of formal succession rules is that such procedures
transform conflict over succession from a single-shot game after the
leader’s death into a dynamic process in which elites can compete for
power peacefully within the political arena. In fact, upon Kenyatta’s
death, as it became clear that Moi would indeed serve as the next pre-
sident, former members of the Change the Constitution group all publicly
proclaimed their loyalty to Moi. Mbiyu Koinange, a long-time cabinet
minister who had served Kenyatta since 1964, had been a participant of
the Change the Constitution group. Upon Moi’s quick ascension to the
presidency after Kenyatta’s death, Koinange made a public loyalty pledge
to Moi, stating “there is no truth whatsoever in the rumours spreading
abroad that I or any other respectable politician I know of in this stable
land of ours will be opposing the President” (as quoted in Karimi and
Ochieng 1980, 3). Koinange actually remained a cabinet minister the
first year of the Moi presidency (although he was given a less prestigious
portfolio) before being removed from the cabinet the following year.
A third benefit of constitutional succession rules is that they prevent
incumbents from reneging on promises made to their designated
3
Two of the leaders of the Change the Constitution movement were Dr. Njoroge
Mungai, Kenyatta’s nephew and personal doctor, and Mbiyu Koinange, his brother-
in-law.
211
successors. The 2012 transition from President Bingu wa Mutharika to his

vice president Joyce Banda in Malawi illustrates this mechanism. Like
Kenya, the Mawali constitution includes an amendment that states:
“Whenever there is a vacancy in the office of President, the First Vice-
President shall assume that office for the reminder of the term” (Section
83(3)). Though Mutharika initially promised to support Banda as his
successor, he reneged on this pledge and instead began to endorse his
younger brother, Peter Mutharika, as the next incumbent.4 Upon
Mutharika’s death in 2012, factions supporting Peter Mutharika tried to
contest Banda’s ascension to the presidency. These attempts ultimately
failed, as the courts and military supported the existing constitutional
succession plans (Cammack 2012; Dionne and Dulani 2013).
In sum, the creation of formal constitutional succession rules promotes
peaceful leadership transitions by protecting the existing regime against
outside challenges, in addition to keeping the incumbent accountable to
her successor. Although scholars of African politics have noted that such
constitutional procedures now play an important role in regulating the
transfer of power in democratizing countries in the post-Cold War era
(Dionne and Dulani 2013; Posner and Young 2007), I show that these
mechanisms played an important role in the transfer of power during the
autocratic era as well.
Hypothesis 1: Regimes that have a constitutional succession rule
are more likely to have peaceful transfers of power.
7.3 designating informal successors

Beyond formal constitutional rules, I also argue that the stable appoint-
ment of a vice president or prime minister can also promote peaceful
transitions by signaling who the designated successor should be. Within
the context of sub-Saharan Africa, vice presidents and prime ministers
have long been understood to be the heir apparent of the president. Vice
presidents and prime ministers are commonly designated as the interim
president in the event of the death or incapacitation of the president. In
many cases, the interim president is even allowed to serve out the rest of
the former incumbent’s term. As Hughes and May (1988) note, this
provides an interim president with “an enormous advantage in
4
Mutharika could not simply remove Banda as vice president because the Malawi
constitution mandates that the vice president can only be removed by impeachment
(Dionne and Dulani 2013).
212
7.4 Succession Policies Versus Term Limits
consolidating his power. His constitutional authority places immediate

patronage in his hands and this is used to reward allies and eliminate
rivals” (14).
Even when the constitution does not specify the succession order, the
“second-in-command” position has generally been perceived as the most
likely stepping-stone to the presidency. As Table 4.1 from Chapter 4
shows, presidents are frequently former vice presidents or prime ministers.
Leon M’Ba, the first president of Gabon, for instance, appointed
a loyal second-in-command, Omar Bongo, to the vice presidency as his
heir apparent. “It is probably true that Bongo’s own political skills served
him well during the transitional period, but this does not gainsay the fact
that he was M’Ba’s personally selected [heir] and that he benefitted from
the approval and legitimacy that such selection bestowed” (Jackson and
Rosberg 1982, 72).
The implication of this argument is that succession planning does not
necessarily have to take place only on paper. Appointing an heir apparent,
similar to the creation of constitutional succession rules, empowers
a specific elite and incentivizes the successor to preserve the existing
regime. Therefore, if my argument about individual elites as focal points
is correct then we should expect that the appointment of de factor succes-
sors is as useful as constitutional succession rules in promoting peaceful
transitions.
Hypothesis 2: Regimes that have an appointed de facto successor
7.4 disaggregat ing autocratic constitutions:

s u c c es s i o n p o l i ci e s v e r s u s t e r m l i m i ts
Although autocratic constitutions may be an important source of regime
stability, I argue that not all constitutional amendments are equally effec-
tive in constraining leaders and solving the problem of leadership succes-
sion. In fact, there is important variation within constitutions: certain
types of amendments matter more and other types matter less. This argu-
ment builds on existing studies of comparative constitutions which high-
light variation in which different kinds of constitutional rights are
enforced (Chilton and Versteeg 2015). Chilton and Versteeg, for instance,
argue that not all rights are equally effective and in an analysis of con-
stitutions from 186 countries, they find that the right to unionize is more
often upheld than freedom of expression.
While I hypothesize that the implementation of constitutional rules
specifying procedures for succession play an important role in promoting
213
peaceful transitions, I predict that the presence of term limits do not.

Existing scholarship on leadership succession procedures in modern auto-
cracies suggest that term limits can be an important type of rule that
regulates the transfer of executive power by specifying when the incum-
bent must leave office. Term limits specify a finite length of time in which
the leader will hold office and may incentivize de facto successors to wait
for the leader to voluntarily leave office, rather than trying to stage a coup
prematurely.
However, term limits do not empower other elites in the same way that
succession rules do because term limits do not specify who the successor
will be. Returning to the example of Kenya, the constitutional amendment
that was implemented under Kenyatta specified that the person in the vice-
presidential role is to be the de facto successor in the event of the death of
the leader. Under this particular constitutional rule, Moi and his suppor-
ters had a lot to lose in the event that this particular amendment was
violated. Succession policies create certainty and establish predictability
about the actors who are expected to come into office and benefit from the
existing regime. Not only do such policies empower specific elites, they
also incentivize these elites to protect the existing regime in order to reap
the benefits of the existing succession order.
Term limits, on the other hand, do not identify a clear alternative leader
that other elites can coordinate around. When term limits are violated by
incumbents – which often occurs in autocracies – it is not clear who exactly is
being hurt the most, unless a specific elite has been designated as a de facto
successor. In sum, constitutional term limits do not help solve coordination
problems as effectively as succession procedures because term limits do not
identify a specific individual that elites should coordinate around.
In sum, we should not expect to see a strong effect of term limits on the
promotion of peaceful leadership transitions:
Hypothesis 3: Regimes that have term limits are not more likely to
have peaceful transfers of power than those who do not.
7.5 e ndogenous i nstitutionalization

re c o n s i d e re d
As Chapter 6 highlighted, a concern that arises when trying to identify the
effects of institutions is that leaders who enter power especially weak
choose to institutionalize; these institutions are certainly not randomly
assigned to regimes. If succession planning is an endogenous decision and
only weak leaders tend to implement succession policies, should we expect
to see effects of these institutions on transition outcomes?
214
7.5 Endogenous Institutionalization Reconsidered
I argue that succession policies should have observable effects

because they affect not only the leader but also the designated suc-
cessor. When weak leaders choose to plan for succession in order to
prevent other elites from preemptively staging a coup, their desig-
nated successors also benefit inadvertently from this strategic plan-
ning. As Sections 7.2 and 7.3 argued, constitutional succession rules
and the appointment of a de facto successor provide certainty and
clarity surrounding the chain of command. Designated successors
have an incentive to protect the incumbent regime in hopes of
becoming the next leader upon the death of the incumbent.
Consistent with the argument in Chapter 6, when weak leaders insti-
tutionalize, we should expect to see an effect on leadership succession
outcomes.
What types of leadership transition outcomes should we expect to see
from regimes with strong leaders who do not institutionalize? Recall that
leaders who enter power already strong do not invest in succession
planning because they are not at risk of being deposed. Strong leaders
therefore do not institutionalize their regimes but are able to remain in
power for long periods of time. However, such regimes are unlikely to
experience a peaceful leadership transition upon the death of the leader.
In fact, such regimes often appear stable while the charismatic leader is
still in power, but then crumble upon the death of the leader due to the
lack of succession planning. Because strong leaders do not plan for
succession (due to the absence of any real threats to their authority
while they are in power), elites often face a coordination problem upon
the leader’s death. Without succession planning, it is unclear who the
next incumbent should be. The resulting power vacuum often leads to
a military coup as elites resort to violence in an attempt to usurp power.
Figure 7.1 summarizes my argument.
The implication of this argument for the empirical section of this
chapter is that we do not need to condition on initial leader strength
when estimating the effect of regime institutionalization on succession
outcomes. We should expect the presence of institutionalized succession

Stable regime

Weak
leader
Peaceful leadership
Figure 7.1 What is the effect of regime institutionalization on leadership succession?
215
rules to result in peaceful leadership transitions, and the absence of such

rules to result in violent transition, regardless of the motivations of the
leaders who created such policies.
7.6 data and descriptive statistics

The remainder of this chapter presents empirical tests of my arguments
about succession procedures and designated successors on leadership
transitions using my dataset.
7.6.1 Dependent Variable

The primary outcome of interest in this chapter is whether the leadership
transition was peaceful, therefore the unit of analysis is leadership transi-
tion rather than leader. The main dependent variable, “peace transition,”
is a dummy variable that takes a value of 1 if the leadership transition is
peaceful and 0 otherwise. I use the following rules to identify peaceful
leadership transitions. First, I require that a leader must have an immedi-
ate successor following their departure from office. To evaluate this, I use
the “start date” and “end date” variables available from the Archigos
dataset (Gleditsch and Chiozza 2009). Second, I require that the method
of exit for the incumbent and method of entry for the successor both be
coded as “regular” by Archigos.5 In other words, in order for the transi-
tion to be coded as peaceful, the incumbent cannot have been deposed
through a coup or civil war. The successor must take office immediately
following the departure of the previous incumbent, and the successor
cannot come to power using military force or foreign imposition.
An example of a peaceful leadership succession is when power was
transferred from Jomo Kenyatta to Daniel arap Moi in Kenya. Kenyatta
died on August 22, 1978, and Moi took office that day: both are coded as
regular entry in Archigos. Another example of a peaceful transition is
subsequently when power was transferred from Ahmadou Ahidjo to Paul
Biya in Cameroon. Ahidjo retired due to ill health on November 6, 1982
5
Archigos breaks down leader exit into the following main categories: assassination,
popular protest, removed by military, removed by other government actors, removed
by rebels, removed through threat of foreign force, regular, still in office. I consider
all exit codes other than assassination, regular, and still in office as an instance of
“irregular” leader exit. I exclude assassination from my list of irregular leader exits
because Archigos codes assassination attempts only by unsupported individuals.
U.S. presidents, for instance, have been assassinated but were peacefully succeeded
by their vice presidents. For leader entry, Archigos uses the following categories:
foreign imposition, irregular, regular, unknown. I only consider cases coded as
“regular” as peaceful entry.
216
7.6 Data and Descriptive Statistics
and Biya took office that same day. On the other hand, from 1960 through
1970, Benin experienced six leadership changes, all of which were driven
by coups. Finally, a number of incumbents were in power through the end
of the dataset in 2010, and I exclude such observations from the analysis
in this chapter. Robert Mugabe of Zimbabwe is such an example.
7.6.2 Independent Variables

For my key explanatory variables, I use three of my regime institutiona-
lization measures. Succession policies and term limits represent formal
constitutional rules that regulate the transfer of power. The stable
appointment of a vice president or prime minister represents the informal
designation of a successor. However, since my desired unit of analysis is
transition units, I collapse my country–year observations to cross-
sectional measures of succession rules.
For succession rules, I use the “specific successor” variable from my
dataset. This variable takes a value of 1 for all years in which the country
had a constitutional rule specifying who would be the interim successor in
the case of the departure or death of the president. Importantly, I only
consider succession rules that were in place in the years leading up to the
leadership transition. If, for example, a leader had succession rules at the
start of her tenure, then eliminated them before existing office, then these
rules would not have been in place to facilitate the transition. Therefore,
the variable, “succession rules” records the number of years during the
leader’s tenure leading up to the transition for which constitutional suc-
cession rules were in place.
In addition to succession rules, I evaluate whether having term limits
promotes peaceful transfers of power. The variable “term limits” records
the number of years during the leader’s tenure leading up to the transition
for which the constitutional term limits were in place.
I also evaluate whether designating a de facto successor has an effect on
the likelihood of peaceful transitions. To code this variable, I focus on vice
president and prime minister appointments that are stable. The variable
“stable VP/PM” records the number of years of the incumbent’s tenure in
which the vice president or prime minister is the same person as the
previous year.
In addition to my main hypotheses about succession planning, I also
create variables to test existing arguments surrounding hereditary succes-
sion and ruling parties. To evaluate whether power transitions between
family members are more stable, I utilize the family ties (“fties”) variable
in Archigos, which identifies the familial relationship between leaders.
The variable, “family ties” is a dummy variable that takes a value of 1 if
217
the transfer of power occurs between two family members. I include blood
ties and ties via marriage and do not exclude irregular transfers of power.
I also create a set of variables to test the party hypothesis. Ruling parties
are thought to provide a clear mechanism for leadership succession, and
this effect is likely to be stronger for more established parties. For data on
ruling parties, I merge the time-series version of my dataset on transitions
with the dataset on institutions from Svolik (2012). Using the party
variable from Svolik, I create the variable “party years,” which documents
the number of years the leader had a ruling party prior to the leadership
transition.6
7 . 7 des c r i p ti v e s t a t i s t i c s
The resulting data for this chapter includes 144 leadership transitions that
occurred between 1960 and 2010 in sub-Saharan Africa. Out of these
transitions, fifty-seven were peaceful and eighty-seven were not.7 As
documented in the literature on authoritarian rule, smooth leadership
transitions do not occur easily; only 40 percent of the transitions in this
sample were peaceful.
A number of interesting descriptive findings emerge from the data. The
first striking observation is that hereditary succession in sub-Saharan
Africa is surprisingly rare. Only 5 percent of power transfers (whether
peaceful or not) occur between family members.8 Out of these nine cases,
four were peaceful,9 and four were not,10 and one leader is still in power.
This finding is similar to patterns of hereditary succession in the global
sample of all authoritarian regimes. Out of all autocratic leadership
6
In order for the party to be included in this variable, I require that (1) The party must
be in power up until the transition. If, for instance, a leader bans a party prior to the
leadership transition, then the party could not have facilitated the transition. (2)
I also require that the party be in power for at least three years. In establishing the
three-year cutoff, I follow conventions in the literature (for example, see Geddes
1999), with the goal of excluding parties that were so new that they most likely did
not have the capacity to play a meaningful role in the leadership transition.
7
There were twenty-seven leaders still in power as of 2010, and I exclude these units
from the analysis.
8
It is useful to recall that I am using a broad definition of the term “family”; I include
relatives who are related by marriage in addition to blood relations.
9
For instance, in 1999, Ismail Omar Guelleh came to power as the president of
Djibouti, following his uncle Hassan Gouled Aptidon who had been the first pre-
sident of Djibouti from 1977 to 1999. Guelleh was the handpicked successor of
Aptidon and took over when his uncle retired.
10
For example, in 1966, Jean-Bedel Bokassa seized power from David Dacko in the
Central African Republic in a military coup. Bokassa was Dacko’s uncle.
218
7.8 Empirical Analysis
transitions that occurred between 1946 and 2016, power was handed
between family members less than 3 percent of the time.
Even family “dynasties” occur very infrequently. In my sample of
African leaders, only 6 percent of leaders have any family ties with earlier
incumbents.11 Once again, this statistic mirrors general patterns of family
dynasties found in the global sample of all autocratic leaders. Summary
statistics of all variables are reported in Appendix Table 7.1.
7 . 8 em p i r i c a l a n a l y s i s
To test the hypotheses presented in this chapter, I estimate a logit model
predicting peaceful transitions for each leadership transition i in country j.
It is represented as the following:
PrðPeaceful leadership transitionij ¼ 1Þ ¼ f ðβXij þ γW ij þ Þ

where Xij is a vector of transition-specific variables (such as the presence
succession rules) as well as leader-specific variables (such as time in office).
Wij is a vector of country-specific controls at the time of the transition
(such as GDP per capita or population size). Standard errors are clustered
at the country level for all model specifications.
Figure 7.2 presents graphs of the marginal effect of constitutional
succession rules and the presence of a de facto successor on the likelihood
of a peaceful transition, and Appendix Table 7.2 reports the full regres-
sion results of the logit analysis. The presence of constitutional rules
specifying succession procedures are positively associated with peaceful
transitions, and this effect is robust across various model specifications.
The presence of a de facto successor (measured by the stability of the vice
president/prime minister appointment) is also significantly associated
with peaceful transitions across all model specifications, suggesting that
informal institutions can play an important role in regulating power. The
size of these effects is not trivial. As the number of years a leader has
a constitutional successor policy increases from 0 to 35, the predicted
probability of a peaceful transition jumps from .3 to almost 1. The size of
the effect of a stable vice president or prime minister is similar in magni-
tude. The effects of term limits are much more inconsistent. For some
model specifications, term limits appear to be positively associated with
11
By family “dynasty” I am referring to leaders who have a familial connection with
an earlier incumbent, but did not immediately come into office following the family
member. For instance, Ian Khama, the current president of Botswana, came into
office in 2008. His father, Sereste Khama, left office in 1980.
219
Stable VP/PM Successor Rules

1 1
Pr(peaceful succession)
.8 .8
.6 .6
.4 .4
.2 .2
0 0
0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1 0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1
Proportion of years in power with Proportion of years in power with
stable VP/PM constitutional succession rules
Figure 7.2 Marginal effects of having a succession policy and designated successor on
likelihood of peaceful transition
peaceful transitions, but this effect is not robust to the addition to controls
(see Appendix Table 7.2).
The presence of a ruling party does not appear to have a significant
effect in any of the model specifications. In fact, the ruling party
variable is never significant in any subsequent models or robustness
checks presented in this chapter. In addition, because there are so few
cases of familial transfers of power (less than 5 percent of the sam-
ple), I drop this variable from the results presented in this chapter.
When included in the models, this variable is never significant.
One potential concern is that parties may be especially weak in
sub-Saharan Africa, particularly during the early decades
immediately following independence. To show that the absence of
a party effect is not simply an artifact of weak parties in Africa,
I analyze a global sample of leadership transitions from 1946 to
2008 and find that the presence of a ruling party does not increase
the likelihood of a peaceful leadership transition. I also find no
effect of hereditary succession on the likelihood of a peaceful
transition in the global sample. Appendix Table 7.3 reports these
results.
7.8.1 Robustness Checks

These findings remain consistent over a number of robustness checks.
First, I rerun the analysis on a subset of the data that excludes units that
do not have constitutions (in order words, only units that have constitu-
tions for the entire period are included in this analysis). The results,
220
7.9 Conclusion
reported in Appendix Table 7.4, remain consistent. The presence of for-

mal succession rules continues to be significantly associated with peaceful
transfers of power, as well as the appointment of a stable vice president or
prime minister.
Second, to ensure that my results are not purely being driven by
institutions that were created after the end of the Cold War in 1990,
I split the sample into two time periods. Models (1) and (2), reported
in Appendix Table 7.5, include leaders who were in power during the
Cold War period, from 1960 to 1990. Models (3) and (4) include
leaders who were in power after the end of the Cold War, from 1991
to 2010. The presence of constitutional successor rules continues to
be significant during both time periods, suggesting that these rules
had an effect on transitions of power even during the most author-
itarian decades.
7.9 conclusion
This chapter examined the effect of institutionalization on leadership
succession outcomes. I argued that the creation of constitutional
succession procedures and the designation of a de facto successor
(through the appointment of a stable vice president or prime minister)
plays a critical role in regulating peaceful leadership turnover.
Although leaders institutionalize out of self-interest, these measures
inadvertently create conditions for peaceful succession. The creation
of a clear line of succession provides certainty of the chain of com-
mand. Designated successors have an incentive to uphold the existing
regime since preserving the status quo is the best guarantee that they
will become the next incumbent.
I also argued that term limits, unlike succession rules, should not
have strong effect on succession outcomes. This is because term limits
do not solve the elite collective action problem by identifying
a specific elite challenger. Whereas the designation of a vice president
or prime minister or the designation of a line of succession in the
constitution identifies specific elites as a clear alternative leader, term
limits place constraints on the leader without empowering a specific
elite.
I analyzed all postindependence leadership transitions within sub-
Saharan Africa and provided empirical evidence for these arguments.
Altogether, Chapters 6 and 7 show that institutionalization does indeed
have stabilizing effects on key leader and regime outcomes.
221
Appendix
Variable N Mean SD Min Max
Peaceful transition 106 .386 .489 0 1

Succession rules 106 2.79 5.08 0 24
Stable VP/PM 106 3.73 5.23 0 35
Term limits 106 1.95 3.88 0 22
Constitution 106 6.84 7.87 0 41
Family ties 106 2043 .205 0 1
Party years 106 4.37 4.37 0 27
GDP per capita 101 1.29 1.61 .160 11.1
Oil production 101 .250 1.13 0 11.8
Population 101 8.71 1.17 6.16 11.6
Ethnic fractionalization 101 .664 .240 .035 .925
Years in power 106 8.50 8.20 1 41
Appendix Table 7.2 Does succession planning increase the likelihood of

peaceful leadership transitions?
DV: peaceful transition (1) (2) (3)
Stable VP/PM 0.257** 0.267** 0.245*

(0.086) (0.103) (0.117)
Successor rules 0.159** 0.159** 0.149*
(0.054) (0.055) (0.059)
Term limits 0.123* 0.123* 0.100
(0.055) (0.058) (0.070)
Constitution −0.093 −0.100 −0.145
(0.115) (0.125) (0.141)
Ruling party 0.042 0.030
(0.091) (0.088)
GDP per capita 0.906*
(0.423)
Oil production −0.060
(0.502)
Population 0.319
(0.225)
Ethnic fractionalization 0.423
(1.264)
(continued)
222
7.9 Conclusion
Cold War −0.911

(0.623)
British −0.103
(0.763)
French −0.388
(0.821)
Years in power 0.005 −0.007 0.066
(0.092) (0.099) (0.108)
Constant −1.842*** −1.806** −5.518*
(0.556) (0.570) (2.502)
Pseudo R-squared 0.286 0.289 0.400
Robust standard errors clustered by country in parentheses

*** p<0.001, ** p<0.01, * p<0.05
Appendix Table 7.3 Global sample
DV: Peaceful transitions (1) (2) (3)
Family ties 0.778 0.824 1.811

(0.560) (0.535) (0.969)
Ruling party 0.008 0.011 −0.027
(0.020) (0.020) (0.044)
Constant −2.936*** −3.130*** −3.147**
(0.720) (0.675) (1.183)
Pseudo R-squared 0.0613 0.0946 0.148
Country FE No No Yes
Region controls No Yes No

*** p < 0.001, ** p < 0.01, * p < 0.05
This sample includes all leadership transitions that have occurred in all dictatorships in the world
(as defined by Cheibub, Gandhi and Vreeland 2010). The family ties variable was retrieved from
the Archigos dataset and the ruling party variable was retrieved from Svolik (2012).
223
Appendix Table 7.4 Robustness check: only units with

constitutions
DV: peaceful transition (1) (2)
Successor rules 0.177** 0.175*

(0.066) (0.075)
Stable VP/PM 0.251** 0.199*
(0.095) (0.093)
Term limits 0.098 0.114
(0.057) (0.077)
Ruling party 0.005
(0.072)
(0.560)
(0.437)
Population 0.276
(0.227)
(1.981)
Years in power −0.078
(0.061)
Constant −1.846* −6.620*
(0.839) (2.599)
Observations 65 62
Pseudo R-squared 0.303 0.436
Note: Units without constitutions for the entire period are excluded
from this analysis.
*** p < 0.001, ** p < 0.01, * p < 0.05
224
7.9 Conclusion
Appendix Table 7.5 Robustness check: time trends
Cold War post-Cold War
DV: peaceful transition (1) (2) (3) (4)
Successor rules 0.112 0.148* 0.165 0.189**

(0.076) (0.071) (0.085) (0.068)
Stable VP/PM 0.215* 0.215 0.297 0.257
(0.101) (0.164) (0.191) (0.176)
Term limits 0.073 0.016 0.136 0.407
(0.064) (0.058) (0.100) (0.323)
Constitution −0.116 −0.241 −0.097 −0.144
(0.160) (0.155) (0.214) (0.262)
(0.106) (0.130)
GDP per capita 0.564 1.224
(0.326) (1.121)
Oil production 0.304 1.209
(0.400) (1.208)
Population 0.990 0.320
(0.667) (0.408)
Ethnic fractionalization −0.085 −0.547
(1.890) (1.499)
British 1.549 −2.020
(1.649) (1.334)
French −0.095 −0.710
(1.420) (1.472)
Years in power 0.036 0.193 −0.021 0.091
(0.130) (0.150) (0.149) (0.203)
Constant −0.895 −11.523 −2.328** −6.863
(0.612) (6.619) (0.762) (3.813)
Observations 47 44 59 57
Pseudo R−squared 0.158 0.339 0.323 0.505

*** p < 0.001, ** p < 0.01, * p < 0.05
225
7
In this chapter, we now turn to examine the effects of institutionalization

on key regime outcomes. In particular, we focus on whether highly insti-
tutionalized regimes are more likely to undergo peaceful leadership tran-
sitions. One of the fundamental challenges of continued authoritarian rule
is the transfer of power from one leader to another. Yet dictatorships
inherently lack mechanisms of electoral competition and leadership suc-
cession. Incumbents often remain in office for long periods of time, mak-
ing it difficult to establish norms of leader turnover. Elections – even when
they exist – are largely meant to sustain, rather than disrupt, the incum-
bent’s regime. As a result, the process of peacefully transferring power
from one leader to another is often quite precarious in dictatorships.
How can autocratic regimes overcome this fundamental challenge of
leadership succession? In this chapter I argue that constitutional succes-
sion rules and the appointment of a de facto successor play a critical role in
regulating the transfer of power. The institutionalization of a clear line of
succession provides certainty in the chain of command. Designated suc-
cessors have an incentive to uphold the existing regime, since preserving
the status quo is the best guarantee that they will become the next incum-
bent. Leaders who plan for succession are therefore less likely to be
deposed in a coup. Moreover, regimes with succession procedures are
significantly more likely to undergo a peaceful transition of power because
succession planning transforms conflict over succession from a single-shot
game after the leader’s death into a dynamic process in which elites
compete for power peacefully within the political arena.
However, I argue that only certain types of rules are effective in pro-
moting peaceful leadership succession. Constitutional amendments that
transfer of power. Such policies establish certainty and predictability
about the actors who are expected to come into office, providing these
elites with a stake in maintaining the existing regime in order to reap
205
accountable.
Importantly, the argument made in this chapter departs from the
argument made in Chapter 6 in the following way. In this chapter,
I argue that when estimating the effect of institutionalization on leader-
ship succession outcomes, we do not need to condition on the initial
strength of the leader. This is because peaceful succession outcomes are
an unintended consequence of regime institutionalization. As Chapter 6
demonstrated, self-interested leaders institutionalize their regimes in
order to insulate themselves against coup attempts. At the same time,
these measures – namely the creation of constitutional succession rules
and the appointment of a de facto successor – create certainty surrounding
the line of succession, therefore paving the way for a peaceful leadership
transition. While leaders do not necessarily care about the outcome of the
regime after their death, the measures they take to stabilize their rule while
in power nonetheless create conditions for the peaceful transfer of power.
This chapter makes two important contributions to theories of leader-
ship succession in dictatorships. First, it is one of the few studies to test
mechanisms of peaceful leadership change in autocracies. Though the
politics of succession is considered to be one of the central challenges of
autocratic rule, the mechanisms that facilitate peaceful leadership transi-
tions are not well understood for modern dictatorships. A limited number
of studies on this topic have focused on the role of hereditary succession
and ruling parties.
Early studies of leadership succession by Tullock (1987) focused on
importance of hereditary succession and primogeniture, and his the-
ories have since been extended and tested on pretwentieth century
European monarchies (Kurrild-Klitgaard 2000; Kokkonen and
Sundell 2014). However, the vast majority of authoritarian regimes
today are not monarchies. In fact, only 7 percent of all autocratic
regimes from 1946 to 2010 are coded as monarchies, and out of those
monarchies only two remain in power today (Geddes, Wright, and
Franz 2014).1 Furthermore, I show in this chapter that hereditary
succession is incredibly rare in modern dictatorships. Within my
1
Out of 280 autocratic regimes from 1946 to 2010, nineteen are coded as monarchies.
206
7.1 Do Dictators Care About Leadership Succession?
dataset only 5 percent of leadership transitions (whether peaceful or

not) were between family members.
A second group of studies suggests that ruling parties play an important
role in facilitating leadership succession. Scholars posit that an essential
role of ruling parties is that they help to mitigate and manage elite conflict,
providing an arena for politicians to compete for power and resources
(Brownlee 2007; Geddes 1999b; Magaloni 2008). Parties can also provide
a mechanism for elite power sharing (Boix and Svolik 2013; Svolik 2012).
Through these mechanisms, ruling parties can help facilitate peaceful
leadership transitions. However, in this chapter, I provide empirical evi-
dence that the presence of a ruling party is never a strong predictor of
peaceful leadership succession.
A second important contribution of this chapter is that it emphasizes
the fact that peaceful transitions of power have indeed occurred within
sub-Saharan African, even during the most authoritarian decades. Past
scholarship on leadership succession in sub-Saharan Africa has generally
held a pessimistic view about the prospect of peaceful or constitutional
transitions. Sylla and Goldhammer (1982), for instance, refer to the
challenge of succession as the “Gordian Knot of African Politics.”
Jackson and Rosberg (1982) predicted that “power politics rather
than institutionalized conventions and procedures may have to resolve
the succession issue” (69). By contrast, this chapter demonstrates that
not all leadership transitions that take place within autocracies in sub-
Saharan Africa are unregulated or driven solely by violence. Instead,
I show that institutional mechanisms can play an important role in
promoting peaceful successions. Regimes that are institutionalized are
significantly more likely to experience continuity across leadership
transitions.
7.1 do dictators care about leadership

s u c c e s si o n ?
As mentioned in the introduction of this chapter, dictators institutionalize
their regimes not necessarily because they care about succession politics
but in order to stabilize their own rule. After all, readers may have been
wondering why a dictator would care about leadership succession if the
transfer of power occurs after the leader has already died or left power.
However, it turns out that the problem of how to extend executive
authority to another elite is a concern not only to others in the regime,
but also to the dictator herself. As Herz (1957) so aptly summarizes: “It
also colors, in anticipation, so to speak, the entire situation during the
dictator’s life and rule. To the dictator it poses a problem and constitutes
207
a danger. To his aids it is a temptation. To the bystander, within and

without, it is fascination” (20).
Scholars originally believed that dictators were quite wary of naming
formal successors due to the “crown prince problem.” Herz famously
argued that by grooming a successor, incumbents create their own worst
enemy. The mere designation of an alternative center of power makes
leaders vulnerable to being deposed prematurely by their own chosen
successor. According to Herz, by naming a successor, the ruler voluntarily
shifts the center of power away from herself, putting herself at risk of
being deposed by her own appointee.
Yet not planning for succession is also equally dangerous for the regime
as well as the dictator herself. When autocratic leaders die without desig-
nating a successor a coordination problem emerges among regime elites
because it is not clear who the de facto successor should be (Kokkonen and
Sundell 2014). Such power vacuums in the aftermath of the death of the
leader often invite coups (Frantz and Stein 2017) or even civil wars
(Kokkonen and Sundell 2020) as elites vie for the incumbency.
In addition, leaders may actually endanger themselves by not planning
for succession. In the absence of an institutionalized succession order,
elites may be incentivized to preemptively try to take power through
coercion in anticipation of eventual conflict over succession. Indeed,
many dictators seem to fall prey to this dilemma, as the majority of
autocratic leaders from 1946 to 2008 have been deposed via nonconstitu-
tional means (Svolik 2012). In fact, leaders themselves often support the
institutionalization of succession policies because it buffers them from
coup attempts (Frantz and Stein 2017). In sum, succession planning
promotes stability during the incumbent’s tenure as well as for the regime
as a whole during the critical moments of leadership transition.
This argument builds on recent literature highlighting the importance
of constitutions in supporting autocratic rule (Albertus and Menaldo
2012; Ginsburg and Simpser 2014). Constitutional rules matter greatly
because they are publicly observable, therefore elites can condition their
behavior and credibly threaten to punish autocratic leaders who violate
existing rules. Autocratic constitutions are most helpful when they can
serve as “focal points” for elites by reducing ambiguity around a clear set
of rules and enforcement behavior (Albertus and Menaldo 2012; Hadfield
and Weingast 2013).2 In fact, Ginsburg and Simpser (2014) argue that
2
This argument is also related to Fearon’s (2011) concept of “self-enforcing democ-
racy” whereby elections provide citizens with a commonly understood set of rules
and procedures that allow them to credibly protest if the leader violates election
principles.
208
constitutional documents may even be especially helpful in autocracies

because the creation of clear and observable rules eases coordination
problems and creates a self-enforcing system – one that can operate in
the absence of a strong judicial system because deviations may be pun-
ished using force. In this way, formal institutions can also serve to “tie the
leader’s hands” and help make autocratic commitments credible, precisely
because creating constitutional rules makes leaders vulnerable to being
deposed if they violate the terms (North and Weingast 1989; Myerson
2008).
In this sense, the “crown prince problem” that Herz highlights actually
helps keep leaders in power. Because the process of designating
a successor and formalizing the line of succession shifts the distribution
of power in favor of elites, they are able to credibly sanction the leader in
the event that the leader reneges on distributive promises. As the model in
Chapter 2 emphasized, autocratic leaders can enhance the stability of their
own tenure by empowering other elites. This dynamic seems to play out
empirically as well. As the results from Chapter 6 demonstrate, leaders
who create constitutional succession rules and appoint a vice president or
prime minister do indeed face fewer coup attempts.
7.2 how does s uccessio n planning promote

regime stability?
Beyond preventing elites from preemptively launching coups in hopes of
capturing power, how does institutionalized succession planning promote
regime stability? I argue that regimes with constitutional amendments
specifying the process of leadership turnover are more likely to undergo
peaceful transitions compared with those without such formal rules.
In particular, three main mechanisms drive the relationship between
formal succession planning and peaceful transfers of power. First, consti-
tutional succession rules provide certainty and clarity surrounding the
chain of command. As a result, designated successors have an incentive
to protect the incumbent and existing regime, since preserving the status
quo is the best guarantee that they will become the next incumbent. Once
named, successors can also start to build their own alliances in anticipa-
tion of the leadership transition. Therefore, a clear line of succession
protects the regime from other potential elite challengers.
The transition in Kenya from President Jomo Kenyatta to Vice
President Daniel arap Moi illustrates this mechanism well. Kenya’s
first presidential transition occurred peacefully in 1978 upon the
death of Jomo Kenyatta, the country’s first president after indepen-
dence, although the struggle over the succession process occurred in
209
the final years of Kenyatta’s presidency. Kenyatta, a highly influential

figure in the country’s struggle for self-determination, took office upon
independence in 1964. In 1967, he named Daniel arap Moi as his vice
president, a position that Moi served for eleven years through the
transition. Notably, Moi was not a member of Kenyatta’s Kikuyu
ethnic group – a power-sharing mechanism by design. Cabinet positions
were often utilized as implicit contracts to maintain support across
various ethnic groups and factions, and Kenyatta always appointed
non-Kikuyus as vice president (Arriola 2009; Karimi and Ochieng
1980; Widner 1992).
The Kenyan constitution, which has been in effect since independence
in 1963, includes specific instructions for succession. Chapter II, Part I,
Section 6 reads:

resignation of the President . . . an election of a President shall be held
Members of the Kikuyu faction within the government did not want
Moi to ascend to the presidency because they feared that they would lose
valuable resources and influential positions that they enjoyed under the
Kenyatta presidency. Furthermore, they feared that the constitutional
succession rules solidified Moi’s role as successor to the presidency. In
the case of Kenyatta’s death, Moi would quickly use his role as interim
president to consolidate power within ninety days to confirm his new role
(Karimi and Ochieng 1980, 15).
In September of 1976, the Kikuyu faction, now calling themselves the
“Change the Constitution” group, proposed that the constitutional suc-
cession rule be changed so that the speaker of the National Assembly (a
position within the faction’s control) would be the interim president,
rather than the vice president. They called a meeting to discuss these
constitutional changes, and twenty MPs attended. Moi and his supporters
immediately responded by obtaining signatures from ninety-eight MPs
(including ten cabinet ministers) who opposed the proposed amendment
change. Importantly, this document showed that the Change the
Constitution group did not have a parliamentary majority needed to
210
pass a constitutional change. (Karimi and Ochieng 1980, 19–25;

Tamarkin 1979, 24; Widner 1992, 115–116).
The attorney general, who was a Moi supporter, further issued
a warning: “It is a criminal offence for any person to encompass, imagine,
devise, or intend the death or deposition of the President.” As a final nail in
the coffin, Kenyatta himself made a statement supporting the attorney
general’s earlier warning, despite the fact that many members of the
Change the Constitution group were from Kenyatta’s own family
(Tamarkin 1979, 24; Widner 1992, 166–117).3
Of course, the constitutional amendment was likely not the only
factor that led to the ascension of Moi to the presidency. Kenyatta’s
influence played a role in supporting Moi’s claims to the office, and the
Change the Constitution group lacked the support of a majority in the
parliament. However, the presence of formal succession rules acted as
a critical barrier against attempts by other elites to dethrone the
appointed successor. As Tamarkin (1979) effectively summarizes: “The
anti-Moi group suffered a severe setback in its first attempt to assert
itself. If they thought of pursuing their struggle they would have to do it
against the backdrop of a grim political reality and with a debilitating
constitutional constraint” (24).
A second benefit of formal succession rules is that such procedures
transform conflict over succession from a single-shot game after the
leader’s death into a dynamic process in which elites can compete for
power peacefully within the political arena. In fact, upon Kenyatta’s
death, as it became clear that Moi would indeed serve as the next pre-
sident, former members of the Change the Constitution group all publicly
proclaimed their loyalty to Moi. Mbiyu Koinange, a long-time cabinet
minister who had served Kenyatta since 1964, had been a participant of
the Change the Constitution group. Upon Moi’s quick ascension to the
presidency after Kenyatta’s death, Koinange made a public loyalty pledge
to Moi, stating “there is no truth whatsoever in the rumours spreading
abroad that I or any other respectable politician I know of in this stable
land of ours will be opposing the President” (as quoted in Karimi and
Ochieng 1980, 3). Koinange actually remained a cabinet minister the
first year of the Moi presidency (although he was given a less prestigious
portfolio) before being removed from the cabinet the following year.
A third benefit of constitutional succession rules is that they prevent
incumbents from reneging on promises made to their designated
3
Two of the leaders of the Change the Constitution movement were Dr. Njoroge
Mungai, Kenyatta’s nephew and personal doctor, and Mbiyu Koinange, his brother-
in-law.
211
successors. The 2012 transition from President Bingu wa Mutharika to his

vice president Joyce Banda in Malawi illustrates this mechanism. Like
Kenya, the Mawali constitution includes an amendment that states:
“Whenever there is a vacancy in the office of President, the First Vice-
President shall assume that office for the reminder of the term” (Section
83(3)). Though Mutharika initially promised to support Banda as his
successor, he reneged on this pledge and instead began to endorse his
younger brother, Peter Mutharika, as the next incumbent.4 Upon
Mutharika’s death in 2012, factions supporting Peter Mutharika tried to
contest Banda’s ascension to the presidency. These attempts ultimately
failed, as the courts and military supported the existing constitutional
succession plans (Cammack 2012; Dionne and Dulani 2013).
In sum, the creation of formal constitutional succession rules promotes
peaceful leadership transitions by protecting the existing regime against
outside challenges, in addition to keeping the incumbent accountable to
her successor. Although scholars of African politics have noted that such
constitutional procedures now play an important role in regulating the
transfer of power in democratizing countries in the post-Cold War era
(Dionne and Dulani 2013; Posner and Young 2007), I show that these
mechanisms played an important role in the transfer of power during the
autocratic era as well.
Hypothesis 1: Regimes that have a constitutional succession rule
7.3 designating informal successors

Beyond formal constitutional rules, I also argue that the stable appoint-
ment of a vice president or prime minister can also promote peaceful
transitions by signaling who the designated successor should be. Within
the context of sub-Saharan Africa, vice presidents and prime ministers
have long been understood to be the heir apparent of the president. Vice
presidents and prime ministers are commonly designated as the interim
president in the event of the death or incapacitation of the president. In
many cases, the interim president is even allowed to serve out the rest of
the former incumbent’s term. As Hughes and May (1988) note, this
provides an interim president with “an enormous advantage in
4
Mutharika could not simply remove Banda as vice president because the Malawi
constitution mandates that the vice president can only be removed by impeachment
(Dionne and Dulani 2013).
212
7.4 Succession Policies Versus Term Limits
consolidating his power. His constitutional authority places immediate

patronage in his hands and this is used to reward allies and eliminate
rivals” (14).
Even when the constitution does not specify the succession order, the
“second-in-command” position has generally been perceived as the most
likely stepping-stone to the presidency. As Table 4.1 from Chapter 4
shows, presidents are frequently former vice presidents or prime ministers.
Leon M’Ba, the first president of Gabon, for instance, appointed
a loyal second-in-command, Omar Bongo, to the vice presidency as his
heir apparent. “It is probably true that Bongo’s own political skills served
him well during the transitional period, but this does not gainsay the fact
that he was M’Ba’s personally selected [heir] and that he benefitted from
the approval and legitimacy that such selection bestowed” (Jackson and
Rosberg 1982, 72).
The implication of this argument is that succession planning does not
necessarily have to take place only on paper. Appointing an heir apparent,
similar to the creation of constitutional succession rules, empowers
a specific elite and incentivizes the successor to preserve the existing
regime. Therefore, if my argument about individual elites as focal points
is correct then we should expect that the appointment of de factor succes-
sors is as useful as constitutional succession rules in promoting peaceful
transitions.
Hypothesis 2: Regimes that have an appointed de facto successor
7.4 disaggregat ing autocratic constitutions:

s u c c es s i o n p o l i ci e s v e r s u s t e r m l i m i ts
Although autocratic constitutions may be an important source of regime
stability, I argue that not all constitutional amendments are equally effec-
tive in constraining leaders and solving the problem of leadership succes-
sion. In fact, there is important variation within constitutions: certain
types of amendments matter more and other types matter less. This argu-
ment builds on existing studies of comparative constitutions which high-
light variation in which different kinds of constitutional rights are
enforced (Chilton and Versteeg 2015). Chilton and Versteeg, for instance,
argue that not all rights are equally effective and in an analysis of con-
stitutions from 186 countries, they find that the right to unionize is more
often upheld than freedom of expression.
While I hypothesize that the implementation of constitutional rules
specifying procedures for succession play an important role in promoting
213
peaceful transitions, I predict that the presence of term limits do not.

Existing scholarship on leadership succession procedures in modern auto-
cracies suggest that term limits can be an important type of rule that
regulates the transfer of executive power by specifying when the incum-
bent must leave office. Term limits specify a finite length of time in which
the leader will hold office and may incentivize de facto successors to wait
for the leader to voluntarily leave office, rather than trying to stage a coup
prematurely.
However, term limits do not empower other elites in the same way that
succession rules do because term limits do not specify who the successor
will be. Returning to the example of Kenya, the constitutional amendment
that was implemented under Kenyatta specified that the person in the vice-
presidential role is to be the de facto successor in the event of the death of
the leader. Under this particular constitutional rule, Moi and his suppor-
ters had a lot to lose in the event that this particular amendment was
violated. Succession policies create certainty and establish predictability
about the actors who are expected to come into office and benefit from the
existing regime. Not only do such policies empower specific elites, they
also incentivize these elites to protect the existing regime in order to reap
the benefits of the existing succession order.
Term limits, on the other hand, do not identify a clear alternative leader
that other elites can coordinate around. When term limits are violated by
incumbents – which often occurs in autocracies – it is not clear who exactly is
being hurt the most, unless a specific elite has been designated as a de facto
successor. In sum, constitutional term limits do not help solve coordination
problems as effectively as succession procedures because term limits do not
identify a specific individual that elites should coordinate around.
In sum, we should not expect to see a strong effect of term limits on the
promotion of peaceful leadership transitions:
Hypothesis 3: Regimes that have term limits are not more likely to
have peaceful transfers of power than those who do not.
7.5 e ndogenous i nstitutionalization

As Chapter 6 highlighted, a concern that arises when trying to identify the
effects of institutions is that leaders who enter power especially weak
choose to institutionalize; these institutions are certainly not randomly
assigned to regimes. If succession planning is an endogenous decision and
only weak leaders tend to implement succession policies, should we expect
to see effects of these institutions on transition outcomes?
214
7.5 Endogenous Institutionalization Reconsidered
I argue that succession policies should have observable effects

because they affect not only the leader but also the designated suc-
cessor. When weak leaders choose to plan for succession in order to
prevent other elites from preemptively staging a coup, their desig-
nated successors also benefit inadvertently from this strategic plan-
ning. As Sections 7.2 and 7.3 argued, constitutional succession rules
and the appointment of a de facto successor provide certainty and
clarity surrounding the chain of command. Designated successors
have an incentive to protect the incumbent regime in hopes of
becoming the next leader upon the death of the incumbent.
Consistent with the argument in Chapter 6, when weak leaders insti-
tutionalize, we should expect to see an effect on leadership succession
outcomes.
What types of leadership transition outcomes should we expect to see
from regimes with strong leaders who do not institutionalize? Recall that
leaders who enter power already strong do not invest in succession
planning because they are not at risk of being deposed. Strong leaders
therefore do not institutionalize their regimes but are able to remain in
power for long periods of time. However, such regimes are unlikely to
experience a peaceful leadership transition upon the death of the leader.
In fact, such regimes often appear stable while the charismatic leader is
still in power, but then crumble upon the death of the leader due to the
lack of succession planning. Because strong leaders do not plan for
succession (due to the absence of any real threats to their authority
while they are in power), elites often face a coordination problem upon
the leader’s death. Without succession planning, it is unclear who the
next incumbent should be. The resulting power vacuum often leads to
a military coup as elites resort to violence in an attempt to usurp power.
Figure 7.1 summarizes my argument.
The implication of this argument for the empirical section of this
chapter is that we do not need to condition on initial leader strength
when estimating the effect of regime institutionalization on succession
outcomes. We should expect the presence of institutionalized succession

Stable regime

Weak
leader
Peaceful leadership
Figure 7.1 What is the effect of regime institutionalization on leadership succession?
215
rules to result in peaceful leadership transitions, and the absence of such

rules to result in violent transition, regardless of the motivations of the
leaders who created such policies.
7.6 data and descriptive statistics

The remainder of this chapter presents empirical tests of my arguments
about succession procedures and designated successors on leadership
transitions using my dataset.
7.6.1 Dependent Variable

The primary outcome of interest in this chapter is whether the leadership
transition was peaceful, therefore the unit of analysis is leadership transi-
tion rather than leader. The main dependent variable, “peace transition,”
is a dummy variable that takes a value of 1 if the leadership transition is
peaceful and 0 otherwise. I use the following rules to identify peaceful
leadership transitions. First, I require that a leader must have an immedi-
ate successor following their departure from office. To evaluate this, I use
the “start date” and “end date” variables available from the Archigos
dataset (Gleditsch and Chiozza 2009). Second, I require that the method
of exit for the incumbent and method of entry for the successor both be
coded as “regular” by Archigos.5 In other words, in order for the transi-
tion to be coded as peaceful, the incumbent cannot have been deposed
through a coup or civil war. The successor must take office immediately
following the departure of the previous incumbent, and the successor
cannot come to power using military force or foreign imposition.
An example of a peaceful leadership succession is when power was
transferred from Jomo Kenyatta to Daniel arap Moi in Kenya. Kenyatta
died on August 22, 1978, and Moi took office that day: both are coded as
regular entry in Archigos. Another example of a peaceful transition is
subsequently when power was transferred from Ahmadou Ahidjo to Paul
Biya in Cameroon. Ahidjo retired due to ill health on November 6, 1982
5
Archigos breaks down leader exit into the following main categories: assassination,
popular protest, removed by military, removed by other government actors, removed
by rebels, removed through threat of foreign force, regular, still in office. I consider
all exit codes other than assassination, regular, and still in office as an instance of
“irregular” leader exit. I exclude assassination from my list of irregular leader exits
because Archigos codes assassination attempts only by unsupported individuals.
U.S. presidents, for instance, have been assassinated but were peacefully succeeded
by their vice presidents. For leader entry, Archigos uses the following categories:
foreign imposition, irregular, regular, unknown. I only consider cases coded as
“regular” as peaceful entry.
216
7.6 Data and Descriptive Statistics
and Biya took office that same day. On the other hand, from 1960 through
1970, Benin experienced six leadership changes, all of which were driven
by coups. Finally, a number of incumbents were in power through the end
of the dataset in 2010, and I exclude such observations from the analysis
in this chapter. Robert Mugabe of Zimbabwe is such an example.
7.6.2 Independent Variables

For my key explanatory variables, I use three of my regime institutiona-
lization measures. Succession policies and term limits represent formal
constitutional rules that regulate the transfer of power. The stable
appointment of a vice president or prime minister represents the informal
designation of a successor. However, since my desired unit of analysis is
transition units, I collapse my country–year observations to cross-
sectional measures of succession rules.
For succession rules, I use the “specific successor” variable from my
dataset. This variable takes a value of 1 for all years in which the country
had a constitutional rule specifying who would be the interim successor in
the case of the departure or death of the president. Importantly, I only
consider succession rules that were in place in the years leading up to the
leadership transition. If, for example, a leader had succession rules at the
start of her tenure, then eliminated them before existing office, then these
rules would not have been in place to facilitate the transition. Therefore,
the variable, “succession rules” records the number of years during the
leader’s tenure leading up to the transition for which constitutional suc-
cession rules were in place.
In addition to succession rules, I evaluate whether having term limits
promotes peaceful transfers of power. The variable “term limits” records
the number of years during the leader’s tenure leading up to the transition
for which the constitutional term limits were in place.
I also evaluate whether designating a de facto successor has an effect on
the likelihood of peaceful transitions. To code this variable, I focus on vice
president and prime minister appointments that are stable. The variable
“stable VP/PM” records the number of years of the incumbent’s tenure in
which the vice president or prime minister is the same person as the
previous year.
In addition to my main hypotheses about succession planning, I also
create variables to test existing arguments surrounding hereditary succes-
sion and ruling parties. To evaluate whether power transitions between
family members are more stable, I utilize the family ties (“fties”) variable
in Archigos, which identifies the familial relationship between leaders.
The variable, “family ties” is a dummy variable that takes a value of 1 if
217
the transfer of power occurs between two family members. I include blood
ties and ties via marriage and do not exclude irregular transfers of power.
I also create a set of variables to test the party hypothesis. Ruling parties
are thought to provide a clear mechanism for leadership succession, and
this effect is likely to be stronger for more established parties. For data on
ruling parties, I merge the time-series version of my dataset on transitions
with the dataset on institutions from Svolik (2012). Using the party
variable from Svolik, I create the variable “party years,” which documents
the number of years the leader had a ruling party prior to the leadership
transition.6
7 . 7 des c r i p ti v e s t a t i s t i c s
The resulting data for this chapter includes 144 leadership transitions that
occurred between 1960 and 2010 in sub-Saharan Africa. Out of these
transitions, fifty-seven were peaceful and eighty-seven were not.7 As
documented in the literature on authoritarian rule, smooth leadership
transitions do not occur easily; only 40 percent of the transitions in this
sample were peaceful.
A number of interesting descriptive findings emerge from the data. The
first striking observation is that hereditary succession in sub-Saharan
Africa is surprisingly rare. Only 5 percent of power transfers (whether
peaceful or not) occur between family members.8 Out of these nine cases,
four were peaceful,9 and four were not,10 and one leader is still in power.
This finding is similar to patterns of hereditary succession in the global
sample of all authoritarian regimes. Out of all autocratic leadership
6
In order for the party to be included in this variable, I require that (1) The party must
be in power up until the transition. If, for instance, a leader bans a party prior to the
leadership transition, then the party could not have facilitated the transition. (2)
I also require that the party be in power for at least three years. In establishing the
three-year cutoff, I follow conventions in the literature (for example, see Geddes
1999), with the goal of excluding parties that were so new that they most likely did
not have the capacity to play a meaningful role in the leadership transition.
7
There were twenty-seven leaders still in power as of 2010, and I exclude these units
from the analysis.
8
It is useful to recall that I am using a broad definition of the term “family”; I include
relatives who are related by marriage in addition to blood relations.
9
For instance, in 1999, Ismail Omar Guelleh came to power as the president of
Djibouti, following his uncle Hassan Gouled Aptidon who had been the first pre-
sident of Djibouti from 1977 to 1999. Guelleh was the handpicked successor of
Aptidon and took over when his uncle retired.
10
For example, in 1966, Jean-Bedel Bokassa seized power from David Dacko in the
Central African Republic in a military coup. Bokassa was Dacko’s uncle.
218
7.8 Empirical Analysis
transitions that occurred between 1946 and 2016, power was handed
between family members less than 3 percent of the time.
Even family “dynasties” occur very infrequently. In my sample of
African leaders, only 6 percent of leaders have any family ties with earlier
incumbents.11 Once again, this statistic mirrors general patterns of family
dynasties found in the global sample of all autocratic leaders. Summary
statistics of all variables are reported in Appendix Table 7.1.
7 . 8 em p i r i c a l a n a l y s i s
To test the hypotheses presented in this chapter, I estimate a logit model
predicting peaceful transitions for each leadership transition i in country j.
It is represented as the following:
PrðPeaceful leadership transitionij ¼ 1Þ ¼ f ðβXij þ γW ij þ Þ

where Xij is a vector of transition-specific variables (such as the presence
succession rules) as well as leader-specific variables (such as time in office).
Wij is a vector of country-specific controls at the time of the transition
(such as GDP per capita or population size). Standard errors are clustered
at the country level for all model specifications.
Figure 7.2 presents graphs of the marginal effect of constitutional
succession rules and the presence of a de facto successor on the likelihood
of a peaceful transition, and Appendix Table 7.2 reports the full regres-
sion results of the logit analysis. The presence of constitutional rules
specifying succession procedures are positively associated with peaceful
transitions, and this effect is robust across various model specifications.
The presence of a de facto successor (measured by the stability of the vice
president/prime minister appointment) is also significantly associated
with peaceful transitions across all model specifications, suggesting that
informal institutions can play an important role in regulating power. The
size of these effects is not trivial. As the number of years a leader has
a constitutional successor policy increases from 0 to 35, the predicted
probability of a peaceful transition jumps from .3 to almost 1. The size of
the effect of a stable vice president or prime minister is similar in magni-
tude. The effects of term limits are much more inconsistent. For some
model specifications, term limits appear to be positively associated with
11
By family “dynasty” I am referring to leaders who have a familial connection with
an earlier incumbent, but did not immediately come into office following the family
member. For instance, Ian Khama, the current president of Botswana, came into
office in 2008. His father, Sereste Khama, left office in 1980.
219
Stable VP/PM Successor Rules

1 1
.8 .8
.6 .6
.4 .4
.2 .2
0 0
0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1 0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1
Proportion of years in power with Proportion of years in power with
stable VP/PM constitutional succession rules
Figure 7.2 Marginal effects of having a succession policy and designated successor on
likelihood of peaceful transition
peaceful transitions, but this effect is not robust to the addition to controls
(see Appendix Table 7.2).
The presence of a ruling party does not appear to have a significant
effect in any of the model specifications. In fact, the ruling party
variable is never significant in any subsequent models or robustness
checks presented in this chapter. In addition, because there are so few
cases of familial transfers of power (less than 5 percent of the sam-
ple), I drop this variable from the results presented in this chapter.
When included in the models, this variable is never significant.
One potential concern is that parties may be especially weak in
sub-Saharan Africa, particularly during the early decades
immediately following independence. To show that the absence of
a party effect is not simply an artifact of weak parties in Africa,
I analyze a global sample of leadership transitions from 1946 to
2008 and find that the presence of a ruling party does not increase
the likelihood of a peaceful leadership transition. I also find no
effect of hereditary succession on the likelihood of a peaceful
transition in the global sample. Appendix Table 7.3 reports these
results.
7.8.1 Robustness Checks

These findings remain consistent over a number of robustness checks.
First, I rerun the analysis on a subset of the data that excludes units that
do not have constitutions (in order words, only units that have constitu-
tions for the entire period are included in this analysis). The results,
220
7.9 Conclusion
reported in Appendix Table 7.4, remain consistent. The presence of for-

mal succession rules continues to be significantly associated with peaceful
transfers of power, as well as the appointment of a stable vice president or
prime minister.
Second, to ensure that my results are not purely being driven by
institutions that were created after the end of the Cold War in 1990,
I split the sample into two time periods. Models (1) and (2), reported
in Appendix Table 7.5, include leaders who were in power during the
Cold War period, from 1960 to 1990. Models (3) and (4) include
leaders who were in power after the end of the Cold War, from 1991
to 2010. The presence of constitutional successor rules continues to
be significant during both time periods, suggesting that these rules
had an effect on transitions of power even during the most author-
itarian decades.
7.9 conclusion
This chapter examined the effect of institutionalization on leadership
succession outcomes. I argued that the creation of constitutional
succession procedures and the designation of a de facto successor
(through the appointment of a stable vice president or prime minister)
plays a critical role in regulating peaceful leadership turnover.
Although leaders institutionalize out of self-interest, these measures
inadvertently create conditions for peaceful succession. The creation
of a clear line of succession provides certainty of the chain of com-
mand. Designated successors have an incentive to uphold the existing
regime since preserving the status quo is the best guarantee that they
will become the next incumbent.
I also argued that term limits, unlike succession rules, should not
have strong effect on succession outcomes. This is because term limits
do not solve the elite collective action problem by identifying
a specific elite challenger. Whereas the designation of a vice president
or prime minister or the designation of a line of succession in the
constitution identifies specific elites as a clear alternative leader, term
limits place constraints on the leader without empowering a specific
elite.
I analyzed all postindependence leadership transitions within sub-
Saharan Africa and provided empirical evidence for these arguments.
Altogether, Chapters 6 and 7 show that institutionalization does indeed
have stabilizing effects on key leader and regime outcomes.
221
Appendix
Variable N Mean SD Min Max
Peaceful transition 106 .386 .489 0 1

Succession rules 106 2.79 5.08 0 24
Stable VP/PM 106 3.73 5.23 0 35
Term limits 106 1.95 3.88 0 22
Constitution 106 6.84 7.87 0 41
Family ties 106 2043 .205 0 1
Party years 106 4.37 4.37 0 27
GDP per capita 101 1.29 1.61 .160 11.1
Oil production 101 .250 1.13 0 11.8
Population 101 8.71 1.17 6.16 11.6
Ethnic fractionalization 101 .664 .240 .035 .925
Years in power 106 8.50 8.20 1 41
Appendix Table 7.2 Does succession planning increase the likelihood of

peaceful leadership transitions?
Stable VP/PM 0.257** 0.267** 0.245*

(0.086) (0.103) (0.117)
Successor rules 0.159** 0.159** 0.149*
(0.054) (0.055) (0.059)
Term limits 0.123* 0.123* 0.100
(0.055) (0.058) (0.070)
Constitution −0.093 −0.100 −0.145
(0.115) (0.125) (0.141)
(0.091) (0.088)
(0.423)
(0.502)
Population 0.319
(0.225)
(1.264)
(continued)
222
7.9 Conclusion
Cold War −0.911

(0.623)
British −0.103
(0.763)
French −0.388
(0.821)
Years in power 0.005 −0.007 0.066
(0.092) (0.099) (0.108)
Constant −1.842*** −1.806** −5.518*
(0.556) (0.570) (2.502)
Pseudo R-squared 0.286 0.289 0.400

*** p<0.001, ** p<0.01, * p<0.05
Appendix Table 7.3 Global sample
DV: Peaceful transitions (1) (2) (3)
Family ties 0.778 0.824 1.811

(0.560) (0.535) (0.969)
Ruling party 0.008 0.011 −0.027
(0.020) (0.020) (0.044)
Constant −2.936*** −3.130*** −3.147**
(0.720) (0.675) (1.183)
Pseudo R-squared 0.0613 0.0946 0.148
Country FE No No Yes
Region controls No Yes No

*** p < 0.001, ** p < 0.01, * p < 0.05
This sample includes all leadership transitions that have occurred in all dictatorships in the world
(as defined by Cheibub, Gandhi and Vreeland 2010). The family ties variable was retrieved from
the Archigos dataset and the ruling party variable was retrieved from Svolik (2012).
223
Appendix Table 7.4 Robustness check: only units with

constitutions
DV: peaceful transition (1) (2)
Successor rules 0.177** 0.175*

(0.066) (0.075)
Stable VP/PM 0.251** 0.199*
(0.095) (0.093)
Term limits 0.098 0.114
(0.057) (0.077)
Ruling party 0.005
(0.072)
(0.560)
(0.437)
Population 0.276
(0.227)
(1.981)
Years in power −0.078
(0.061)
Constant −1.846* −6.620*
(0.839) (2.599)
Observations 65 62
Pseudo R-squared 0.303 0.436
Note: Units without constitutions for the entire period are excluded
from this analysis.
*** p < 0.001, ** p < 0.01, * p < 0.05
224
7.9 Conclusion
Appendix Table 7.5 Robustness check: time trends
Cold War post-Cold War
DV: peaceful transition (1) (2) (3) (4)
Successor rules 0.112 0.148* 0.165 0.189**

(0.076) (0.071) (0.085) (0.068)
Stable VP/PM 0.215* 0.215 0.297 0.257
(0.101) (0.164) (0.191) (0.176)
Term limits 0.073 0.016 0.136 0.407
(0.064) (0.058) (0.100) (0.323)
Constitution −0.116 −0.241 −0.097 −0.144
(0.160) (0.155) (0.214) (0.262)
(0.106) (0.130)
GDP per capita 0.564 1.224
(0.326) (1.121)
Oil production 0.304 1.209
(0.400) (1.208)
Population 0.990 0.320
(0.667) (0.408)
Ethnic fractionalization −0.085 −0.547
(1.890) (1.499)
British 1.549 −2.020
(1.649) (1.334)
French −0.095 −0.710
(1.420) (1.472)
Years in power 0.036 0.193 −0.021 0.091
(0.130) (0.150) (0.149) (0.203)
Constant −0.895 −11.523 −2.328** −6.863
(0.612) (6.619) (0.762) (3.813)
Observations 47 44 59 57
Pseudo R−squared 0.158 0.339 0.323 0.505

*** p < 0.001, ** p < 0.01, * p < 0.05
225
8
Conclusion
8.1 t aking s tock: authoritarian institutions

This book is motivated by the question of how executive power becomes
institutionalized in authoritarian regimes. Under what conditions does the
leader at the apex of power delegate authority to other elites, and why
would such a leader want to do so? The book opened with a quote from
Samuel Huntington’s famous volume, Political Order in Changing
Societies (1968). In that study, he examined the fundamental question of
how political systems become institutionalized. Though Huntington did
not make distinctions between regime types in his study, since then
political scientists have established the very important observation that
the institutional makeup of dictatorships does indeed vary across time and
space. Dictatorships, like democracies, differ greatly in the extent to which
they are institutionalized. This book explains how institutionalized forms
of dictatorship emerge.
Why is this topic so important? Although the field of comparative
authoritarianism has bourgeoned in the last twenty years, two major
challenges remain.
First, the existing literature has largely been focused on studying the
presence of nominally democratic institutions, such as parties, legisla-
tures, and elections. Scholars have developed a number of theories
explaining why autocratic leaders would voluntarily adopt institutions
such as parties and legislatures (Boix and Svolik 2013; Gandhi 2008;
Gandhi and Przeworski 2007; Svolik 2012). These studies highlight the
important idea that leaders choose to embrace nominally democratic
institutions because doing so helps to strengthen their rule. Whereas
political scientists and observers used to find it puzzling that dictatorships
would adopt institutions that resemble those found in democracies, these
studies emphasized the important idea that one-man rule was ultimately
226
8.1 Taking Stock
destabilizing and that autocrats were willing to trade unconstrained

power for more stable rule.
Yet, simply having institutions, such as parties or legislatures, does not
necessarily constrain dictators. This is easily demonstrated by the fact that
autocracies with ruling parties and legislatures are incredibly common in
the post-Second World War era. From 1946 to 2008, autocratic leaders
maintained a ruling party 87 percent of the time, and a legislature 85 per-
cent of the time. These percentages have approached 100 percent since the
end of the Cold War. Despite the prevalence of these institutions, scholars
agree that the vast majority of dictatorships are not highly institutiona-
lized or stable. During the same period, almost half of all leadership
transitions that occurred in dictatorships were violent and coups were
the most common way in which leaders were deposed. In fact, the appear-
ance of democratic-like institutions, such as ruling parties, often obscures
the true lack of constraints on the leader.
To sum up, although leaders do make strategic decisions regarding the
adoption of autocratic institutions, the mere presence of nominally demo-
cratic institutions cannot explain why some regimes persist while others
fall. Ruling parties, legislatures, or elections can often project and amplify
autocratic power, rather than constrain it, and this is true especially when
the institutions are weak. This book contends that the existing focus on
nominally democratic institutions – many of which are quite weak – has
diverted our focus of real and binding constraints on autocratic power. I
argue that rather than studying regime institutions, scholars should focus
on studying regime institutionalization.
Second, existing scholarship has largely focused on explaining the
effects of institutions on regime stability, rather than the origins of con-
strained dictatorship. In fact, this omission has been noted by scholars
(Magaloni 2006; 2008; Reuter 2017) but remains extremely understu-
died. If constrained forms of dictatorship do indeed have important
stabilizing and regime maintenance properties, then why don’t all auto-
cratic leaders adopt such constraints? How do institutionalized forms of
dictatorship emerge in the first place?
In addition to being an important scholarly omission, the origins of
these constraints affect our understanding of the effects of regime institu-
tionalization. As Pepinsky (2014) notes, no contemporary scholar of
authoritarian regimes believes that these institutions are randomly
assigned yet few studies have taken into account the endogenous adoption
of institutions when theorizing and estimating the effects of such institu-
tions. Leaders base decisions to institutionalize precisely on the presence
or absence of underlying regime threats, and these strategic decisions
affect the relationship between institutionalization and resulting regime
227
8 Conclusion
stability. Studies that examine the effects of autocratic institutions must

also necessarily consider the origins of these institutions.
There is one exception to the general lack of studies focusing on the
origins of autocratic institutions: a number of studies do examine the
emergence of institutionalized ruling parties or dominant-party regimes
(Brownlee 2007; Gehlbach and Keefer 2011; Levitsky and Way 2012;
Reuter 2017; Slater 2010; Smith 2005).1 However, the vast majority of
ruling parties in autocracies are simply not strong or institutionalized, and
most dictatorships are not true party-based regimes (Meng 2019). The
vast majority of ruling parties fail to survive past the death or departure of
the founding leader, suggesting that these organizations lack the institu-
tional infrastructure to rule beyond the influence of a particular leader.
Such regimes may have the façade of appearing as a party-based regime
while under the tutelage of a charismatic leader but quickly crumble upon
the death of the leader. Studies focusing on regimes with exceptionally
strong ruling parties, such as the CCP in China, Communist Party of the
Soviet Union, or PRI in Mexico, are likely explaining cases that resemble
outliers rather than the modal dictatorship. In other words, previous
studies lacked generalizable theories and empirical studies of how
constrained forms of authoritarian rule emerge.
Furthermore, by emphasizing the regime stabilizing effects of institu-
tions, existing studies have adopted a primarily functionalist approach to
explaining the existence of parties and legislatures. Yet if we examine the
historical record, it becomes quickly apparent that many ruling parties are
not strategically created, but instead inherited by leaders. Virtually all
leaders who come into power in newly postcolonial states do so with
already existing pro-independence parties. Even coup leaders often ally
with party elites and inherit already existing regime parties after taking
office. In other words, parties and legislatures often exist not because
incumbents strategically created them but because they already existed.
By contrast, this book argues that the real strategic decision incumbents
make is not whether to create these institutions, but whether to institu-
tionalize their regime.
This book makes five broad claims that address shortcomings in the
existing literature on comparative authoritarianism.
1. Regime institutionalization rather than the adoption of nominally
democratic institutions drives authoritarian stability.
1
Yet these studies still do not link the emergence of strong ruling parties to regime
outcomes and instead focus primarily on explaining the origins of party-based
regimes.
228
8.2 Empirical Contribution
2. In contrast to the prominence of “big men” theories of rule, African

dictatorships vary in their levels of institutionalization over space and
time. Beginning in the 1990s, institutionalization has allowed incum-
bents to solidify their hold on power.
3. Institutions provide credible commitments only when they change the
underlying distribution of power between actors. In authoritarian
regimes this can be done by providing elites with access to the state.
4. Autocratic leaders make decisions to institutionalize based largely on
the conditions under which they first come into power. These institu-
tional decisions lock in and shape the rest of their rule.
5. We must take into account the endogenous adoption of these con-
straints when evaluating the effects of institutionalization on regime
outcomes.
For the rest of this chapter, I will discuss each of these claims in more
detail and summarize the empirical and theoretical contributions of these
key findings. I will then conclude by considering the implications of my
argument for future studies of institutional design and democratic
backsliding.
8.2 e mpirical contribution: reexamining regime

i n st i t u t ion a l i z a ti o n
The first set of contributions of this book are empirical. I highlight the
importance of examining regime institutionalization as a distinct concept
rather than simply focusing on the presence of nominally democratic
institutions. Highly institutionalized regimes have explicit constraints on
presidential power, and I introduce a new dataset measuring institutiona-
lization by looking at constitutions and presidential cabinets. My mea-
sures demonstrate that existing variables of authoritarian institutions
serve as poor proxies for regime institutionalization. The data also reveals
a number of important findings for African politics, namely that institu-
tions in Africa are not uniformly weak and that the regional trend toward
institutionalization in the 1990s was an effort to reinforce authoritarian
rule rather than usher in true democracy.
8.2.1 Measuring Autocratic Regime Institutionalization

Although autocratic regime institutionalization is a variable which scho-
lars often reference, it has not been carefully defined or operationalized.
Developing high quality indicators of authoritarian institutions poses
a number of challenges, as dictatorships are often closed off or limit access
229
8 Conclusion
to accurate information. As a result, scholars are limited to a small num-

ber of large-N datasets that rely either on subjective expert coding or on
overly broad measures of the existence of institutions.
In this book I present a new cross-sectional time-series dataset measur-
ing autocratic regime institutionalization across all countries within sub-
Saharan Africa from 1960 to 2010. Using historical records of state
constitutions and presidential cabinets, I created seven indicators of
regime institutionalization:
1. whether the constitution had a leadership succession policy;
2. whether the constitution had a leadership succession policy that
designated a specific successor;
3. whether the constitution included term limits;
4. whether a vice president or prime minister was appointed to the
presidential cabinet;
5. whether the person in the vice president or prime minister post was
the same as the year before;
6. whether a defense minister was appointed in the presidential cabinet;
7. whether the person in the defense post was the same as the year
before.
that can be easily verified and replicated. Rather than hand-coding cases
according to subjective criteria, the data was constructed around
indicators that do not rely on the researcher to make subjective decisions
when coding the variables. Each variable takes the form of a dummy
variable, and simply asks whether a specific rule exists in the constitution
or whether a specific position is filled within the cabinet.
An additional advantage is that the data is disaggregated and time-
variant. Rather than collapsing multiple dimensions of institutionaliza-
tion into a single regime category, the data is presented as separate
indicators, therefore allowing the researcher to determine which dimen-
sion of institutionalization is most relevant to the research question at
hand. Moreover, yearly measures allow us to observe not only differences
across countries but also changes in levels of institutionalization over
time. Finally, this approach of measuring regime institutionalization
using data on constitutions and presidential cabinets is not unique to sub-
Saharan Africa and can be replicated in other regions of the world.
A comparison between my measures of institutionalization in
Africa against existing datasets reveals that simply having parties
and legislatures does not necessarily mean that explicit constraints
on executive power exist. Many regimes that have been coded as
dominant-party regime types are actually not very institutionalized,
230
8.2 Empirical Contribution
and expert-coded Polity and V-Dem scores display inconsistencies

when compared to my disaggregated indicators of regime institutio-
nalization. In sum, existing variables of autocratic institutions are
poor indicators of regime institutionalization because they often do
not reflect the extent to which the leader is actually constrained.
8.2.2 Regime Institutionalization in African Autocracies

This study also makes a number of important contributions to scholarship
on African politics. It focuses explicitly on strategies of autocratic rule in
sub-Saharan Africa, separating it from most existing studies that examine
governance, development, conflict, and democratization in Africa.
Furthermore, as a result of the proliferation of field, lab, and survey
experiments in the region, much of the existing literature retains
a strong focus on local-level politics. By contrast, this book presents
theory and empirics on elite politics in sub-Saharan Africa.
I show that autocratic regimes in Africa vary substantially in their levels
of institutionalization both across countries and over time. Autocratic
regimes in Botswana, Kenya, Mozambique, and Tanzania, for instance,
were highly institutionalized. Autocracies in countries such as Benin,
Mali, Malawi, and Niger exhibited low levels of institutionalization.
Whereas some leaders implemented few constraints on their authority
while they were in power, other leaders created constitutional rules and
delegated important cabinet positions to elites.
These empirical patterns establish an important finding: the image of
African dictatorships as uniformly weak and uninstitutionalized is
a false narrative. By providing empirical measures of autocratic regime
institutionalization in sub-Saharan Africa, this book challenges
a number of conventional wisdoms that prevail in the study of African
politics. Scholars have historically argued that African dictatorships
were ruled by “big men” who relied primarily on patronage-based
strategies rather than institutions to remain in power. These studies
generally disregarded constitutional rules and institutions in African
autocracies as meaningless or window dressing. In contrast with these
dominant perspectives, I demonstrate that there is rich variation in the
extent to which dictatorships in Africa are institutionalized. While some
autocrats, such as Mobutu Sese Seko of Zaire or Modibo Keita of Mali,
were indeed strongman dictators who ruled largely without constraints,
other leaders, such as Ahmadou Ahidjo of Cameroon or Seretse Khama
of Botswana created institutionalized mechanisms of power sharing that
limited their despotic power. This book shows that some African auto-
crats do indeed rule through institutions that regulate and depersonalize
231
8 Conclusion
power, and this is true even during the most authoritarian decades prior
to the end of the Cold War. Institutions are not uniformly weak in sub-
Saharan Africa.
The data also highlights that patterns of institutionalization change
over time. There is a clear difference between the modal African autocracy
before and after the end of the Cold War. In the decades from the 1960s
through the 1980s, weak leaders were compelled to institutionalize their
regimes during the independence era. As a result, about half of African
autocracies during this time established institutionalized power-sharing
mechanisms through cabinet ministerial appointments and roughly
a quarter created constitutional succession procedures. Following the
end of the Cold War from the 1990s through the present, most African
autocrats were forced to institutionalize their regimes and we see a sharp
increase in the creation of constitutional rules and stable cabinet
appointments.
Moreover, this book provides an alternative perspective on democra-
tization in Africa. I contend that trends toward institutionalization in the
1990s were an attempt by incumbents to recalibrate the political system in
order to entrench authoritarian stability, rather than genuine movement
towards political liberalization. Despite the introduction of multiparty
elections across most African nations following the end of the Cold War,
democratic consolidation remains stubbornly out of reach in most coun-
tries within the region. This book demonstrates how regime institutiona-
lization allowed incumbents and elites to strike credible and lasting
bargains of joint rule, even as external conditions changed. Rather than
ushering in real citizen-led democracies, the institutionalization of African
regimes in the 1990s introduced more durable rule-based forms of author-
itarian rule.
8.3 t heoretical contribution: t he origins and

co ns equences of regime institutionalization
The second key contribution of this book is theoretical. I examine the
causes and contributions of regime institutionalization. I provide a specific
mechanism of how institutions can provide credible commitment power,
rather than assuming that de jure rules on paper will be binding.
Importantly, this book considers how the strategic origins of institutiona-
lization shape the effects of such rules. The consequences of institutions
are intimately related to where these institutions came from. Rather than
considering the origins and effects of regime institutionalization in isola-
tion, this book provides a unifying account of the creation and conse-
quences of institutionalized dictatorship.
232
8.3 Theoretical Contribution
8.3.1 Sources of Credible Commitment Power in Autocracies

This book tackles the question of how certain types of institutions can
credibly constrain autocratic leaders. After all, as Svolik (2012) highlights,
one of the defining features of an authoritarian regime is the lack of an
independent authority with the power to enforce agreements amongst
actors. Furthermore, in dictatorships violence is always available as an
outside option. How then, do institutions become binding and self-
reinforcing? After all, a dictator who can create an institution can also
disassemble it as well.
I argue that institutions matter, not when they establish de jure rules,
but when they affect de facto political power. Institutions credibly con-
strain leaders only when they change the underlying distribution of power
between the incumbent and elites. Within the context of this book, the
establishment of executive constraints empowers elites by providing them
with access to the state. When leaders delegate key cabinet positions, such
as the vice presidency or minister of defense position, that particular elite
is given access to the power and resources that are associated with that
ministry. He then can then reward his own constituency, therefore con-
solidating his own base of support. Furthermore, elites that are assigned to
key positions within the cabinet, such as the vice president or prime
minister, then become focal points for other elites. This is particularly
true when the appointee is designated in the constitution as the legal
successor to the presidency. The designated successor is identified as
a potential challenger to the incumbent if she were to renege on promises
to distribute rent. To sum up, institutions that empower and identify
specific challengers help to solve elite collective action problems, and
this mechanism provides the key source of commitment power in auto-
cracies. Institutionalization limits despotic power by creating conditions
that actually threaten the leader.
8.3.2 The Origins of Institutionalized Dictatorship

This book examines how institutionalized forms of dictatorship emerge.
I argue that autocratic leaders make decisions to institutionalize at the
start of their tenure based largely on the conditions under which they
come into power. Since regime institutionalization places binding con-
straints on leaders, autocrats do not institutionalize unless they must.
Leaders who enter power weak and vulnerable to being deposed are
most likely to place constraints on their own authority. Because tempor-
ary offers of patronage are insufficient to buy quiescence from exception-
ally strong elites, weak leaders remain in power by delegating authority to
233
8 Conclusion
elites. Providing elites with access to the state shifts the distribution of
power away from the autocrat in favor of her appointees, providing them
with an institutional mechanism to guarantee their access to future rents.
Leaders who enter power extremely strong and having already consoli-
dated power have no incentives to create institutionalized mechanism for
rent distribution because they do not face credible threats to their rule.
Initially weak autocrats therefore adopt institutional constraints under
duress. They trade absolute power for regime stability.
This theory presents an interesting twist of fate. Initially strong leaders
are not incentivized to invest in mechanisms of joint rule because they are
able to remain in power without making institutional commitments to
other elites. Yet, this strategy of one-man rule is ultimately destabilizing
and does not last beyond the founding leader. On the other hand, initially
weak leaders must pursue the strategy of empowering other elites in order
to buy their support. These self-interested actions generate power-sharing
institutions, which can last beyond the tenure of that particular leader.
Durable authoritarianism emerges out of initial leader weakness.
My theory also stresses the path dependent nature of regime institutio-
nalization. Because violence is an ever-present option in autocracies,
incumbents necessarily worry about being deposed from the day that
they enter office. The initial distribution of power between the incumbent
and her elites shape her decision about the types of institutions to set up at
the very start of her rule. However, as institutionalization empowers other
elites, it becomes self-reinforcing and difficult to remove. Even if the
incumbent becomes stronger in later periods, she remains constrained by
the very rules she implemented at the start of her tenure.
This argument attempts to address critiques of the authoritarian institu-
tions literature as being overly functionalist. Although in my account, leaders
are indeed making strategic decisions to institutionalize or not, these deci-
sions have a “lock-in” effect and are not easily disposable. Leaders cannot
construct and disassemble institutions on a whim. Whether a leader is
especially vulnerable to being deposed at the start of her tenure is historically
determined by the way in which she obtained power. This initial distribution
of power affects her decision to institutionalize, which in turn has long-
lasting consequences on the institutional makeup of the regime.
8.3.3 Reassessing the Effects of Institutionalization on Autocratic

Durability
After providing a theory of the endogenous emergence of executive con-
straints, the book reexamines the relationship between institutionaliza-
tion and autocratic durability. Do institutionalized regimes actually
234
8.3 Theoretical Contribution
perform better on outcomes such as the length of leader tenure, vulner-

ability to coup attempts, and leadership succession? By contrast to most
previous accounts, this book considers the fact that rules are strategically
adopted by autocrats when determining the effects of these institutions on
regime outcomes.
Building on the theory developed in the first half of the book, I argue that
we should observe differential effects of institutionalization on autocratic
durability for initially strong versus initially weak leaders. Since leaders
who enter office already strong do not need to rely on power-sharing
institutions to remain in power, we should observe little or no effect of
institutionalization on durability for strong leaders. Their rule is secure
with or without institutions. For leaders who enter office weak, however,
institutionalization should have an effect on autocratic durability because it
provides a credible mechanism for long-term rent distribution to elites. In
sum, institutionalization does have an effect on regime outcomes, such as
leader tenure or vulnerability to coups, but primarily for weak leaders.
These findings make an important contribution to a large body of
existing scholarship focusing on estimating the effects of autocratic insti-
tutions on regime durability. Despite the general consensus that institu-
tionalized forms of dictatorship tend to be the most stable, the empirical
record of the relationship between institutions and regime stability has
been very mixed. While some studies (such as Boix and Svolik 2013;
Brownlee 2007; and Geddes 1999a) have found a significant effect of
institutions on regime durability, other studies (such as Gandhi 2008;
and Smith 2005) have found no systematic relationship.
This book provides two reasons why existing studies have produced
divergent empirical findings. First, the existing findings may be subject to
a number of measurement problems due to the difficulties of obtaining
fine-grained data that accurately reflects the extent to which the leader is
constrained. As the book demonstrated in Chapter 4, simply having
institutions such as parties and legislatures is not a strong predictor of
credible executive constraints, and many regimes that have been coded as
party-based are actually not very institutionalized. Second, existing
studies are also subject to biased theoretical modeling. This book argues
that different types of leaders make different institutional decisions when
they come into power. These strategic decisions are based on underlying
threat perceptions. Leaders who enter office strong do not institutionalize
because they do not need to do so in order to remain in power.
A regression model that regresses institutionalization on regime stability
without taking into account differential effects based on leader type will
likely result in either a diminished effect or null effect of institutionaliza-
tion on autocratic stability, both of which would be inaccurate.
235
8 Conclusion
Finally, I also show that highly institutionalized regimes are more likely
to survive peaceful leadership transitions. However, the book notes that
peaceful transitions of power are an unintended consequence of strategic
institutionalization by leaders. Incumbents create constitutional rules
designating the succession order and appoint elites to high-ranking posi-
tions within the presidential cabinet not necessarily out of concern for the
stability of regime after they die. Instead, they make these institutional
decisions to stabilize their own rule and prevent elites from preemptively
trying to depose them. Yet these self-interested actions empower the
designated successor, who has every incentive to remain loyal to the
current regime in the hope of being the next incumbent.
The findings from these two chapters produce two distinct types of
autocratic regimes. The first is that of the institutionalized dictatorship.
Such regimes may appear initially weak due to the absence of
a charismatic and influential figurehead. Yet, durable institutions emerge
out of autocratic weakness. Since initially weak leaders cannot rely on
personalist strategies of rule to stay in power, such leaders must necessa-
rily delegate authority to other elites. These institutional arrangements
become self-reinforcing over time and joint rule allows the regime to
survive past the tenure of a single leader.
The second is that of the strongman dictatorship. This leader enters
office having already consolidated power and is often a charismatic
“founding father” figure. Strongman leaders do not need to institutio-
nalize their regimes in order to maintain support from other elites
because they do not face any viable elite challengers of equal status.
However, due to the lack of succession planning and absence of institu-
tionalized rules and organizational structure that can operate indepen-
dently of the ruler, these regimes fall apart upon the death of the
strongman leader.
Yet it is important to note that personalist dictators can often remain in
power for very long periods of time and in fact often die in office. Dictators
such as Francisco Franco who ruled Spain from 1939 until his death in
1975 or Mobutu Sese Seko who ruled Zaire from 1965 to 1997, remained
in power for decades. Because leaders endogenously respond to threats,
regime institutionalization and the length of leader tenure are not strongly
correlated. However, this mode of personalist rule generally cannot be
passed on beyond the tenure of one leader. While strongman dictatorships
appear stable while the incumbent is still alive, such regimes do not survive
leadership transitions. As a result, institutionalized regimes do indeed
survive longer than uninstitutionalized regimes by solving the leadership
succession problem. The longest surviving autocracies, such as those in
China, Tanzania, and Mexico under the PRI, reflect this pattern.
236
8.4 Looking Ahead
8.4 looking ahead: i mplications

for institutional design, democratization,
and democratic backsliding
This chapter concludes by considering the implications of autocratic
regime institutionalization for future studies of institutional design,
democratization, and democratic backsliding. One of the themes of
this book is that institutional change is path dependent. Since effective
constraints are those that bind, the institutional decisions that leaders
make early on in their tenure often shape the rest of their rule. Another
important implication of this study is that the initial period of a leader’s
tenure often constitutes a critical juncture. How a leader enters office
determine the power dynamics within the regime coalition, and the
relative strength of the incumbent versus elites shapes the institutional
arrangements that emerge at the start of the regime. Founding leaders
tend to be dangerously personalist. Even though the strategy of strong-
man rule may be sustainable through the lifespan of the first leader, such
systems frequently crumble when the leader dies, revealing underlying
institutional weaknesses within the regime. On this front, scholars and
practitioners should pay close attention to first leadership transitions,
which often expose the extent to which the regime can operate indepen-
dently of any particular leader.
This book also has important implications for theories of democratiza-
tion. I demonstrate that autocratic leaders can often appear democratic by
introducing nominally democratic institutions, such as parties, legisla-
tures, and elections. Yet these incumbents and their elites retain political
power through institutionalized arrangements. This suggests that
researchers and policymakers may be misinterpreting highly institutiona-
lized autocracies for countries undergoing democratic transitions, when in
fact, real access to power remains out of reach for ordinary citizens.
Future research should pay greater attention to the quality of democratic
institutions in transitioning countries rather than simply observing the
presence or absence of these institutions.
In fact, even the introduction of multiparty elections does not necessa-
rily result in democratic consolidation. As recent scholarship on elections
in sub-Saharan Africa has demonstrated, incumbents routinely exploit
access to state resources to coopt opposition parties and politicians
(Arriola, DeVaro, and Meng 2018, Bleck and van de Walle 2018).
Importantly, these strategies can occur even in the absence of fraud or
explicit rule breaking. Leaders can exploit free and fair elections to
amplify their own incumbency advantages. Electoral competition may
not necessarily become institutionalized even when formal electoral
237
8 Conclusion
institutions are in place (Posner and Young 2007). These themes suggest
that future scholarship on democratic transitions and consolidation
should pay more attention to the conditions under which the introduction
of multiparty elections actually produce electoral turnover.
Finally, this book also provides insights that address growing concerns
about democratic backsliding. Democratic backsliding, also referred to as
democratic erosion, broadly describes the process of “state-led debilita-
tion or elimination of any of the political institutions that sustain an
existing democracy” (Bermeo 2016, 5). It is a topic that has gained
increasing attention but has received little scholarly research in political
science thus far. Scholars are now in the early stages of conceptualizing
and measuring democratic backsliding in addition to understanding the
causes and consequences of it.
On this topic, this book offers two important perspectives. First, any
regime – whether democratic or authoritarian – can have institutions that
constrain executive power. Therefore, the debilitation or elimination of
political institutions is not and should not be limited to the study of
democracies. Institutionalized forms of dictatorship can backslide as
well. This has been demonstrated by the removal of constitutional term
limits by President Xi Jinping of China in 2018 and President Paul Biya of
Cameroon in 2008. Scholars should therefore pursue the study of de-
institutionalization across all regime types, without restricting analysis
of institutional erosion to democracies.
Second, this book demonstrates that institutional constraints are most
effective when they empower specific political actors who can check the
authority of the president. Not all constitutional rules have the same stick-
ing power, and certain types of institutions may be easier to remove or
weaken than others. As Chapter 7 argued, term limits are easier to ignore or
remove compared with constitutional succession procedures because term
limits do not empower particular elite challengers. Therefore when an
incumbent violates term limits, elites have a more difficult time overcoming
coordination problems required to sanction the incumbent because it is not
clear who exactly is losing out the most. Future research can consider this
mechanism of elite empowerment when trying to understand the conditions
under which leaders can weaken or remove institutional constraints on
their power. Constitutional rules matter, not because they exist on paper,
but only when they reshape political power.
238
References
Acemoglu, D. and J. Robinson (2001). A theory of political transitions. American

Economic Review 91(4), 938–963.
Acemoglu, D. and J. Robinson (2006). Economic Origins of Dictatorship and
Democracy. New York: Cambridge University Press.
Adamolekun, L. (1976). Towards development-oriented bureaucracies in Africa.
International Review of Administrative Sciences 42(3), 257–265.
Albertus, M. and V. Menaldo (2012). Dictators as founding fathers? The role of
constitutions under autocracy. Economics and Politics 24(3), 279–206.
Albertus, M. and V. Menaldo (2018). Authoritarianism and the Elite Origins of
Democracy. Cambridge University Press.
Alt, J., J. Aldrich, and A. Lupia (Eds.) (2007). Positive Changes in Political Science:
The Legacy of Richard D. McKelvey’s Most Influential Writings. University of
Michigan Press.
Arriola, L. (2009). Patronage and political stability in Africa. Comparative
Political Studies 42(10), 1339–1362.
Arriola, L. (2013). Multi-ethnic Coalitions in Africa: Business Financing of
Opposition Election Campaigns. Cambridge University Press.
Arriola, L., J. DeVaro, and A. Meng (2018). Elite cooptation and opposition
fragmentation in electoral authoritarian regimes. Working paper. University
of Virginia.
Bermeo, N. (2016). On democratic backsliding. Journal of Democracy 27(1),
5–19.
Bienen, H. (1970). Tanzania: Party Transformation and Economic Development.
Princeton University Press.
Bienen, H. (1978). Armies and Parties in Africa. New York: Africana Publishing.
Bienen, H. and N. van de Walle (1989). Time and power in Africa. American
Journal of Political Science 83(1),19–34.
Bienen, H. and N. Van de Walle (1991). Of Time and Power: Leadership Duration
in the Modern World. Stanford University Press.
Blaustein, A. P. and G. H. Flanz (Eds.) (1976). Constitutions of the Countries of
the World: Angola. New York: Oceana Publications.
Blaydes, L. (2010). Elections and Distributive Politics in Mubarak’s Egypt.
New York: Cambridge University Press.
Bleck, J. and N. van de Walle (2018). Electoral Politics in Africa Since 1990.
Cambridge University Press.
239
References
Boix, C. and M. Svolik (2013). The foundations of limited authoritarian

government: institutions, commitment, and power-sharing in dictatorships.
The Journal of Politics 75(2), 300–316.
Boone, C. (2014). Property and Political Order in Africa. New York: Cambridge
University Press.
Borocz, J. (2000). Informality rules. East European Politics and Societies 14(2),
348–380.
Bracanti, D. (2014). Democratic authoritarianism: origins and effects. Annual
Review of Political Science 17, 313–326.
Bratton, M. and N. van de Walle (1997). Democratic Experiments in Africa.
New York: Cambridge University Press.
Brownlee, J. (2007). Authoritarianism in an Age of Democratization. New York:
Buchenau, J. (2006). Plutarco Elías Calles and the Mexican Revolution. Lanham,
MD: Rowman & Littlefield.
Buckles, G. and J. Gandhi (2017). Opposition unity and cooptation in hybrid
regimes. Working paper. Atlanta, GA: Emory University.
Camara, M. S. (2005). His Master’s Voice: Mass Communication and Single-party
Politics in Guinea Under Sekou Toure. Trenton, NJ: Africa World Press.
Cammack, D. (2012). Malawi in crisis, 2011–12. Review of African Political
Economy 39(132), 375–388.
Carey, J. M. (2000). Parchment, equilibria, and institutions. Comparative
Political Studies 33(6–7), 735–761.
Carter, B. L. and M. Hassan (2019). Regional governance in divided societies:
evidence from the Republic of Congo and Kenya. Journal of Politics. https://doi
.org/10.1086/708915.
Cheeseman, N. (2015). Democracy in Africa: Successes, Failures, and the Struggle
for Political Reform. New Approaches to African History. Cambridge
University Press.
Chehabi, H. and J. Linz (Eds.) (1998). Sultanistic Regimes. Baltimore, MD: John
Hopkins University Press.
Cheibub, J. A., J. Gandhi, and J. R. Vreeland (2010). Democracy and dictatorship
revisited. Public Choice 143(1), 67–101.
Chilton, A. S. and M. Versteeg (2015). Do constitutional rights make a difference?
American Journal of Political Science 60(3), 575–589.
CIA (1983). Cameroon: Biya’s Political Prospects. Washington, DC: Central
Intelligence Agency.
CIA (1984). Cameroon: Challenges to Biya’s Leadership. Washington, DC:
Central Intelligence Agency.
Clodfelter, M. (1992). Warfare and Armed Conflicts. Jefferson, NC: McFarland.
Collier, R. (1982). Regimes in Tropical Africa. Berkeley: University of California
Press.
Committee on Foreign Relations (1978). Uganda: The Human Rights Situation:
Hearings Before the Subcommittee on Foreign Economic Policy of the
Committee on Foreign Relations. June 15, 21, 26. Washington: US
Government Printing Office.
Coppedge, M., J. Teorell, S. Lindberg, and S.-E. Skaaning (2016). Measuring high
level democratic principles using the V-Dem data. International Political
Science Review 37(5), 580–593.
240
References
Cox, G. (2016). Marketing Sovereign Promises: Monopoly Brokerage and the

Growth of the English State. Cambridge University Press.
Daddieh, C. (2001). Elections and ethnic violence in Cote d’Ivoire: the unfinished
business of succession and democratic transition. African Issues 29(12), 14–19.
Dahl, R. (1972). Polyarchy. New Haven, CT: Yale University Press.
De Ayala, R. (2009). The Theory and Practice of Item Response Theory.
Methodology in the Social Sciences. New York: Guilford Publications.
Debs, A. and N. P. Monteiro (2014). Known unknowns: Power shifts, uncertainty,
and war. International Organization 68(1), 1–31.
Dal Bó, E. and R. Powell (2009). A model of spoils politics. American Journal of
Political Science 53(1), 207–222.
Decalo, S. (1976). Coups and Army Rule in Africa. Studies in Military Style. New
Haven, CT: Yale University Press.
Decalo, S. (1990). Coups and Army Rule in Africa: Motivations and Constraints.
New Haven, CT: Yale University Press.
DeLancey, M. (1989). Dependence and Independence. Boulder, CO: Westview
Press.
Diamond, L. (2002). Thinking about hybrid regimes. Journal of Democracy 13(2),
21–35.
Dickie, J. and A. Rake (1973). Who’s Who in Africa: The Political, Military and
Business Leaders of Africa. London: African Development.
Dionne, K. Y. and B. Dulani (2013). Constitutional provisions and executive
succession: Malawi’s 2012 transition in comparative perspective. African
Affairs 112(446), 111–137.
Dower, P. C., E. Finkel, S. Gehlbach, and S. Nafziger (2017). Collective action and
representation in autocracies: Evidence from Russia’s great reforms. American
Political Science Review 112(1), 125–147.
Dresden, J. R. (2017). From combatants to candidates: electoral competition and
the legacy of armed conflict. Conflict Management and Peace Science 34(3),
240–263.
Dunning, T. (2008). Improving causal inference: strengths and limitations of
natural experiments. Political Research Quarterly 61(2), 282–293.
Elkins, Z., T. Ginsburg, and J. Melton (2014). Imagining a world without the
universal declaration of human rights. SSRN. March, 8. http://dx.doi.org/10
.2139/ssrn.2469194.
Embretson, S. E. and S. P. Reise (2000). Item Response Theory for Psychologists.
Mahwah, NJ: L. Erlbaum Associates.
Europa Publications (1960–2010). The Europa World Year Book. New York:
Europa Publications.
Fearon, J. (1995). Rationalist explanations for war. International Organization
49(3), 379–414.
Fearon, J. (2011). Self-enforcing democracy. Quarterly Journal of Economics 126
(4), 1661–1708.
Francois, P., I. Rainer, and F. Trebbi (2015). How is power shared in Africa?
Econometrica 83(2), 465–503.
Frantz, E. and E. A. Stein (2017). Countering coups: leadership succession rules in
dictatorships. Comparative Political Studies 50(7), 935–962.
Gandhi, J. (2008). Political Institutions under Dictatorship. New York:
241
References
Gandhi, J. and E. Lust-Okar (2009). Elections under authoritarianism. Annual

Gandhi, J. and A. Przeworski (2007). Authoritarian institutions and the survival
of autocrats. Comparative Political Studies 40(11), 1279–1301.
Gandhi, J. and R. Ruiz-Rufino (2015). Routledge Handbook of Comparative
Political Institutions. Abingdon: Routledge.
Geddes, B. (1999a). Authoritarian breakdown: empirical test of a game-theoretic
argument. Paper presented at the annual meeting of the American Political
Science Association, Atlanta, September 1999. https://pdfs.semanticscholar.org/
ff90/c7b0c6f0633dc894d4c2ced78c78ddfef463.pdf.
Geddes, B. (1999b). What do we know about democratization after twenty years?
Annual Review of Political Science 2, 115–144.
Geddes, B. (2003). Paradigms and Sand Castles: Theory Building and Research
Design in Comparative Politics. University of Michigan Press.
Geddes, B. (2009). How autocrats defend themselves against armed rivals. Paper
presented at the American Political Science Association annual meeting, Toronto,
2009. https://pdfs.semanticscholar.org/9c63/fea96ba579bccf8f4898777c0d01
fa16a29c.pdf.
Geddes, B., J. Wright, and E. Frantz (2012). Autocratic regimes code book.
Version 1.2. Working Paper. Los Angeles, UCLA.
Geddes, B., J. Wright, and E. Frantz (2014). Autocratic breakdown and regime
transitions: a new data set. Perspectives on Politics 12(2), 313–331.
Geddes, B., J. Wright, and E. Frantz (2018). How Dictatorships Work: Power,
Personalization, and Collapse. Cambridge University Press.
Gehlbach, S. and P. Keefer (2011). Investment without democracy: ruling-party
institutionalization and credible commitment in autocracies. Journal of Com-
parative Economics 39(2), 123–139.
Gehlbach, S., K. Sonin, and M. W. Svolik (2016). Formal models of
nondemocratic politics. Annual Review of Political Science 19, 565–584.
Ginsburg, T. and A. Simpser (2014). Constitutions in Authoritarian Regimes.
Gleditsch, N. P., P. Wallensteen, M. Eriksson, M. Sollenberg, and H. Strand
(2002). Armed conflict 1946–2001: a new dataset. Journal of Peace Research
39(5), 615–637.
Glickman, H. (ed.) (1992). Political Leaders of Contemporary Africa South of the
Sahara. Westport, CT: Greenwood Press.
Goemans, H., K. S. Gleditsch, and G. Chiozza (2009). Introducing Archigos:
a dataset of political leaders. 46(2), 269–283.
Granato, J. and F. Scioli (2004). Puzzles, proverbs, and omega matrices: the
scientific and social significance of empirical implications of theoretical
models (EITM). Perspectives on Politics 2(2), 313–323.
Greif, A. and D. D. Laitin (2004). A theory of endogenous institutional change.
American Political Science Review 98(4), 633–652.
Grzymala-Busse, A. (2007). Encouraging effective democratic competition. East
European Politics and Societies 21(1), 91–110.
Grzymala-Busse, A. (2010). Institutions and the fate of democracy. Journal of
Cold War Studies 12(1), 181–183.
Grzymala-Busse, A. (2011). Time will tell? temporality and the analysis of causal
mechanisms and processes. Comparative Political Studies 44(9), 1267–1297.
242
References
Gyimah-Boadi, E. and C. Dad (1999). Economic Reform and Political

Liberalization in Ghana and Côte d’Ivoire: a Preliminary Assessment of
Implications for Nation Building. Westport, CT: Praeger.
Haber, S. (2008). Authoritarian government in B. R. Weingast and D. A. Wittman
(eds.) The Oxford Handbook of Political Economy. Oxford University Press.
693–707.
Haber, S., N. Maurer, and A. Razo (2003). The Politics of Property Rights:
Political Instability, Credible Commitments, and Economic Growth in
Mexico, 1876–1929. Cambridge University Press.
Hadfield, G. and B. Weingast (2013). Constitutions as coordinating devices.
SSRN. September, 1. https://papers.ssrn.com/sol3/papers.cfm?abstract_
id=1963451.
Haggard, S. and R. R. Kaufman (1995). The Political Economy of Democratic
Transitions. Princeton University Press.
Hambleton, R., H. Swaminathan, and H. Rogers (1991). Fundamentals of Item
Response Theory. London: SAGE.
Handloff, R. (1991). Cote d’Ivoire: A Country Study. Washington, DC: Library of
Congress.
Hartlyn, J. (1994). Crisis-ridden elections (again) in the Dominican Republic:
neopatrimonialism, presidentialism, and weak electoral oversight. Latin
American Politics and Society 36(4), 91–144.
Hassan, M. (2017). The strategic shuffle: ethnic geography, the internal security
apparatus, and elections in Kenya. American Journal of Political Science 61(2),
382–395.
Hassan, M. (2020). Regime Threats and State Solutions: Bureaucratic Loyalty and
Embeddedness in Kenya. Cambridge University Press.
Hayek, F. V. (1960). The Constitution of Liberty. University of Chicago Press.
Helmke, G. (2017). Institutions on the Edge: The Origins and Consequences of
Inter-Branch Crises in Latin America. Cambridge Studies in Comparative
Politics. Cambridge University Press.
Helmke, G. and S. Levitsky (2004). Informal institutions and comparative politics:
a research agenda. Perspectives on Politics 2(4), 725–740.
Herz, J. H. (1957). Rise and demise of the territorial state. World Politics 9(4),
473–493.
Hicken, A. (2011). Clientelism. Annual Review of Political Science 14(1),
289–310.
Hollyer, J. R. and L. Wantchekon (2015). Corruption and ideology in autocracies.
The Journal of Law, Economics, and Organization 31(3), 499–533.
Hughes, A. and R. May (1988). The politics of succession in black Africa. Third
World Quarterly 10(1), 1–22.
Huntington, S. (1968). Political Order in Changing Societies. New Haven, CT:
Yale University Press.
Hyde, S. D. and N. Marinov (2012). Which elections can be lost? Political Analysis
20(2), 191–210.
Jackson, R. and C. Rosberg (1982). Personal Rule in Black Africa. Berkeley:
University of California Press.
Joseph, R. (1978). Gaulist Africa: Cameroon under Ahidjo. Enugu: Fourth
Dimension Publishing.
Karimi, J. and P. Ochieng (1980). The Kenyatta Succession. Jos: Transafrica.
243
References
Knight, J. (1992). Institutions and Social Conflict. Cambridge University

Press.
Kofele-Kale, N. (1986). Ethnicity, regionalism, and political power: a post
mortem of Ahidjo’s Cameroon in M. Schatzberg and W. Zartman (eds.) The
Political Economy of Cameroon. New York: Praeger. 53–82.
Kokkonen, A. and A. Sundell (2014). Delivering stability: primogeniture and
autocratic survival in European monarchies 1000–1800. American Political
Science Review 108(2), 438–453.
Kokkonen, A. and A. Sundell (2020). Leader Succession and Civil War.
Comparative Political Studies 53(3–4): 434–468.
Konate, S. (2004). The politics of identity and violence in Cote d’Ivoire. West
Africa Review 5.
Konde, E. (2012). Cameroon’s political succession schism of 1983 revisited.
International Journal of Arts and Commerce 1(6), 25–37.
Konings, P. (2007). The neoliberalising African state and private capital
accumulation: the case of Cameroon in A. E. Fernandez Jilberto and
B. Hogenboom (eds.) Big Business and Economic Development: Conglomerates
and Economic Groups in Developing Countries and Transition Economies under
Globalisation. London: Routledge. 251–273.
Kurrild-Klitgaard, P. (2000). The constitutional economics of autocratic
succession. Public Choice 103(1), 63–84.
La Porta, R., F. Lopez-de Silanes, A. Shleifer, and R. Vishny (1998). Law and
finance. Journal of Political Economy 106(6), 1113–1155.
Lagace, C. B. and J. Gandhi (2015). Authoritarian Institutions. Abingdon:
Routledge.
Landes, D. (1998). The Wealth and Poverty of Nations. New York: Norton.
Langston, J. (2006). The changing party of the institutional revolution: electoral
competition and decentralized candidate selection. Party Politics 12(3),
395–413.
Lauth, H. (2015). The matrix of democracy: a three-dimensional approach to
measuring the quality of democracy and regime transformations. Würzburger
Arbeitspapiere zur Politikwissenschaft und Sozialforschung. Universität
Würzburg.
Lauth, H.-J. (2000). Informal institutions and democracy. Democratization 7(4),
21–50.
LeBas, A. (2013). From Protest to Parties: Party-building and Democratization in
Africa. Oxford University Press.
LeVine, V. (1964). The Cameroons: from Mandate to Independence. Oakland:
University of California Press.
Levitsky, S. (1998) Institutionalization and Peronism: the concept, the case and the
case for unpacking the concept. Party Politics 4(1), 77–92.
Levitsky, S. and G. Helmke (2006). Informal Institutions and Democracy:
Lessons from Latin America. Baltimore, MD: Johns Hopkins University Press.
Levitsky, S. and M. V. Murillo (2009). Variation in institutional strength. Annual
Review of Political Science 12(1), 115–133.
Levitsky, S. and L. A. Way (2010). Competitive Authoritarianism: Hybrid
Regimes After the Cold War. New York: Cambridge University Press.
Levitsky, S. and L. Way (2012). Beyond patronage: violent struggle, ruling party
cohesion, and authoritarian durability. Perspectives on Politics 20(4), 869–889.
244
References
Levitsky, S. and L. Way (2013). The durability of revolutionary regimes. Journal

of Democracy 24(3), 5–17.
Louw-Vaudran, L. (2016) Term-limit changes the biggest threat to democracy in
Africa? Institute for Security Studies. https://issafrica.org/amp/iss-today/term-
limit-changes-the-biggest-threat-to-democracy-in-africa.
Lucardi, A. (2017). Authoritarian party institutionalization. Working paper.
Mexico City: Instituto Tecnológico Autónomo de México.
Lust-Okar, E. (2006). Elections under authoritarianism: preliminary lessons from
Jordan. Democratization 13(3), 456–471.
Lyons, T. (2016). The importance of winning: victorious insurgent groups and
authoritarian politics. Comparative Politics 48(2), 167–184.
Magaloni, B. (2006). Voting for Autocracy: Hegemonic Party Survival and Its
Demise in Mexico. Vol. 296. Cambridge University Press.
Magaloni, B. (2008). Credible power sharing and the longevity of authoritarian
rule. Comparative Political Studies 41(4), 715–741.
Magaloni, B. and R. Kricheli (2010). Political order and one-party rule. Annual
Mahoney, J. and K. Thelen (2010). A Gradual Theory of Institutional Change.
Mainwaring, S. and T. R. Scully (Eds.) (1995). Building Democratic Institutions:
Party Systems in Latin America. Stanford University Press.
Mamdani, M. (1996). Citizen and Subject: Contemporary Africa and the Legacy
of Late Colonialism. Princeton University Press.
Marshall, M. G., T. R. Gurr and K. Jaggers (2016) Polity™ IV Project: Political
Regime Characteristics and Transitions, 1800-2015. Dataset Users’ Manual.
Vienna: Centre for Systemic Peace.
McCauley, J. (2017). The Logic of Ethnic and Religious Conflict in Africa.
McDonald, R. (1999). Test Theory: A Unified Treatment. New York: Taylor &
Francis.
McGovern, M. (2011). Making War in Cote d’Ivoire. University of Chicago Press.
Meisner, M. (1986). Mao’s China and After: A History of the People’s Republic.
Transformation of Modern China series. New York: Free Press.
Meng, A. (2019). Ruling parties in authoritarian regimes: rethinking institutional
strength. British Journal of Political Science. https://doi.org/10.1017/S000
7123419000115.
Meng, A. and J. Paine (2020). Rebel regimes in africa: military powersharing and
authoritarian durability. Working paper: University of Virginia and University
of Rochester.
Mershon, C. A. (1994). Expectations and informal rules in coalition formation.
Comparative Political Studies 27(1), 40–79.
Miller, A. (2010). How to read Xi Jinping’s 19th party congress political report.
China Leadership Monitor 53, 1–7.
Miller. A. (2017). Xi Jinping and the party’s “guiding ideology.” China
Leadership Monitor 54, 1–15.
Miller, M. (2020). The strategic origins of electoral authoritarianism. British
Journal of Political Science 50(1), 17–44.
Morrison, D. (1989). Black Africa: A Comparative Handbook. St. Paul: Paragon
House.
245
References
Mulford, D. C. (1967). Zambia: The Politics of Independence, 1957–1964.

London: Oxford University Press.
Myerson, R. B. (2008). The autocrat’s credibility problem and foundations of the
constitutional state. American Political Science Review 102(1), 125–139.
Nalepa, M. (2010). Skeletons in the Closet: Transitional Justice in Post-
Communist Europe. Cambridge University Press.
Negretto, G. L. (2013). Making Constitutions: Presidents, Parties, and
Institutional Choice in Latin America. Cambridge University Press.
Ngoh, V. J. (1987). Cameroon: 1884–1985 a Hundred Years of History. Limbe:
Navi-Group Publications.
North, D. C. (1990). Institutions, Institutional Change, and Economic
Performance. Cambridge University Press.
North, D. and B. Weingast (1989). Constitutions and commitment: the evolution
of institutions governing public choice in seventeenth-century England. The
Journal of Economic History 49(4), 803–832.
Opalo, K. O. (2019). Legislative Development in Africa: Politics and Postcolonial
Legacies. Cambridge University Press.
O’Donell, G. A. (1994). Delegative democracy. Journal of Democracy 5(1),
55–69.
Paine, J. (2019). The dictator’s power-sharing dilemma: countering dual outsider
threats. Working paper. New York: University of Rochester.
Pande, R. and C. Udry (2006). Institutions and development: a view from below.
Working papers 928. November. New Haven, CT: Yale University.
Panebianco, A. (1988). Political Parties: Organization and Power. Cambridge
University Press.
Peaslee, A. J. and D. Peaslee Xydis (1974) Constitutions of Nations. 4th ed. The
Hague: Martinus Nijhoff.
Pepinsky, T. (2009). Economic Crises and the Breakdown of Authoritarian
Regimes. Cambridge: Cambridge University Press.
Pepinsky, T. (2014). The institutional turn in comparative authoritarianism.
British Journal of Political Science 44(3),631–653.
Pierson, P. (2000). Increasing returns, path dependence, and the study of politics.
The American Political Science Review 94(2), 251–267.
Posner, D. and D. J. Young (2007). The institutionalization of political power in
Africa. Journal of Democracy 18(3), 126–140.
Powell, J. M. and C. L. Thyne (2011). Global instances of coups from 1950 to
2010: a new dataset. Journal of Peace Research 48(2), 249–259.
Powell, R. (1999). In the Shadow of Power: States and Strategies in International
Politics. Princeton University Press.
Powell, R. (2006). War as a commitment problem. International Organization 60
(1), 169–203.
Powell, R. (2012). Persistent fighting and shifting power. American Journal of
Political Science 56(3), 620–637.
Przeworski, A. (2015). Endogenous change of institutions in J. Gandhi and
R. Ruiz-Rufino (eds.) Routledge Handbook of Comparative Political
Institutions, Abingdon, Routledge, 43–55.
Przeworski, A., M. Alvarez, J. Cheibub, and F. Limongi (2000). Democracy and
Development: Political Institutions and Well-Being in the World, 1950–1990.
Cambridge Studies in the Theory of Democracy. Cambridge University Press.
246
References
Putnam, R. D. (1993). What makes democracy work? National Civic Re-view 82

(2), 101–107.
Rabinowitz, B. (2018). Coups, Rivals, and the Modern State. Cambridge
University Press.
Radu, M. and K. Sommerville. The People’s Republic of The Congo in C. Allen,
M. Radu, K. Sommerville, and J. Baxter (eds.) Benin, The Congo, Burkina Faso
(1989). London: Pinter. 145–236.
Reuter, O. J. (2017). The Origins of Dominant Parties: Building Authoritarian
Institutions in Post-Soviet Russia. Cambridge University Press.
Reuter, O. J. and J. Gandhi (2011). Economic performance and elite defection
from hegemonic parties. British Journal of Political Science 41(1), 83–110.
Riedl, R. B. (2014). Authoritarian Origins of Democratic Party Systems in Africa.
Roessler, P. (2011). The enemy within: personal rule, coups, and civil war in
Africa. World Politics 63(2), 300–346.
Roessler, P. (2016). Ethnic Politics and State Power in Africa: the Logic of the
Coup-Civil War Trap. Cambridge University Press.
Rubin, N. N. (1971). Cameroun: an African Federation. London: Pall Mall Press.
Schumpeter, J. (1947). Capitalism, Socialism, and Democracy. New York:
Harper.
Singh, N. (2014). Seizing Power: The Strategic Logic of Military Coups.
Baltimore, MD: JHU Press.
Slater, D. (2003). Iron cage in an iron fist: authoritarian institutions and the per-
sonalization of power in Malaysia. Comparative Politics 36(1), 81–101.
Slater, D. (2010). Ordering Power: Contentious Politics and Authoritarian
Leviathan in Southeast Asia. New York: Cambridge University Press.
Smith, B. (2005). Life of the party: the origins of regime breakdown and
persistence under single-party rule. World Politics 57(3), 421–451.
Stasavage, D. (2003). Public Debt and the Birth of the Democratic State: France
and Great Britain 1688–1789. Cambridge University Press.
Svolik, M. (2009). Power sharing and leadership dynamics in authoritarian
regimes. American Journal of Political Science 53(2), 477–494.
Svolik, M. (2012). The Politics of Authoritarian Rule. New York: Cambridge
University Press.
Sylla, L. and A. Goldhammer (1982). Succession of the charismatic leader: the
Gordian knot of African politics. Daedalus 111(2), 11–28.
Takougang, J. and M. Krieger (2000). African State and Society in the 1990s:
Cameroon’s Political Crossroads. Boulder, CO: Westview Press.
Takougang, J. and J. M. Mbaku (2004a). The Leadership Challenge in Africa:
Cameroon under Paul Biya. Africa World Press.
Takougang, J. and J. M. Mbaku (2004b). The Leadership Challenge in Africa:
Cameroon under Paul Biya. Trenton, NJ: Africa World Press.
Tamarkin, M. (1979). From Kenyatta to Moi: the anatomy of a peaceful transition
of power. Africa Today 26(3), 21–38.
Thelen, K. (1999). Historical institutionalism in comparative politics. Annual
Review of Political Science 2(1), 369–404.
Thelen, K. (2004). How Institutions Evolve: The Political Economy of Skills in
Germany, Britain, the United States, and Japan. Cambridge Studies in
Comparative Politics. Cambridge University Press.
247
References
Treisman, D. (2007). What have we learned about the causes of corruption from
ten years of cross-national empirical research? Annual Review of Political
Science 10(1), 211–244.
Tsai, L. L. (2007). Accountability Without Democracy: Solidary Groups and
Public Goods Provision in Rural China. Cambridge University Press.
Tullock, G. (1987). Autocracy. Berlin: Springer.
Van de Walle, N. (1993). The politics of non-reform in Cameroon in
T. M. Callaghy and J. Ravenhill (eds.) Hemmed In: Responses to Africa’s
Economic Decline. New York: Columbia University Press. 357–397.
Vogt, M., N.-C. Bormann, S. Raijegger, L.-E. Cederman, P. Hunziker, and
L. Girardin (2015). Integrating data on ethnicity, geography, and conflict: the
ethnic power relations data set family. Journal of Conflict Resolution 59(7),
1327–1342.
Welch, C. E. (1970). Soldier and State in Africa: A Comparative Analysis of
Military Intervention and Political Change. Evanston, IL: Northwestern
University Press.
Widner, J. A. (1992). The Rise of a Party-State in Kenya: From “Harambee!” to
“Nyayo!”. Berkeley: University of California.
World Bank (2016). Cameroon. https://data.worldbank.org/country/cameroon.
World Bank (2017). Côte d’Ivoire. https://data.worldbank.org/country/cote-
divoire.
Woronoff, J. (1972). West African Wager: Houphouet Versus Nkrumah.
Metuchen, NJ: Scarecrow Press.
Zolberg, A. (1966). Creating Political Order: The Party-States of West Africa.
Chicago: Rand McNally.
Zolberg, A. (1969). One Party Government in the Ivory Coast. Princeton
University Press.
Zolberg, A. (1971). Political development in the Ivory Coast since independence in
P. Foster and A. Zolberg (eds.) Ghana and the Ivory Coast: Perspectives on
Modernization. University of Chicago Press. 22–23.
248
Index
Ahidjo, Ahmadou, 64–65, 137 Bongo, Omar, 213

distribution of power under, 69 Botswana, 109, 219
ethnic representation under, 89–91 Botswana Rule, 102–103
executive constraints under, 74–76 British colonies, former, 69, 70–71
initial leadership strength, 67, Burkina Faso, 123
144
institutionalization under, 68 Calles, Plutarco Elías, 6, 117–118
leadership transitions, 67, 76–79, 107, Cambodia, 118
216–217 Cameroon, 64–65
repression under, 76 Ahidjo rise to power, 71–74
rise to power, 71–74 background and history, 70–71
tenure of, 68 colonial rule, 64, 70–71
Akoué tribe (Côte d’Ivoire), 82 Côte d’Ivoire, similarity of, 64
Amin Dada, Idi, 34–35 distribution of power, 66–67, 69
Angola elections, 78–79
distribution of power, 13 ethnic and regional representation in,
rebel-group regimes, 156 75–76, 89–91
ruling party, 103 executive constraints in, 74–76
succession procedures, 106, 136 GDP, xii
Aptidon, Hassan Gouled, 218 initial leadership strength, 67–68,
Assale, Charles, 75 144
assassination, 216 institutionalization in, 68
leadership transitions, 67, 76–79, 107,
Bamileke ethnic group (Cameroon), 73–74, 216–217
90 separatist conflict in, 70
Banda, Hastings, 13 succession procedures, 69, 77, 137
Baoulé ethnic group (Côte d’Ivoire), 81, 82, unification, 72–74
90 Cameroonian National Union (CNU) party,
Bassa ethnic group (Cameroon), 90 77
Bédié, Henri Konan, 65, 67, 87–89 Cape Verde, 136
Benin, 217 Cárdenas, Lázaro, 117
Biya, Paul, 65, 238 Central African Republic, 218
institutionalization and, 68 Chad, 109
leadership transitions, 67, 76–79, 107, Chama Cha Mapinduzi (CCM) party, 1,
216–217 103, 155
Bloc Democratique Camerouanis (BDC), Change the Constitution group (Kenya),
72 210–211
Bokassa, Jean-Bedel, 218 clientelism, 99
249
Index
Cold War coup d’états, leadership strength and, 135,

institutionalization after, 18, 135, 145–146, 151–154, 192–193
157–161, 184, 185, 232 aggregate statistical scores, 170
peaceful transitions after, 221 high ranking leaders statistical scores, 176
commitment problems in authoritarian post-1991 sample statistical scores, 174
regimes pre-1991 sample statistical scores, 172
distribution of power, 36–37, 38–39 variation within leaders statistical scores,
dynamic power shifts and, 38 175
executive constraints, 39 coup traps, 153
Communist Party of China (CCP), 5, 24, crown-prince problem, 209
118
Communist Party of Kampuchea (CPK), Dacko, David, 218
118 Deby, Idriss, 109
Compaore, Blaise, 123 democracies
constitutions coding as, 103
adherence to, 187–188 defined, 102
commonality in authoritarian regimes, institutional strength measurement in,
9–10 101
in Côte d’Ivoire, weakness of, 86–87 democratic backsliding, 238
importance in autocracies, 208–209 Democratic Republic of Congo, 6, 106
leadership control over, 40–41 democratization
in measurement model, 105–106 in sub-Saharan Africa, 14–15, 27–29,
persistence of, 138–139 158, 232
succession procedures and regime theories of, 237–238
stability, 209–212 Deng Xiaoping, 118
term limits and succession procedures in depersonalization, 96
peaceful transitions, 213–214, 219 Diori, Hamani, 40, 137
Conte, Lansana, 2 distribution of power
correlation in measurement model, 114–116 in Cameroon, 69, 74–76
Côte d’Ivoire, 40–41, 65–66 Cameroon and Côte d’Ivoire, compared,
background and history, 80–81 66–67
Cameroon, similarity of, 64 commitment problems in authoritarian
colonial rule, 81 regimes, 36–37, 38–39
distribution of power, 66–67, 69 in Côte d’Ivoire, 69
economic crisis, 87 ethnic representation in, 89–91
economic growth, 85–86 executive constraints and, 4–5, 35–36,
elections in, 87 96, 110, 135–136, 233
ethnic representation, 86, 89–91 as informal institutionalization,
forced labor, 83, 84–85 13–14
GDP, xii, 81 leadership strength and, 16–18
Houphouët rise to power, 81–85 in measurement model, 104, 106–111
independence, 85 Djibouti, 218
initial leadership strength, 67–68, Dominican Republic, 6–7
143–144 dos Santos, Jose Eduardo, 13, 107
institutionalization in, 68 dynamic power shifts, commitment
leadership transitions, 67, 87–89 problems and, 38
personalist rule, 79–80, 85–87
succession procedures, 69, 79–80, 87–88 elections
coup d’états, 7, 18, 36–37 in authoritarian regime definition, 102
Ahidjo attempt against Biya, 78 Botswana Rule, 102–103
in Côte d’Ivoire, 88 in Cameroon, 78–79
previous positions held in, 107–108 commonality in authoritarian regimes,
regime stability and, 192, 193–194 9–10
vulnerability to, 198–199, 200 in Côte d’Ivoire, 87
250
Index
in democratization, 237–238 Guinea

in sub-Saharan Africa, 28–29 executive constraints, in comparison with
England, commitment in, 39 Tanzania, 2–3
Equatorial Guinea, 40 instability in, 1–2
Ethiopia, 40 regime topology and institutionalization
ethnic representation score, 120
in Cameroon, 75–76, 89–91 as single-party regime, 24, 118
in Côte d’Ivoire, 86, 89–91 Guinea Bissau, 120
Europe, commitment in, 39
executive constraints hereditary succession, 206–207, 217–219
in autocratic institutions, 3–4, Houphouët-Boigny, Félix, 18, 40–41,
226–227 65–66
in Cameroon, 74–76 distribution of power under, 69
commitment problems in authoritarian ethnic representation under, 89–91
regimes, 39 initial leadership strength, 67–68,
in Côte d’Ivoire, lack of, 85–87 143–144
distribution of power and, 4–5, 35–36, institutionalization under, 68
96, 110, 135–136, 233 leadership transitions, 67, 87–89
in informal institutionalization, 98–100 personalist rule, 85–87
leadership strength and, 16–18, 35, rise to power, 81–85
134–136, 233–234 second in command position under, 107,
in measurement model, 104 109
reasons for adopting, 16, 34 succession procedures, 79–80
stability and, 19–20 tenure of, 68
in sub-Saharan Africa, 13, 14,
111–116 independence wars
Tanzania and Guinea, compared, 2–3 participation in, 151, 168
time period of adoption, 136–138 rebel-group regimes, 154–157, 178, 180,
V-Dem scores, comparison with, 182
124–126 inherited leadership, 135, 143–145,
Eyadema, Gnassingbe, 18 150–151, 166, 168
institutional spells, 139
family dynasties, 219 institutional strength, 95–96, 97–98, 101,
forced labor in Côte d’Ivoire, 83, 118
84–85 institutionalization
foreign aid, 160–161 after Cold War, 18, 135, 157–161, 184,
Franco, Francisco, 236 185, 232
French colonies, former autocratic institutions, compared,
Cameroon, history of, 70–71 97–98
Côte d’Ivoire, history of, 81 autonomous organizations, compared, 97
French citizenship in, 83 British and French colonies, compared, 69
regime institutionalization, strength of, causes of, 16–18, 134–161, 227–228
69 consequences of, 19–22
French Constituent Assembly, 82, defined, 3, 12, 93, 95–96, 104
83, 84 depersonalization in, 96
Front Populaire Ivoirien (FPI) party, 87 distribution of power in, 35–36, 96,
Fulani ethnic group (Cameroon), 73, 89 135–136
foreign aid and, 160–161
Gabon, 213 implications, 96–97, 237–238
Gambia, 94, 120 leadership strength and, 20, 68, 134–135,
Gbagbo, Laurent, 87 190–192, 214–216
Ghana, 94, 106, 138 leadership transitions, constraints on,
Gnassingbe, Eyadema, 114 12–13
Guelleh, Ismail Omar, 218 measurement model, 31, 229–231
251
Index
institutionalization (cont.) sample criteria, 102–104

natural resources and, 135, 159–160, 185 trends and patterns, 111–116
path dependent nature of, 17–18, 234, V-Dem scores, comparison with,
237 124–126, 230–231
peaceful transitions and, 20–22 institutionalization, regime stability and,
persistence of, 138–139 227–228, 234–236
personalist rule, compared, 5–9, 96–97 adherence to constitutional rules,
ruling parties, compared, 40, 228 187–188
at societal level, 12 conflicting studies, 189–190
when becomes effective, 100–101 coup vulnerability, 198–199
institutionalization, causes of endogenous institutionalization effects,
cross-sectional analysis, 147–157 estimating, 192–195
initial leadership strength, 135–136, initial leadership strength, 187, 235
233–234 leadership tenure, 195–198
leadership strength measurement, measurement problems with existing
140–141 studies, 190, 235
post-Cold War analysis, 157–161, 184, theoretical problems with existing
185 studies, 190–192, 235
institutionalization, leadership succession institutionalized dictatorships, 236
and, 236 institutions, autocratic
constitutions robustness check, 224 appearance of democratization in,
descriptive statistics, 218–219 237–238
dictators, effect of succession rules on, equilibrium in, 25–26, 98
207–209 executive constraints in, 3–4,
global sample, 223 226–227
hereditary succession, 217–218 inconsistency of data, 190
informal successor designation, 212–213, institutionalization, compared,
217 97–98
initial leadership strength, 214–216 regime stability and, 9–11, 37–38,
logit analysis, 219–221 189–190
peaceful transition identification, institutions, formal
216–217 defined, 98
post-Cold War analysis, 221 informal institutions, effect on, 115
regime stability, effect of succession rules in measurement model, 105–106
on, 209–212 institutions, informal
ruling parties and, 218 defined, 98
summary statistics, 222 distribution of power as, 13–14
term limits versus succession policies in executive constraints in, 98–100
constitutions, 213–214, 217 formal institutions, effect on, 115
institutionalization, measurement model in measurement model, 106–111
advantages of, 111, 230
assumptions in, 136–139 Jawara, Gawda, 94
cabinet appointments, 106–111 judicial constraints in V-Dem scores,
causality versus correlationality, 148 125–126
constitutions, 105–106
correlation of indicators, 114–116 Kabila, Laurent-Désiré, 6
cross-sectional scores, 147 Keita, Modibo, 118–119
executive constraints in, 104 Kenya
indicators used, 101, 111, 230 founding president, power of, 142
Polity scores, comparison with, 123–124, succession procedures, 105, 209–211,
230–231 214, 216
for regime stability, 193 successors, influence of, 142–143
regime typologies, comparison with, Kenya African Democratic Union (KADU),
119–123 143
252
Index
Kenyatta, Jomo, 21, 105, 142–143, institutionalization and, 236

209–211, 214, 216 in Kenya, 209–211, 216
Khama, Ian, 219 previous positions held for presidents,
Khama, Seretse, 109, 219 107–108
Khmer Rouge, 118 ruling parties and, 207
Kikuyu ethnic group (Kenya), 210–211 in sub-Saharan Africa, 207
Koinange, Mbiyu, 211 legislatures
commonality in authoritarian regimes, 9
lamidos (Fulani chiefs) in Cameroon, 76, constraints in V-Dem scores,
144 124–125
leadership strength Loi Cadre, 72, 83
Ahidjo rise to power and, 71–74
Cameroon and Côte d’Ivoire, compared, M’Ba, Leon, 213
67–68 Madagascar, 120
coup attempts and, 194, 198–199 Malawi, 13, 212
executive constraints and, 16–18, 35, Mali, 118–119, 120
134–136, 233–234 Mao Zedong, 5, 24, 118
Houphouët rise to power and, 81–85 Masire, Quett, 109
inherited versus mass movement mass movement leaders, 135, 143–145,
leaders, 135, 143–145, 150–151, 166, 150–151, 166, 168
168 Mauritius, 161
institutionalization and, 20, 68, 134–135, Maximato Period, 117
190–192, 214–216 Mbida, Andre-Marie, 72
leadership tenure and, 195–198 Mexico
peaceful transitions and, 206 commitment in, 39
regime stability and, 68, 187, 192–195, informal successor designation in,
235 99
leadership strength, measuring, 140–141, as outlier in data model, 29–30
192–193 personalist rule in, 117–118
coup leaders, 145–146, 151–154, 170, PRI regime in, 6
172, 174, 175, 176 regime stability, 189
cross-sectional analysis, 147–157 military commissions, 153–154
founding presidents, 141–143, 148–149, military officer rankings, coup leaders and,
163 145–146, 151–152, 175, 176
nationalist leaders, 143–145, 150–151, military regimes
166, 168 defined, 117
post-Cold War analysis, 157–161, 184, stability of, 189
185 minister of defense
rebel-group regimes, 154–157, 178, 180, appointment by rebel-group regimes,
182 156–157
leadership tenure consistency in, 110, 137
institutionalization and, 195–198 Mobutu Sese Seko, 6, 7, 18,
regime stability and, 192, 193–194 34, 236
summary statistics, 200 Moi, Daniel arap, 21, 105, 142–143,
leadership transitions 209–211, 214, 216
adherence to constitutional rules, monarchies, 206–207
187–188 Moviemento Popular de Libertacao de
in authoritarian regimes, 7–9 Angola (MPLA) party, 103
in Botswana, 109 Mozambique, 156
in Cameroon, 67, 76–79, 107, Mugabe, Robert, 217
216–217 Mungai, Njoroge, 211
constraints on, 12–13 Mutharika, Bingu wa, 212
in Côte d’Ivoire, 67, 87–89 Mutharika, Peter, 212
influence of successors, 142–143 Mwinyi, Ali Hassan, 1
253
Index
Namibia, 114 personalist rule in, 118

National Action Party (PAN), 6 as single-party regime, 118
National Assembly of Guinea, 2, 3 regime typology limitations, 24
natural resources, institutionalization and, Pereira, Aristides, 136
135, 159–160, 185 personalist rule
Neto, Agostinho, 136 in authoritarian regimes, 6–9,
Niger, 40, 120, 121–123, 137 236
Nigerian Progressive Party (PPN), 40, 120, changes over time, 117–118
121–123 conflation with single-party regimes,
Nkrumah, Kwame, 18, 82 118–119
Nujoma, Sam, 114 in Côte d’Ivoire, 79–80, 85–87
Nyerere, Julius, 1, 2, 155 defined, 117
institutionalization, compared, 5–9,
Obiang Nguema Mbasogo, Teodoro, 40 96–97
organizations, autonomous, 97 persistence of institutionalization, lack of,
Ouattara, Alassane, 88–89 138
stability of, 189
Parti democratique de Guinee (PDG) party, Pohamba, Hifikepunye, 114
2, 3, 24 Politburo Standing Committee, 100
Parti Democratique de la Cote d’Ivoire political institutionalization, 95
(PDCI) party, 83–84, 87, 143 Polity scores, 103–104, 123–124
Partido Revolucionario Institucional (PRI) presidents
party, 6, 117–118 founding, 134, 141–143, 148–149, 163,
party institutionalization, 95 192–193
peaceful transitions previous positions held, 107–108
defined, 7
previous positions held in, 107 Rassemblement Democratique Africain
peaceful transitions, institutionalization (RDA), 143
and, 20–22, 236 Rawlings, Jerry, 94
constitutions robustness check, rebel-group regimes, 154–157, 178, 180,
224 182
descriptive statistics, 218–219 regime typologies
dictators, effect of succession rules on, changes over time, 117–118
207–209 as composite indices, 118–119
global sample, 223 defined, 116–117
hereditary succession, 217–218 institutional strength variations
informal successor designation, 212–213, in, 118
217 institutionalization measurement,
initial leadership strength, 214–216 comparison with, 119–123
logit analysis, 219–221 subjectivity in, 119
peaceful transition identification, regimes
216–217 constitutions, control over, 40–41
post-Cold War analysis, 221 coup d’états, 7
regime stability, effect of succession rules defined, 94–95
on, 209–212 as institutionalized dictatorships,
ruling parties, 218 236
summary statistics, 222 institutional equilibria, 25–26, 98
term limits versus succession policies in origins of, effect on institutionalization,
constitutions, 213–214, 217 26–27
People’s Republic of China personalist rule in, 6–9, 236
as hierarchical system, 5 presence of autocratic institutions, effect
informal successor designation, of, 9–11
100 rules, procedures, norms governing, 5–6
as outlier in data model, 29–30 typologies, 23–24
254
Index
regimes, commitment problems in creation of, 40

distribution of power, 36–37, 38–39 death of founder, effect of, 10–11
dynamic power shifts, 38 in Côte d’Ivoire, weakness of, 87
executive constraints, 39 institutionalization, compared, 40, 228
regimes, leadership transitions in, 7–9, 236 leadership transitions and, 207, 218,
constitutions robustness check, 224 220
descriptive statistics, 218–219 stability and, 19
dictators, effect of succession rules on,
207–209 Sassou Nguesso, Denis, 41, 137
global sample, 223 second-in-command positions
hereditary succession, 217–218 consistency in, 109–110
informal successor designation, 212–213, creation of, 107
217 peaceful transitions and, 212–213, 217,
initial leadership strength, 214–216 219
logit analysis, 219–221 Sekou Toure, Ahmed, 1, 2, 3, 73
peaceful transition identification, 216–217 Senghor, Léopold Sédar, 65
post-Cold War analysis, 221 separatist conflict in Cameroon, 70
regime stability, effect of succession rules Soviet Union, 118, 189
on, 209–212 Statue of the Cameroun, 73
ruling parties, 218 sub-Saharan Africa
summary statistics, 222 democratization in, 14–15, 27–29, 158,
term limits versus succession policies in 232
constitutions, 213–214, 217 executive constraints, 13, 14,
regimes, single-party 111–116
conflation with personalist rule, 118–119 leadership transitions in, 207
defined, 117 patterns of institutionalization in, 18,
institutional strength variations in, 118 93–94, 111–116, 231–232
stability of, 189 postcolonial nature of, 30
regimes, stability of, 227–228, 234–236 reasons for studying, 27–30
adherence to constitutional rules, regime stability in, 29–30
187–188 succession procedures
conflicting studies, 189–190 adherence to, 187–188
coup vulnerability, 198–199 in Cameroon, 69, 77, 137
endogenous institutionalization effects, in Côte d’Ivoire, 69, 79–80, 87–88
estimating, 192–195 effect on dictators, 207–209
initial leadership strength, 68, 187, 235 in Guinea, 3
leadership tenure, 195–198 hereditary succession, 206–207,
measurement problems with existing 217–219
studies, 190, 235 informal successor designation, 212–213,
in sub-Saharan Africa, 29–30 217
succession procedures, 209–212 in Kenya, 209–211
theoretical problems with existing in measurement model, 104, 105–106
studies, 190–192, 235 in Mexico, 99
regional representation in Cameroon, peaceful transitions and, 20–22,
75–76 205–206, 217, 219
repression under Ahidjo in Cameroon, 76 persistence of, 138–139
Republic of Congo, 41, 137 as ranking of elites, 66–67
rotation rates regime stability and, 209–212
in minister of defense positions, 110 in Tanzania, 2
in second-in-command positions, term limits, compared, 213–214
109–110 time period of adoption, 136–138
ruling parties Sudanese Union, 120
Botswana Rule, 102–103 Swaziland, 40
commonality in authoritarian regimes, 9 Syndicat Agricole Africain (SAA), 82
255
Index
Tabi, Emmanuel Egbe, 75 Togo, 114, 161

Tanganyika African National Union traditional chiefs in Cameroon, 76
(TANU) party, 1, 155 transparency, lack of, 37
Tanzania Trujillo, Rafael, 6–7
executive constraints, in comparison with
Guinea, 2–3 Union Camerounaise (UC), 72
persistence of institutionalization, 138 Union des Populations du Cameroun (UPC),
rebel-group regime weakness, 155 71, 72
ruling party, 103 Union Soudanaise - Rassemblement
stability in, 1, 94–95 Democratique Africain (US-RDA)
term limits, 96 party, 118–119
adherence to, 187–188, 238 United National Independence Party
in Guinea, 3 (UNIP), 155
in measurement model, 106 United States, foreign aid from, 160–161
in Mexico, 6
peaceful transitions and, 206, 217, V-Dem scores, 124–126
219–220
persistence of, 138–139 XCONST variable (Polity scores), 123–124
succession procedures, compared, 21, Xi Jinping, 238
213–214
in Tanzania, 2 Zambia, 155
time period of adoption, 136–138 Zimbabwe, 217
256

Cambridge Core Constraining Dictatorship 9nov2022

Uploaded by

Copyright:

Available Formats

Cambridge Core Constraining Dictatorship 9nov2022

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Cambridge Core Constraining Dictatorship 9nov2022

Uploaded by

Copyright:

Available Formats

C O N S T R A IN I N G D I C T A T O R S H I P

How do some dictatorships become institutionalized ruled-based systems,

Anne Meng is Assistant Professor in the Department of Politics, University of

Published online by Cambridge University Press

Other books in the series

Alberto Alesina and Howard Rosenthal, Partisan Politics, Divided Government

Published online by Cambridge University Press

Published online by Cambridge University Press

Published online by Cambridge University Press

Published online by Cambridge University Press

Published online by Cambridge University Press

Cambridge University Press is part of the University of Cambridge.

Published online by Cambridge University Press

How do some dictatorships become institutionalized ruled-based systems,

Anne Meng is Assistant Professor in the Department of Politics, University of

Published online by Cambridge University Press

Other books in the series

Alberto Alesina and Howard Rosenthal, Partisan Politics, Divided Government

Published online by Cambridge University Press

Published online by Cambridge University Press

Published online by Cambridge University Press

Published online by Cambridge University Press

Published online by Cambridge University Press

Cambridge University Press is part of the University of Cambridge.

Published online by Cambridge University Press

How do some dictatorships become institutionalized ruled-based systems,

Anne Meng is Assistant Professor in the Department of Politics, University of

Published online by Cambridge University Press

Other books in the series

Alberto Alesina and Howard Rosenthal, Partisan Politics, Divided Government

Published online by Cambridge University Press

Published online by Cambridge University Press

Published online by Cambridge University Press

Published online by Cambridge University Press

Published online by Cambridge University Press

Cambridge University Press is part of the University of Cambridge.

Published online by Cambridge University Press

How do some dictatorships become institutionalized ruled-based systems,

Anne Meng is Assistant Professor in the Department of Politics, University of

Published online by Cambridge University Press

Other books in the series

Alberto Alesina and Howard Rosenthal, Partisan Politics, Divided Government

Published online by Cambridge University Press

Published online by Cambridge University Press

Published online by Cambridge University Press

Published online by Cambridge University Press

Published online by Cambridge University Press

Cambridge University Press is part of the University of Cambridge.

Published online by Cambridge University Press

List of Figures page xii

3.3 Cô te D’Ivoire Under Houphouët: Low Levels of Regime

7.3 Designating Informal Successors 212

1.1 Number of failed and successful coups in autocracies page 8

5.3 Coup leaders versus noncoup leaders 152

Appendix 3.1 Summary statistics page 92

Appendix 5.13 Differences-in-differences test 185

friendship: Andrew Bertoli, Tara Buss, Christopher Chambers-Ju,

African politics community, including Jaimie Bleck, Ruth Carlitz, Mai

BDC Bloc Democratique Camerouanis