Presidents, Monarchs, and Prime Ministers: Executive Power Sharing in the World (Palgrave Studies in Presidential Politics) 3031039599, 9783031039591

Citation preview

PALGRAVE STUDIES IN PRESIDENTIAL POLITICS SERIES EDITOR: GIANLUCA PASSARELLI

Presidents, Monarchs, and Prime Ministers Executive Power Sharing in the World

Carsten Anckar

Palgrave Studies in Presidential Politics

Series Editor Gianluca Passarelli, Sapienza University of Rome, Rome, Italy

Palgrave Studies in Presidential Politics publishes books on all aspects of presidential politics. We welcome proposals for monographs, edited volumes and Pivots on topics such as: • • • • • •

Contemporary presidencies and presidential powers Presidential elections and presidential party politics Presidential relations with the legislature The media and presidential communication The administrative presidency and presidential advisers The history of presidential offices and presidential biographies

The series focuses on presidents throughout the world, including both directly elected and indirectly elected presidents, both single-country and comparative studies of presidential politics. It also includes volumes on conceptual or theoretical aspects, such as how to measure presidential power. Moreover, the series considers book projects on the reform of presidential politics, e.g. the reform of presidential elections. For further information on the series and to submit a proposal for consideration, please get in touch with: • Commissioning Editor Ambra Finotello ambra.finotello@palgrave. com • Series Editor Gianluca Passarelli [email protected]

More information about this series at https://link.springer.com/bookseries/15600

Carsten Anckar

Presidents, Monarchs, and Prime Ministers Executive Power Sharing in the World

Carsten Anckar Department of Political Science Åbo Akademi University Turku, Finland

Palgrave Studies in Presidential Politics ISBN 978-3-031-03959-1 ISBN 978-3-031-03960-7 (eBook) https://doi.org/10.1007/978-3-031-03960-7 © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Cover image: © Tuomas A. Lehtinen/Moment/Getty Image This Palgrave Macmillan imprint is published by the registered company Springer Nature Switzerland AG The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland

For Oxana, Nadine, Beatrice, and Richard

Preface

In 1978, Urho Kekkonen was elected president of Finland, receiving 81.7% of the popular vote. His candidature was supported by nine parties, representing ideologies from the farthest left to the right. In the first and only round of voting in the Electoral College, Kekkonen’s vote share rose to 86.3%. Such constellations and figures are normal in authoritarian states but hardly flattering for a country that had been a stable democracy for 60 years, under a constitution, an important ambition of which was to divide powers between the president and the prime minister. I was eight years old when the election took place, but I do remember watching the election night coverage together with my father, a professor of political science and a staunch opponent of Kekkonen. Three years later, Kekkonen, marked by dementia, resigned, leaving the presidency in the hands of Mauno Koivisto. This marked the beginning of the dismantling of the powers of the Finnish president, a process that has been slow and continues to date. I am convinced that the experience of growing up in a country where executive power sharing has been a central and topical feature of the political life is the most important explanation for why this book has been written. Ever since Kekkonen left office, Finnish prime ministers and presidents have regularly clashed on the issue how powers should be divided between them. Studying political science in such a milieu is likely to direct your research interest to questions of executive power sharing. Indeed, I have noticed that I share the experience of having grown up in a system

vii

viii

PREFACE

with a divided executive with many of my most active colleagues in this particular field of research. In workshops and panels on dual executives, the representation of Irish, Portuguese, French, and, especially, Central European political scientists, tends to be quite impressive. Most research efforts in the area are explicitly concerned with the semi-presidentialist form of government. However, as the concept of dual executive is much broader than that of semi-presidentialism the explicit aim of the present study is to include a wide variety of democratic systems where powers are shared between the head of state and the head of government. Constellations where a monarch shares powers with a prime minister responsible to the legislature are of particular interest, as these power sharing arrangements have been largely disregarded in the literature on dual executives. Furthermore, in this book, the concept of ‘sharing’ is stretched to its limits. Essentially, the question of the division of powers becomes relevant under all circumstances where the head of state is separated from the head of government. In some cases, like Sweden, the role of head of state is completely ceremonial, in other countries, like Sri Lanka and Peru, the prime minister is clearly subordinate to the president. Although constitutions set limits to the exercise of powers, they do not always reflect the political reality. This is especially the case in democracies with a monarch as head of state, where constitutional provisions often provide monarchs with wide reaching powers in the executive sphere. In other cases, constitutions are vaguely formulated, thus giving ambitious and unscrupulous actors a possibility to stretch their provisions to their limits. Indeed, the previous Finnish constitution in force 1919–2000, described by my father as a ‘smorgasbord’ of prerogatives, largely explain why President Kekkonen could raise to the position as the completely dominating political figure in Finnish politics. In contrast to many other works in the field, the present book measures the powers of the head of state with respect to political practice rather than constitutional provisions. A little more than a decade ago, such an attempt would have been faced with insurmountable obstacles. However, with the introduction of the V-dem dataset, it has become much easier to make cross-country comparisons on the basis of behavioural aspects. Although the book is single authored, I owe my gratitude to a lot of persons who have helped me complete the manuscript. First, I wish to thank the series editor Gianluca Passarella for the encouragement to write a book on dual executives and for accepting the manuscript in the

PREFACE

ix

series Palgrave Studies in Presidential Politics. Like so many times before, Professor Thomas Denk has provided many extremely useful comments on the methodologies applied. In recent years, I have presented various papers in panels organized by the ECPR standing group on Presidential Politics. I thank various participants for many valuable comments and suggestions. I also thank my colleagues at the Department of Political Science at Åbo Akademi University for providing me with a highly stimulating research environment. Last but certainly not least, I thank my wife Oxana and my children Nadine, Beatrice, and Richard for their support, patience, and understanding which made this book possible. Turku, Finland January 2022

Carsten Anckar

Contents

1

Introduction 1.1 The Research Task 1.2 Mapping the Dual Executives 1850–2019 1.3 How Dual Executives Emerge 1.4 How Dual Executives Transform 1.5 How Dual Executives End 1.6 Outline of the Book 1.7 Data and Methods 1.8 Contribution to the Field of Research References

1 1 4 6 9 10 12 13 15 16

2

Classifying Political Regimes 2.1 Introduction 2.2 Classifying Authoritarian Regimes 2.3 Classifying Democratic Regimes 2.4 Conclusion: The Difficult Art of Classifying Regimes References

21 21 22 32 44 46

3

The 3.1 3.2 3.3 3.4 3.5

51 51 53 56 57 61

Research Population Separating Democracies from Autocracies Separating Dual Executives from Single Executives The Selection of the Head of State The Powers of the Head of State Dual Executives 1850–2019

xi

xii

CONTENTS

Appendix: Measuring the Powers of the Head of State Based on the V-dem Dataset References 4

65 69 71 71 73

How Dual Executives Emerge 4.1 Introduction 4.2 Previous Regime Form and Regime Choice 4.3 Previous Regime Form and Introduction of Dual Executives—Empirical Patterns 4.4 Diffusion 4.5 Diffusion and Introduction of Dual Executives—Empirical Patterns 4.6 The Comprehensive Picture 4.7 Discussion References

98 103 113 116

5

Patterns of Transformation in Dual Executives 5.1 Theoretical Patterns of Transformation 5.2 Empirical Patterns of Transformations 5.3 Multivariate Patterns 5.4 Discussion Sources References

121 121 126 132 149 159 159

6

How Dual Executives End 6.1 Introduction 6.2 When Do We Expect Dual Executives to Die? 6.3 Selection Mode, Powers, and Regime Breakdown—Empirical Patterns 6.4 Multivariate Patterns 6.5 Discussion Sources References

163 163 164

Conclusions 7.1 Summary of Findings 7.2 Lessons, Limitations, and Prospects for Future Research References

201 201 209 215

7

Index

81 93

167 174 188 194 194

217

List of Figures

Fig. 5.1

Fig. 5.2

Fig. 5.3

Fig. 5.4

Fig. 6.1

Fig. 6.2

Fig. 6.3

Predicted probability of loss of powers at t according to amount of powers at t − 1 in dual executives with non-popularly elected presidents Predicted probability of loss of powers at t according to amount of powers at t − 1 in dual executives with popularly elected presidents Predicted probability of loss of powers at t according to amount of powers at t − 1 in dual executives with a hereditary monarch Predicted probability of monarch losing powers at t according to amount of powers at t − 1 in democratic monarchies with a population of at least 500,000 individuals Predicted probability of regime breakdown according to amount of power of president in dual executives with non-popularly elected presidents Predicted probability of regime breakdown according to amount of power of president in dual executives with popularly elected presidents Predicted probability of regime breakdown according to amount of power of monarch in dual executives with a monarch as head of state

152

152

153

155

187

188

189

xiii

xiv

LIST OF FIGURES

Fig. 6.4

Predicted probability of regime breakdown according to amount of power of monarch in dual executives with a monarch as head of state and a population of at least 500,000 individuals

191

List of Tables

Table 3.1 Table 3.2 Table 4.1 Table 4.2 Table 4.3 Table 4.4 Table 4.5 Table 4.6 Table 4.7 Table 4.8 Table 4.9 Table 4.10

Dual executives according to mode of selection of head of state in democracies 1850–2019 Executive power sharing patterns in the world 1850–2019. V-dem figures Previous regime form and type of executive at time of independence or at transition to democracy Previous regime type and powers of president in dual executives with a non-popularly elected president Previous regime type and powers of president in dual executives with a popularly elected president Previous regime type and powers of monarch in dual executives with a hereditary monarch as head of state Colonial heritage and democratic regime form Colonial heritage and powers of the president in dual executives with a non-popularly elected president Colonial heritage and powers of the president in dual executives with a popularly elected president Colonial heritage and powers of the monarch in dual executives with a monarch as head of state Similar regime density and choice of dual executives. Logistic regression analyses Explaining choice of a dual executive with a non-popularly elected president as head of state 1850–2019. Logistic regression analyses

62 64 82 83 84 85 99 101 102 103 104

110

xv

xvi

LIST OF TABLES

Table 4.11

Table 4.12

Table 5.1 Table 5.2 Table 5.3

Table 5.4 Table 5.5 Table 5.6 Table 5.7 Table 5.8

Table 5.9

Table 5.10

Table 5.11

Table 5.12

Table 6.1

Explaining choice of a dual executive with a popularly elected president as head of state 1850–2019. Logistic regression analyses Explaining choice of a dual executive with a hereditary monarch as head of state 1850–2019. Logistic regression analyses Expected patterns of transformations in democratic countries with dual executives Mode of selection and changes in powers of heads of state 1850–2019 Transition probabilities of powers of the head of state in democratic countries with presidents not elected by popular vote Transition probabilities of powers of the head of state in democratic countries with popularly elected presidents Transition probabilities of powers of the head of state in democratic countries with a monarch as head of state Explaining increasing powers of the head of state 1850–2019. Logistic regression analyses Explaining decreasing powers of the head of state 1850–2019. Logistic regression analyses Explaining increasing powers of the head of state 1850–2019 in democratic republics where the president is not popularly elected. Logistic regression analyses Explaining decreasing powers of the head of state 1850–2019 in democratic republics where the president is not popularly elected. Logistic regression analyses Explaining increasing powers of the head of state 1850–2019 in democratic republics where the president is popularly elected. Logistic regression analyses Explaining decreasing powers of the head of state 1850–2019 in democratic republics where the president is popularly elected. Logistic regression analyses Explaining increasing and decreasing powers of the head of state 1850–2019 in democratic republics with a monarch as head of state. Logistic regression analyses Powers of the head of state and transformation to other regime types from dual executives with presidents not elected by popular vote

111

112 125 127

129 130 132 142 144

145

146

147

148

150

168

LIST OF TABLES

Table 6.2

Table 6.3

Table 6.4 Table 6.5 Table 6.6 Table 6.7

Powers of the head of state and transformation to other regime types from dual executives with popularly elected presidents Powers of the head of state and transformation to other regime types from dual executives with a monarch as head of state Explaining end of dual executives 1850–2019 Explaining end of dual executives in countries with non-popularly elected presidents 1850–2019 Explaining end of dual executives in countries with popularly elected presidents 1850–2019 Explaining end of dual executives in countries with a monarch as head of state 1850–2019

xvii

170

171 181 183 185 186

CHAPTER 1

Introduction

1.1

The Research Task

About thirty years ago, four seminal works, Presidents and Assemblies (Shugart & Carey, 1992), Parliamentary versus Presidential Government (Lijphart, 1992), Comparative Constitutional Engineering (Sartori, 1994), and The Failure of Presidential Democracy (Linz & Valenzuela, 1994), were published. The books emerged quite early during the ‘new institutionalism’ wave in political science. They were particularly important for the purpose of the present study as they directed a great deal of attention to democratic regimes that were neither presidential nor parliamentary. The concept of semi-presidentialism had been introduced by Maurice Duverger (1970, 277, 1978, 1980) some decades earlier, but had so far received very little scholarly attention in Anglo-Saxon political science literature (e.g. Åberg & Sedelius, 2020).1 Outside the English-speaking world, the semi-presidential form of government (albeit not always referred to by that term) had, by that time, been a more popular topic for decades (Anckar, 1984; Chapsal, 1981; Colliard, 1978; Debré, 1958; Duhamel, 1988; Finger, 1923; Haungs, 1968; Lindman, 1937, 1974; Pereira, 1984; Preuss, 1926; Sousa, 1984; Tarkiainen, 1938). The Anglo-Saxon lack of interest in the subject is most probably explained by the fact that the semi-presidentialist regime form was commonly not in use in English-speaking countries. The only possible © The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 C. Anckar, Presidents, Monarchs, and Prime Ministers, Palgrave Studies in Presidential Politics, https://doi.org/10.1007/978-3-031-03960-7_1

1

2

C. ANCKAR

exception was Ireland, which in 1937 introduced a constitution where executive powers were divided by a popularly elected president and a prime minister responsible to parliament. However, in comparison with other countries with the same arrangement, the powers of the Irish presidents were weak; as noted by Elgie (2012, 511), ‘from the late 1930s to the late 1980s the Irish President had the fewest constitutional powers of all countries with a directly elected president…’. The increased scholarly attention on semi-presidentialism in the 1990s was not only a consequence of the publication of the pioneering works cited above, but also by the fact that the dual executive model became quite popular in democratizing countries in Eastern Europe in particular but also in a number of newly democratized states in Africa and Asia (Elgie, 2016, 54). At the same time, authors struggled with the definitional criteria of the concept. Duverger (1980, 166) had defined semi-presidentialism as the situation in which ‘(1) the president of the republic is elected by universal suffrage; (2) he possesses quite considerable powers; (3) he has opposite him, however, a prime minister and ministers who possess executive and governmental power and can stay in office only if the parliament does not show its opposition to them’. The problem with the definition was that it was vague with respect to how much powers the president should be in possession of. Naturally, ‘quite considerable’ powers could mean more or less anything. As a consequence, authors found it difficult to agree on the population of semi-presidential countries. Another issue of concern was whether one should measure powers with respect to constitutional provisions or political practice. Although Duverger relied on constitutional provisions when identifying semi-presidential systems, he also dwelled extensively on practical powers, noting, for instance, that ‘although the constitution plays a certain part in the application of presidential powers, this role remains secondary compared to that of other parameters’ (Duverger, 1980, 179). One ‘solution’ to the problem was to disregard the concept of semipresidentialism altogether. In their highly influential work, Shugart and Carey (1992) effectively split semi-presidential systems into the categories premier-presidentialism and president-parliamentarism. The crucial difference between the categories was that in systems situated in the latter category, the president could dismiss the government whereas he or she was unable to do so in premier-presidentialist countries. The book was highly influential and their subcategorization continues to be used in contemporary works on dual executives. However, their work

1

INTRODUCTION

3

was pioneering also in another respect. Reasoning from a principal-agent theory, the authors made the argument that the popular election of the head of state was necessary for inclusion among the categories of premier-presidentialism and president-parliamentarism. A few years later, Robert Elgie entered the scene, proposing a definition of semi-presidentialism that effectively built on Shugart and Carey, combining premier-presidential and president-parliamentary systems under the umbrella term semi-presidentialism. According to Elgie (2011, 3), semi-presidentialism is ‘the situation where there is both a directly elected fixed-term president and a prime minister and cabinet who are collectively responsible to the legislature’. Elgie’s definition completely disregards the power dimension and focuses exclusively on the mode of selection of the head of state. In recent decades, Elgie’s definition of semi-presidentialism has become the most popular one. The main advantage is that authors find it easy to agree on the population of semi-presidential countries. However, disregarding the power dimension also has disadvantages. Most importantly, it blurs the distinction between parliamentarism and semi-presidentialism. A country with a non-popularly elected, but very powerful president is automatically classified as parliamentary whereas a country with a popularly elected president, holding ceremonial powers only, is regarded as semi-presidential. Although research on semi-presidentialism constitutes a natural point of departure for the present work, it is important to point out that irrespective of how semi-presidentialism is defined, the concept of dual executives is wider than that of semi-presidentialism. This was exemplified elegantly in Alan Siaroff’s (2003) work, where he made use of several dispositional categories when classifying the democratic republics of the world. Siaroff’s (2003) article is particularly important as it demonstrated that many hybrid regime forms remain unexplored and cannot be covered by the concept of semi-presidentialism, whether broadly or narrowly defined. The present research effort follows in the footsteps of Siaroff’s (2003) work. It differs from most comparative studies of dual executives in three important respects: (1) it moves beyond the category of semipresidentialism, accounting for a larger set of executive power sharing systems, (2) it is extensive in time and space as it covers all democratic dual executives in the world during the time period 1850–2019, and (3) its focus is on a wide variety of actual powers of the head of state rather than on constitutional provisions. The book sets out to fulfil four purposes.

4

C. ANCKAR

The first one is to classify dual executives in terms of how the head of state is selected and with regard to the powers he or she possesses. The second ambition is to explain why different types of dual executives emerge. The third objective is to account for transformation patterns along the dimensions selection mode and powers of the head of state, whereas the fourth, and final, purpose is to explain why dual executives die.

1.2

Mapping the Dual Executives 1850–2019

In democracies we make a distinction between countries with a presidential form of government and countries with a parliamentary form of government. In addition, a number of countries combine features of presidentialism and parliamentarism. This category of hybrid systems is rather heterogeneous but a common denominator for most of these systems is that the executive power is divided between the head of state and the prime minister. In the present contribution, the ambition is to dwell more deeply into the category of these hybrid systems. The first step of the monograph is to categorize the different variants of executive power sharing in democratic systems during the time period 1850–2019. As a point of departure, I operate with a broad conception of powersharing, including all democratic systems in which the head of state is different than the head of government. This population is identified in a stepwise process. First, all non-democratic countries are excluded from the population. In the second step, all democracies where the position of head of state coincides with the position as head of government are excluded, since, in these cases, the concept of power sharing is irrelevant. The elimination of these systems effectively means that all countries with a presidential form of government are excluded. However, the same thing applies for the few parliamentary systems, notably South Africa and Botswana, where the president acts as both head of state and head of government, at the same time as he or she is subject to a parliamentary vote of no confidence. In principle, all remaining systems are of interest for the study. It is evident that by combining a number of important institutional features we can come up with many different combinations of power sharing (e.g. Siaroff, 2003). It is, however, equally evident that many of these combinations have few, if any, empirical instances. Instead, it is important to focus on those aspects that are expected to be of importance for the distribution of power between the prime minister and the head of state.

1

INTRODUCTION

5

A minimal requirement for a dual executive is that the actors who share powers have an independent room for manoeuvre. This is most clearly the case in dual executives where the president is popularly elected. Here the independence of the president vis-à-vis the prime minister and vice versa is primarily manifested by the fact that the two actors base their legitimacy on different institutions (the president on the people) and the prime minister on the parliament (Shugart & Carey, 1992; Shugart, 2005). However, although the popular election of the president is likely to enhance his or her level of independence in relation to the prime minister, it is not a necessary criterion for inclusion in the population. As pointed out by Tavits (2009, 34) presidents elected by parliaments do not necessarily act as agents of parliament. The central aspects are that the legislature cannot remove the president, who, in his/her turn, has to take into account the partisan constellation in parliament when appointing (and dismissing) the prime minister. Thus, the independence of the actors is achieved when the prime minister is accountable to the legislature at the same time as the president cannot be removed from office by parliament (except under special circumstances, for instance as a result of an impeachment process). An even milder criterion of independence of the actors occurs in those rare cases where both the president and the prime minister can be ousted of office by a parliamentary vote of no confidence, but the fate of the president is not tied to the fate of the prime minister and vice versa. Finally, not only republics are of interest for the purpose of the study. There are also a substantial number of democratic countries with a hereditary monarch as head of state and a prime minister and government responsible to the legislature. In these cases, the head of government bases his or her legitimacy on the people and the monarch on a god given right to rule. Countries that meet any of these criteria are included in the population of dual executives. However, the extent to which actual power is distributed between the head of state and the head of government varies a lot. A large part of countries in the population consists of democracies with a monarch as head of state and a prime minister as head of government. In many of these countries, the two actors do not really share political power. All power is concentrated into the hands of the prime minister, whereas the monarch is in possession of ceremonial powers only. It is important to point out that similar arrangements occur in republics as well; there are many instances where the president has very

6

C. ANCKAR

few prerogatives, regardless of whether he or she is popularly elected or not. Therefore, as a next step, the power dimension is introduced. The purpose is thereby to assess to what extent the head of state can impinge on the position of the prime minister. Here, the powers of the head of state are measured with regard to behavioural traits rather than constitutional provisions. The main source for measuring the powers has been the V-dem-dataset (Coppedge et al., 2020), which contains a number of variables which measure the powers of the head of state based explicitly on political practice. The dataset currently covers the years 1789–2019, which means that it is possible to conduct global cross-country comparisons over a very long period of time. Altogether the V-dem dataset contains nine prerogatives that are of importance for the purpose of the present study. As some of the questions are overlapping, not all of them can be used for measuring the degree of power of the head of state. In the present volume, I make use of seven questions from the V-dem dataset. The first three questions refer to the executive sphere, more specifically to the ability of the head of state to appoint and dismiss ministers. In addition, there are two questions concerning legislative powers, namely the ability of the head of state to propose legislation and veto legislation. There is also one question which concerns the legislative sphere directly and the executive sphere indirectly, namely the power to dissolve the legislature. Finally, one question in the dataset concerns the extent to which the head of state has powers over domestic policy decisions. Some of the questions are dichotomous and some ordinal. The ordinal ones are converted to take values from 0 to 1. By adding the values, an index of powers ranging from 0 to 7 is obtained.

1.3

How Dual Executives Emerge

The second aim of the study is to explain why different variants of executive power sharing occur. The literature on the determinants of regime choice is rather scarce, and this is particularly the case regarding hybrid forms, where authors have almost exclusively focused on the semi-presidential form of government. In most of these studies, semipresidentialism has been conceived of in a broad sense, i.e. according to Elgie’s definition cited above. It should nevertheless be acknowledged, that theories explaining semi-presidentialism are not very well developed. The spread of semi-presidentialism occurred in two phases: one extremely

1

INTRODUCTION

7

slow phase, which began in 1919 and another, fast one, which began in 1990 and continues to date. From the 1920s to the 1970s, semipresidential systems were rare, but at the same time in use in very different contexts; from various parts in Europe (e.g. Finland, Weimar Germany, France), in the Caribbean (Cuba 1940–1952), and on the Southern Cone (Brazil 1961–1962). However, the countries that adopted semipresidentialism at this early stage in history were rarely or ever influenced by other semi-presidential countries (Elgie, 2011, 7). The same thing does not apply to countries that chose semipresidentialism during the later phase. Here, diffusion patterns are clearly at work. First, as the former colonies of France and Portugal became independent and eventually democratized they adopted forms of government that resembled those of their former mother countries. Second, as the former Eastern European countries regained their freedom after the fall of the Berlin Wall, they tended to opt for semi-presidentialism. The reason for this, it has been claimed, is that they wanted to retain the dual power structure, which characterized the communist systems, namely a party hierarchy with the general secretary of the communist party on top alongside a government led by a prime minister, formally subject to a parliamentary vote of no confidence (Blondel, 1992; Wu, 2011, 24–25). Based on the aforesaid I expect diffusion to be of importance for the emergence of dual executives. However, based on the patterns observed with regard to the early variants of semi-presidentialism, other explanations must be accounted for as well. This is particularly the case since there are results indicating that the relationship between diffusion and semipresidentialism is highly complex and not as strong as the one existing between diffusion and parliamentarism (Åberg & Denk, 2020). Along with diffusion patterns, I expect the previous regime type to be of importance for explaining why dual executives emerge. In fact, separating regime heritage from diffusion is difficult, as regime transformations often occur in countries geographically close to each other, which share a common authoritarian regime form but also a common former ‘mother country’. As described above, the choice of the semi-presidential systems in former communist countries is often traced to the dual power structure that characterized the communist systems. In other words, the spread of semi-presidentialism in Eastern Europe is explained both with reference to constitutional heritage and diffusion; as countries framed their constitutions they tended to opt for systems that resembled their former constitutions, while, at the same time, being influenced by the

8

C. ANCKAR

political system imposed on them by the Soviet Union and constitutional choices made by surrounding countries with a similar past. The legacy of the previous regime form has been largely overlooked as a potential explanation for the birth and stability of subsequent regimes. Therefore, I pay extensive regard to the transition process from the previous regime form to the regime with a dual executive. During the last two decades, a handful of works have suggested that the previous regime type is important for explaining regime choice (Geddes, 1999; Hadenius & Teorell, 2007; Magaloni, 2008; Geddes et al., 2014; Anckar, 2018). Interestingly enough, authors have paid more attention to authoritarian regime forms than to democratic regime types, merely focusing on the question if certain types of authoritarian regimes are more likely than others to generate democracy. Nevertheless, the argument can be extended, and the ambition here is consequently to answer both the question if certain authoritarian regime types are more likely than others to generate dual executives, and also if certain democratic regime types are more likely than others to give rise to dual executives. Authors disagree strongly on how authoritarian regime forms should be classified. Within the framework of the present study, I apply the following categorization: absolute monarchy, single-party systems, multi-party authoritarian rule, military rule, personalist rule, monarchic oligarchy, and other oligarchy. The classification of the cases is based on Anckar and Fredriksson (2019). The general assumption is that absolute monarchies, monarchic oligarchies, and single-party systems are particularly likely to give rise to dual executives. The theoretical justification for linking monarchy to dual executives goes back to a line of reasoning which Huntington (1968, 177–191) described as ‘the king’s dilemma’. Absolute monarchy was the prevailing authoritarian regime form during the nineteenth century. We expect a dual executive with a monarch possessing more than ceremonial powers to exist for a limited period of time, as a transitionary stage, when countries move from absolute monarchy to constitutional monarchy with a monarch possessing ceremonial powers only. As discussed above with reference to the semi-presidential systems in Eastern Europe, single-party systems are expected to give rise to executive power sharing due to their dual power structure; one within the leading party and another within the formal institutions (parliament, government, head of state) (Blondel, 1992; Wu, 2011, 24–25). To some extent, the same constellation is valid also for multi-party authoritarian systems. However, in these

1

INTRODUCTION

9

systems, the formal institutions are far more important than in singleparty systems, which means that they should be somewhat less likely to transform into dual executives. The remaining authoritarian regime forms are not expected to transform into system with executive power sharing. Finally, dual executives can emerge from democratic regime forms, either parliamentary systems with a single head of state and government or from presidential regimes. Although there is little theoretical evidence suggesting that either of the two regime forms would be particularly likely to generate transitions towards dual executives, we cannot rule out the possibility that dual executives are more likely to emerge from one of these regime forms than from the other. However, the question is empirical rather than theoretical.

1.4

How Dual Executives Transform

The third aim of the study is to explain why certain forms of dual executives are more stable than others. Here, the point of departure is an assumption that dual executives vary extensively in terms of their bases of legitimacy. The discussion is largely based on Samuels and Shugart, 2010, 25–34) who used a principle-agent framework for classifying democratic regimes. In parliamentary systems, the government functions as an agent of parliament; it emerges from the parliament and can be brought down by the parliament. In presidential systems, the executive, i.e. the president, emerges from the electorate and does not depend on parliament for his or her survival in office (Samuels and Shugart, 2010, 26–27). Dual executives are more complicated as the bases of legitimacy of the head of state tend to vary significantly. In cases where the president is popularly elected for a fixed term in office, the position of the president resembles that of the position of the president in a presidential system. However, the two branches of the executive originate from separate institutions (the president from the electorate and the prime minister from the assembly and possibly the president) and are dependent on their survival on different institutions (the president is elected for a fixed term whereas the prime minister is dependent on the legislature and possibly the president). Things are different in situations where the president is elected by parliament. In these cases, both the prime minister and the president originate from the parliament. Nevertheless, as long as the president is elected for a fixed term in office, he or she does not depend on parliamentary support for remaining in office. The prime minister, on the other

10

C. ANCKAR

hand, depends on the parliamentary majority both for taking office and remaining in office. With regard to monarchies, the situation is the same as in systems with popularly elected presidents in the sense that the head of state does not depend on the legislature for his or her origin or survival in office. However, in contrast to dual executives with popularly as well as non-popularly elected presidents, the monarch is in complete lack of democratic legitimacy. Instead, the monarch’s legitimacy is based on the divine right of kingship. My general assumption is that the bases of legitimacy profoundly determines the exercise of powers of the head of state. Here, the difference between popularly and non-popularly elected presidents is crucial.2 The popular election of the president should enhance the legitimacy of the head of state and make it possible for him or her to be a real power holder, i.e. an actor who actively participates or even has a dominating role in the policy process. Since the president does not depend on the parliament for his or her survival in office, the president can challenge the prime minister and other ministers without having to fear the threat of being removed from office. Overall, then, the strong legitimacy of the popularly elected president is expected to strengthen the position of the president in relation to the prime minister (Duverger, 1980; Shugart and Carey, 1992; Baylis, 1996; Neto and Lobo, 2009). As non-popularly elected presidents do not have a popular basis of legitimacy, their room for manoeuvre is more restricted than for popularly elected presidents. However, since they are almost always elected for a fixed term in office, they are not necessarily confined to the position as puppets of the parliamentary majority. Anyhow, very powerful, nonpopularly elected presidents should be more difficult to find than strong, popularly elected presidents. The same assumption is valid, even more so, for powerful monarchs, who have no democratic basis of legitimacy whatsoever. Here, we can assume that any time period during which a monarch is in possession of substantial powers is limited, and that pressures for reducing the powers of the monarch will grow as democracy consolidates.

1.5

How Dual Executives End

As discussed above, dual executives can experience changes with regard to how the head of state is (s)elected and with regard to how much powers

1

INTRODUCTION

11

the head of state possesses. However, sometimes dual executives die. In such cases, they either transform into autocratic systems or to single executives.3 It is reasonable to suspect that the risk of democratic breakdown should be enhanced in situations where the head of state is popularly elected and when he or she is in possession of strong powers. This line of reasoning is largely based on Juan Linz (1994) discussions on the ‘perils of presidentialism’. In presidential systems, both the president and the legislature base their legitimacy on the people. This means that the propensity for conflicts between the two institutions is high. Popularly elected presidents, particularly in cases where they portray themselves as outsiders of the political establishment, can feel tempted to position themselves above their political contenders as ‘[t]he feeling of having independent power, a mandate from the people … is likely to give a president a sense of power and mission that might be out of proportion to the limited plurality that elected him’ (Linz, 1994, 19). There is every reason to suspect that the power dimension is of importance as well. The basic assumption is that the more powers the president possesses, the stronger the odds that he or she will be successful in challenging the parliament. A strong president, then, can even be tempted to concentrate even more powers into his or her hands, and, ultimately move the country into authoritarianism. We can assume that Linz’ warnings about the presidential form of government are of relevance also for dual executives. This is particularly the case in situations where the head of state is popularly elected, where political processes risk becoming personalized much in the same way as in presidential systems (Elgie, 2008, 52). There is also theoretical support for an assumption that countries with dual executives with powerful heads of state are particularly likely to suffer a democratic breakdown as ‘the dual control of the government is likely to generate institutional conflict,…, which may trigger potential crises of governability and eventually endanger democracy’ (Elgie and Schleiter, 2011, 45). Another question of interest concerns the regime types that dual executives transform into when they end. As we recall, we expected dual executives to emerge from single-party systems, absolute monarchies, or monarchic oligarchies. However, I do not know of any study which has attempted to map transformation patterns from dual executives to various types of autocratic systems. Admittedly, there are few theoretical arguments suggesting that some authoritarian regime types should be more likely than others to emerge in the wake of a democratic collapse in

12

C. ANCKAR

countries with dual executives. The only exception is a link between dual executives to personalist rule. Juan Linz (1994) in particular has pointed out that there is a risk that a political outsider, unconstrained by political parties, could become president and then challenge the democratic institutions and concentrate powers into his or her own hands. In such instances, the outcome would most likely be a personalist authoritarian system, where the person in power would remain the same but the regime type would change. The same uncertainty concerns transformation patterns from dual executives to democratic single head executives, either in the form of presidentialism or in the form of parliamentarism with a single-head government. To a considerable extent, therefore, the assessments of such transformation patterns are conducted by following an inductive approach rather than a deductive one.

1.6

Outline of the Book

The present contribution is organized in seven chapters. Chapter 1 introduces the theme of the book, the dual executive, and goes on by introducing the three overarching research questions, how dual executives come into existence, what patterns of transformation they follow in terms of mode of selection and powers of the head of state, and lastly how they end. Chapter 2 deals with the question how political regimes can be classified. The chapter functions as a stepping-stone to the main research objective. As regime transformations are central for two of the research questions, it is necessary to dwell on the subject quite extensively. Authoritarian regimes are classified into seven categories. With regard to democratic regimes, I discuss the problems involved in drawing lines between presidentialism, parliamentarism, and semi-presidentialism. I also introduce the category semi-constitutional democracy. In Chapter 3, I make the argument that for the purpose of the present book, the separation of dual executives from single executives is more relevant than the traditional categorization of democratic regimes discussed in Chapter 2. The chapter ends with a presentation of the research population. Chapter 4 sets out to explain why countries choose dual executives. The ambition is to test the proposition that dual executives are explained with reference to two plausible determinants, the regime form the country had prior to its transition to democracy or diffusion mechanisms. In Chapter 5, the ambition is to explain why certain forms of dual executives are more stable than others. Based

1

INTRODUCTION

13

on a principle-agent framework, I propose that powerful, non-popularly elected presidents and monarchs should be hard to find and in situations where they exist, there should be pressures for reducing their powers or, alternatively, introduce a popularly elected head of state. I run a number of statistical analyses in order to answer the question if increases or decreases of the powers of the heads of state are related to their previous power levels. A number of control variables are introduced and theoretically linked to the dependent variables. In Chapter 6, I provide an assessment of how dual executives die. Here, the ambition is to test, in each dual executive variant, if countries where the powers of the head of state are very strong are more likely to suffer from a democratic breakdown than countries where the powers of the head of state are weak or moderate. In addition, I give an assessment of transformation patterns from dual executives to other regimes forms, aiming to establish if certain variants of dual executives are more likely than others to transform into certain variants of authoritarian regime forms. Chapter 7, finally, summarizes the findings and discusses the implications for future research in the area.

1.7

Data and Methods

Until recently, measuring de facto powers of heads of state for the purpose of global historical studies was extremely difficult. With the introduction of the V-dem dataset, such comparisons are now possible. This is not the place to give a detailed account of the methodology applied in the dataset. For a comprehensive presentation of how the V-dem measures are calculated, the reader is referred to Coppedge et al. (2020). Suffice it here to say, that V-dem makes use of expert-coded data, where each observation in the dataset typically is based on judgements from five experts. Throughout the book, I employ a mixed-methods strategy. Variable interactions are typically measured by means of statistical analyses. Crosstabulations are useful to give a general view of how the variables of interest are related to each other and are used particularly in Chapter 4, in order to assess the relationship between, on the one hand, the previous regime form and the colonial heritage, and on the other hand, choice of dual executives. Cross-tabulations are also applied in Chapter 6, where the task is to trace transformation patterns from the various dual executive forms to authoritarian regimes.

14

C. ANCKAR

I also make use of Markov-chain probability analyses. This technique is particularly useful when assessing the stability of dual executives in Chapter 5. Here, for instance, we make the assumption that the probability of a head of state gaining or losing powers at time t is dependent on the level of powers the head of state possessed at time t −1, i.e. the previous year. For multivariate analyses, more sophisticated techniques are required. Since most of the analyses operate with a dichotomous dependent variable, logistic regression analyses are generally used in order to determine the significance of each of the independent variables. Logistic regressions are used in Chapter 4, in order to explain why countries choose a particular dual executive variant, in Chapter 5, in order to explain why heads of state lose or gain powers, and in Chapter 6, for the purpose of explaining why dual executives end. Although statistical techniques are important when assessing the overall relevance of the explanatory variables, they should be complemented with qualitative techniques. Although the number of cases included in the study is very high, the central dimensions, selection mode and powers of the head of state, do not vary much across time and this is particularly the case with regard to selection mode. Especially when multivariate statistical models are run separately in the three selection mode categories, instances where the phenomenon of interest occurs are sometimes limited. It should be readily admittedly, that the quantitative methods are in the foreground; the ambition is not to conduct deep-rooted cases studies or advanced process tracing techniques for a limited number of cases. The goal is more modest. Sometimes the qualitative ambition is restricted to merely listing the cases reduced to numerical terms in the tables. In other instances, cases placed in a specific category are scrutinized in somewhat more detail, in order to find a common characteristic that could explain their placement within the category in question. The use of qualitative techniques is also motivated by the fact that the criteria of categorizations are far from always crystal clear. This is particularly the case with regime classifications, where the distinction between the categories single-party rule, personalist rule, and military rule can be extremely difficult as many regimes incorporate elements of all three regime types. Moreover, some regimes exist for very short periods of time, during transitionary stages between two ‘ordinary’ regimes. The same problem arises also with regard to the mode of selection of the president. In a number of countries, the president has been elected by parliament during an initial period, for instance after the country has transformed

1

INTRODUCTION

15

into a democracy from an autocracy or when it has received its independence, while a new constitution has been drafted. Isolating and discussing such ambiguous instances makes it possible to provide a more nuanced picture of the relationships between the variables of interest.

1.8

Contribution to the Field of Research

During the last two decades, there has been a growing interest in intraexecutive relationships. Robert Elgie (2016, 54) has identified three waves of studies in semi-presidentialism: the first one focusing on how semipresidentialism should be defined, the second one on the relationship between semi-presidentialism and democratization, and the third, more heterogeneous one, focusing on ‘government formation and termination’ as well as ‘parties, power and parliament’. Within each of these waves, a significant number of research efforts have been made, resulting in many articles but also a number of monographs (e.g. Brunclík & Kubát, 2019; Köker, 2017; Raunio & Sedelius, 2020; Sedelius, 2006; Tavits, 2009). Despite the growing literature in the field, there are still some gaps that need to be filled. Åberg and Sedelius (2020, 22–23) have noted that the research field semi-presidentialism is characterized by a ‘bias towards single-case studies of European and post-communist countries’ and called for ‘more methodologically sophisticated and empirically sound large-N studies’. Köker (2017, 2), on his part, has noted that while there has been a growing focus on presidential powers, surprisingly few authors have actually studied how these powers are used. Finally, studies on intraexecutive relationships have generally focused on republics only, thereby neglecting monarchies. The fact that executive power sharing in monarchies has received very little scholarly attention is understandable in the sense that, by definition, there is no place for a powerful hereditary monarch in a democratic system. Consequently, it is self-evident that executive power sharing mechanisms are of less relevance in democratic monarchies than in republics. Nevertheless, there are examples of systems classified as democracies where the monarch has, or has had, more or less the same position as a president in many semi-presidential systems. It is therefore essential, that these systems are accounted for as well. The present work contributes to the literature on intra-executive relations in three important respects. First, it moves beyond the category of semi-presidentialism, accounting not only for systems with popularly

16

C. ANCKAR

and non-popularly elected presidents but also for constitutional monarchies. Second, the study is extensive in time and space as it covers all democratic countries with shared executives during the time period 1850–2019. A longer time perspective is particularly important when studying phenomena like regime choice and regime transformations, since these tend to occur infrequently. Third, whereas authors have generally measured presidential powers according to constitutional provisions, the present work sets out to assess the powers of the head of state in practice. The V-dem dataset is particularly useful for the purpose of the present study as it provides measures of many prerogatives not only for presidents but also for monarchs during a time period of more than two hundred years.

Notes 1. Although Maurice Duverger is largely recognized as the ‘father’ of the concept of semi-presidentialism, the term ‘régime semi-présidentiel’ had been introduced already in 1959 by Hubert Beuve-Méry (cited in Chevrier, 2010, 65–97). 2. Here, a terminological clarification is needed. In the present contribution, I separate popularly elected presidents from non-popularly elected ones. An alternative strategy would be to apply the categories directly elected and indirectly elected presidents. This is the strategy adopted by, for instance Tavits (2009). However, indirectly elected presidents can also be popularly elected in cases where there is a popularly elected assembly elected for the sole purpose of conducting the final election of the head of state. In other words, the American presidents are popularly but indirectly elected, through the Electoral College, much in the same way the Finnish presidents were popularly but indirectly elected during the years 1925–1988, when the Finnish Electoral College existed. 3. A third, mostly theoretical, possibility is that they transform into a collegial presidential executive.

References Åberg, J., & Denk, T. (2020). Diffusion and the choice of democratic government system at the time of democratisation. Zeitschrift Für Vergleichende Politikwissenschaft, 14(2), 75–98.

1

INTRODUCTION

17

Åberg, J., & Sedelius, T. (2020). A structured review of semi-presidential studies: Debates, results and missing pieces. British Journal of Political Science, 50(3), 1111–1136. Anckar, C. (2018). Global patterns of regime change 1800–2015. Political Science, 70(3), 207–233. Anckar, C., & Fredriksson, C. (2019). Classifying political regimes 1800–2016: A typology and a new dataset. European Political Science, 18(1), 84–96. Anckar, D. (1984). Folket och presidenten. En författningspolitisk studie. Helsingfors: Finska Vetenskaps-Societeten (Bidrag till kännedom av Finlands natur och folk H. 132). Baylis, T. A. (1996). Presidents versus Prime Ministers: Shaping executive authority in Eastern Europe. World Politics, 48(3), 297–323. Blondel, J. (1992). Dual leadership in the contemporary world. In Lijphart, A. (Ed.). Parliamentary versus presidential government. Oxford University Press. Brunclík, M., & Kubát, M. (2019). Semi-presidentialism, Parliamentarism, and Presidents: presidential politics in Central Europe. Routledge. Chapsal, J. (1981). La vie politique sous la Ve République. Presses universitaires de France. Chevrier, M. (2010). D’une contre-démocratie à l’autre’. In M. Chevrier & I. Gusse (Eds.), La France depuis de Gaulle: La Ve République en perspective (pp. 65–97). Presses de l’Université de Montréal. Colliard, J.-C. (1978). Les régimes parlamentaires contemporains. Presse de la Fondation Natoinale de Sciences Politiques. Coppedge, M., Gerring, J., Glynn, A., Knutsen, C. H., Lindberg, S., Pemstein, D., Seim, B., Skanning, S., & Teorell, J. (2020). Varieties of democracy: Measuring two centuries of political change. Cambridge University Press. Coppedge, M., Gerring, J., Knutsen, C. H., Lindberg, S., Teorell, J., Altman, D., Bernhard, M., Fish, M. S, Glynn, A., Hicken, A., Lührmann, A, Marquardt, K. L., McMann, K., Paxton, P., Pemstein, D., Seim, B., Sigman, R., Skaaning, S.-E., Staton, J., Cornell, A., Gastaldi, L., Gjerløw, H., Mechkova, V., von Römer, J., Sundström, A., Tzelgov, E., Uberti, L., Wang, Y.-t., Wig, T., Ziblatt, D. (2020). V-Dem codebook v10. Varieties of Democracy (VDem) Project. Debré, M., (1985). Discours de M. Michel Debré devant le Conseil d’Etat le 27 août. 1958. In D. Maus (Ed.), 1998 (p. 3). Les grands textes de la pratique constitutionelle de la Veme République. Paris. Duhamel, O. (1988). Remarques sur la notion du régime semi-présidentiel. In D. Colas & C. Emeri (Eds.), Droit, Institutions, et Systèmes Politiques: Mélanges en hommage à Maurice Duverger (pp. 581–590). Presses Universitaire de France. Duverger, M. (1970). Institutions et droit constitutionnelle, 11e édition. Presses Universitaire de France. Duverger, M. (1978). Échec au roi. Albin Michel.

18

C. ANCKAR

Duverger, M. (1980). A new political system model: Semi-presidential government. European Journal of Political Research, 8(2), 165–187. Elgie, R. (2008). The perils of semi-presidentialism: Are they exaggerated? Democratization, 15(1), 49–66. Elgie, R. (2011). Semi-presidentialism: An Increasingly common constitutional choice. In R. Elgie, S. Moestrup, & Y. S. Wu (Eds.), Semi-presidentialism and democracy. Palgrave Macmillan. Elgie, R. (2012). The President of Ireland in Comparative Perspective. Irish Political Studies, 27 (4), 502–521. Elgie, R. (2016). Three waves of semi-presidential studies. Democratization, 23(1), 54. Elgie, R., & Schleiter, P. (2011). Variation in the durability of semi-presidential democracies. Palgrave MacMillan Finger, A. (1923). Das Staatsrecht des Deutschen Reichs: Der Verfassung vom 11. August 1919. Ferdinand Enke Verlag. Geddes, B. (1999). What do we know about democratization after twenty years? Annual Review of Political Science, 2(1), 115–144. Geddes, B., Wright, J., & Frantz, E. (2014). Autocratic regimes and transitions. Perspectives on Politics, 12(2), 313–331. Hadenius, A., & Teorell, J. (2007). Pathways from authoritarianism. Journal of Democracy, 18(1), 143–156. Haungs, F. (1968). Reichspräsident und Parlamentarische Kabinettsregierung. Westdeutscher Verlag. Huntington, S. P. (1968). Political order in changing societies. Yale University Press. Köker, P. (2017). Presidential activism and veto power in Central and Eastern Europe. Palgrave Studies in Presidential Politics. Lijphart, A. (Ed.). (1992). Parliamentary versus presidential government. Oxford University Press. Lindman, S. (1937). Studier över parlamentarismens tillämpning i Finland 1919– 1926. Åbo Akademi (Acta Academia Aboensis, Humaniora, Vol. XH, nr. 1). Lin, S. (1974). Presidentens Ställning Och Grundlagsreformen. Finsk Tidskrift, 98, 439–456. Linz, J. (1994). Presidential or parliamentary democracy: Does it make a difference? In J. Linz & A. Valenzuela (Eds.), The Failure of Presidential Democracy (Vol. I, pp. 3–87). Johns Hopkins University Press. Linz, J., & Valenzuela, A. (Eds.). (1994). The failure of presidential democracy, volumes I and II . Johns Hopkins University Press. Magaloni, B. (2008). Credible power-sharing and the longevity of authoritarian rule. Comparative Political Studies, 41(4/5), 715–741.

1

INTRODUCTION

19

Neto, O. A., & Lobo, M. C. (2009). Portugal’s semi-presidentialism (re)considered: An assessment of the president’s role in the policy process, 1976–2006. European Journal of Political Research, 48(2), 234–255. Pereira, A. G. (1984). O semipresidencialismo em Portugal. Ática. Preuss, H. (1926). Staat, Recht und Freiheit: Aus 40 Jahren deutscher Politik und Geschichte. Mohr. Raunio, T., & Sedelius, T. (2020). Semi-presidential policy-making in Europe: Executive coordination and political leadership. Palgrave Studies in Presidential Politics. Samuels, D. J., & Shugart, M. S. (2010). Presidents, Parties, and Prime Ministers. Cambridge University Press. Sartori, G. (1994). Comparative constitutional engineering: An inquiry into structures, incentives and outcomes. Palgrave Macmillan. Sedelius, T. (2006). The tug-of-war between presidents and prime ministers: Semi presidentialism in Central and Eastern Europe. Örebro University Press. Shugart, M. S. (2005). Semi-presidential systems: Dual executive and mixed authority patterns. French Politics, 3(3), 323–351. Shugart, M. S. & Carey, J. M. (1992). Presidents and assemblies: Constitutional design and electoral dynamics. Cambridge University Press. Siaroff, A. (2003). Comparative presidencies: The inadequacy of the presidential, semi presidential and parliamentary distinction. European Journal of Political Research, 42(3), 287–312. Sousa, M. R. (1984). O sistema de governo português antes e depois da Revisão Constitucional. Cognitio. Tarkiainen, T. (1938). Tasavallan presidentin asema Suomen parlamenttaarisessa hallitusjärjestelmässä. K. J. Gumrnerus Osakeyhtiö. Tavits, M. (2009). Presidents with prime ministers: Do direct elections matter? Oxford University Press. Wu, Y. S. (2011). Clustering of semi-presidentialism: A first cut. In R. Elgie, S. Moestrup, & Y. S. Wu (Eds.), Semi-presidentialism and democracy (pp. 21– 41). Palgrave Macmillan.

CHAPTER 2

Classifying Political Regimes

2.1

Introduction

Any attempt to classify regimes must start by specifying what we mean by ‘regime’. In recent years, a substantial number of new global datasets on political regimes have emerged (e.g. Alvarez et al., 1996; Boix 2013; Cheibub et al., 2010; Coppedge et al., 2020; Knutsen et al., 2016). The growing amount of longitudinal data on regime characteristics is natural, given the popularity of comparative historical analysis. However, with the proliferation of data on regime characteristics, it has also become increasingly obvious that authors differ widely with regard to how regimes are conceptualized and classified. The most basic regime distinction is the one that exists between democratic systems and autocratic ones. Here, two strategies are available: The first alternative is to consider differences between the systems as differences in degree rather than kind. This means that countries are placed on a continuum ranging from a maximal level of democracy to a maximal level of autocracy (e.g. Coppedge et al., 2020; Hadenius 1992; Vanhanen, 2013). The second alternative is to make a qualitative distinction between the categories and thus place countries in either the category democracy or autocracy (Alvarez et al., 1996; Boix 2013). In the present research

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 C. Anckar, Presidents, Monarchs, and Prime Ministers, Palgrave Studies in Presidential Politics, https://doi.org/10.1007/978-3-031-03960-7_2

21

22

C. ANCKAR

effort, I make use of the latter strategy, applying a qualitative distinction between autocracy and democracy and then continue by conducting further subdivisions within each of these two categories. Evidently, drawing a line between democratic and authoritarian regime forms is a delicate task as many countries balance on the edge between the categories. Nevertheless, such a procedure is necessary, as the aim of the study is to account for patterns of regime transformations between qualitatively distinct regimes. I shall return to the subject in the next chapter, when I discuss the criteria for establishing the research population. The present chapter will be devoted to a general presentation of the problems involved in the classification of authoritarian and democratic regime forms. I first present a classification of authoritarian systems, which will be used in subsequent analyses throughout this book. I then discuss the classification of democratic regime forms, with a special focus on the problems involved in classifying systems that are neither fully presidential nor fully parliamentary. I begin with an assessment of the authoritarian variants, and then shift focus to the classifications of democratic regimes.

2.2

Classifying Authoritarian Regimes

Three of the most widely used data sets on autocratic political regimes, namely Geddes (2014), Cheibub et al. (2010), and Hadenius and Teorell (2007) (Wahman et al., 2013) differ to a considerable extent from each other in terms of both conceptualization and measurement of ‘regime’ (e.g. Roller, 2013). A first question concerns if a regime should be based on personal traits or institutional characteristics. Hadenius and Teorell base their categorization of regimes ‘on the institutions on which … elites rely in order to regulate the access to and maintenance of public authority’ (Wahman et al., 2013: 21, emphasis omitted). As a consequence of this definition, ‘the identity or longevity of dictators is left out of the picture as are the ideological stance of the regime and its source of legitimation’ (Wahman et al., 2013: 20). Geddes et al. (2014, 314), on their part, choose a different strategy, where emphasis is on ‘the rules that identify the group from which leaders can come and determine who influences leadership choice and policy’. As a practical consequence of the two different definitions, Geddes, Wright, and Frantz sometimes identify several regimes under a period when the regime type remains the same (Geddes 2014, 315, 323). In contrast, in the dataset by Wahman et al. (2013), regime and regime type coalesce. A

2

CLASSIFYING POLITICAL REGIMES

23

similar conception of regime change is held by Cheibub et al. (2010) in their global dataset. A second question concerns what categories we should apply when we classify autocratic regimes. The question is pertinent as authors tend to disagree heavily on the categorization of autocratic regimes (Bell, 2016; Brooker, 2013; Cheibub et al., 2010; Diamond, 2002; Geddes et al., 2014; Kailitz, 2013; Linz & Stepan, 1996; Wahman et al., 2013) and some authors go as far as questioning the meaningfulness of conducting such classifications altogether (Barros, 2016). Accordingly, Hadenius and Teorell identify four authoritarian regime types: monarchy, military, one party, and multi-party. Geddes, Wright, and Frantz, again, operate with four broad categories, monarchy, military, party-based, and personalist regimes. Kailitz (2013) introduces the categories monarchy, personalist autocracy, military autocracy, communist ideocracy, one-party autocracy, and electoral autocracy. The broadest categorization is the one suggested by Cheibub et al. (2010) who only distinguish between monarchy, military, and civilian authoritarian regime forms. Since I strongly subscribe to the principle that that social science should progress cumulatively, and that each of the schemes of classification has its own merits, the categorization applied within the framework of the present research effort is largely derived from the discussions conducted in these pioneering works. At the same time, as will be evident from the presentation, I do not consider any of the three existing regime categorizations to be unassailable, wherefore many of the criteria of classifications are modified. This strategy is motivated also by practical circumstances; since the time period covered here is much longer than in previous studies, we need operational criteria that are parsimonious. Following Hadenius and Teorell (2007), I rely on institutional characteristics when defining political regimes. In other words, the mere fact that a political leader is replaced by another does not necessarily imply a regime change. The crucial issue is whether or not the regime type remains the same or not. In the present chapter, authoritarian countries are classified into the following regime forms: single-party rule, multi-party authoritarian rule, personalist rule, military rule, absolute monarchy, monarchic oligarchy, and other oligarchy. The categorization is discussed somewhat more briefly in a work that introduces a global database on regime forms (Anckar & Fredriksson, 2019). Here, I provide

24

C. ANCKAR

a more extensive account of the central arguments for applying this categorization and also give a short description of the basic strategy applied when conducting this classification. The category absolute monarchy is uncontroversial as it is included in all reputable datasets on authoritarian regime forms. The classical notion of absolute monarchy involves a power holder who bears the title King, exercises unlimited powers, and has inherited his position from his father. Although hereditary succession, usually from father to son, is strongly associated with the term monarchy, many of the countries currently regarded as monarchies do not practise primogeniture. Instead, the process of selecting a monarch is in the hands of the royal family. This is the case in countries like Kuwait, Oman, and Saudi Arabia (Cheibub et al., 2010, 85; Herb, 1999). There tends to be little disagreement among authors regarding the defining characteristics of a monarchy. Cheibub et al. (2009, 7) apply two criteria of monarchy (or, in their terminology, hereditary royalty), namely that the power holders bear ‘a title such as kings, emirs, sultans, and … have been preceded or succeeded by a relative’. In addition, they disregard self-proclaimed royals, ‘such as Bokassa in the Central African Republic’) from the category. Geddes et al. (2014, 318) define as a monarchy a system where ‘control over policy, leadership selection, and the security apparatus is in the hands of … a royal family’. Hadenius and Teorell regard as monarchies ‘those regimes in which a person of royal descent has inherited the position of head of state in accordance with accepted practice and/or the constitution’ (Wahman et al., 2013, 25). Furthermore, they add the criterion ‘one cannot proclaim oneself a monarch’ (Wahman et al., 2013, 25). In the present work, I apply the definition by Hadenius and Teorell and use a generous definition of ‘accepted practice’, meaning that instances where the monarch is ousted of office and replaced by family members are accepted as hereditary succession. Let it be clearly spelled out, though, that ‘accepted practice’ does not mean that all cases where power is passed from father to son count as monarchies (Wahman et al., 2013, 25). For instance, in Syria, Bashar al Assad succeeded his father Hafez al-Assad as president in 2000 but he did not inherit the position as president; instead he was approved for the position in a public referendum according to the constitutional provisions of the country. Also, in line with Hadenius and Teorell, I stipulate that a monarchy does not emerge simply because a political leader, like for instance,

2

CLASSIFYING POLITICAL REGIMES

25

Bokassa in Central African Republic in 1976, proclaims him- or herself king, queen, or emperor etc. On the other hand, cases where a royal dynasty is substituted with another royal dynasty are generally counted as a continuation of the monarchic regime type if the new dynasty was installed by another actor or institution than the new monarch himself. Such scenarios were not uncommon in Europe during the nineteenth century, where new monarchical dynasties were generally installed by the parliament. The same strategy applies to countries where a monarch has been (s)elected by parliament or third parties at the time the country received its independence. Thus, Bulgaria is classified as a monarchy from its independence in 1879, when the Grand National Assembly elected Alexander I, of the House of Battenberg and nephew of the Russian Tsar, to the throne and the same classification applies to the regime of Ferdinand I, of the House of Saxe-Coburg and Gotha, who was elected King by the Bulgarian parliament in 1887, after the abdication of Alexander I. The definition of military rule is somewhat more controversial than the definition of monarchy. Cheibub et al. (2010, 88) make use of a wide and straightforward definition of military regimes, stipulating that ‘[t]he effective head of government is a military ruler if he is or was a member of the institutionalized military prior to taking power’. Another strategy is the one proposed by Geddes et al. (2014, 319), who define military rule as ‘rule by an officer constrained by other officers’. Thus, their conception of military rule is based on a separation of ‘rule by “the military as an institution” … from rule by a military strongman’(Geddes et al., 2014, 319). The requirement that the leader should be constrained by other officers can, however, be questioned on the ground that autocratic rule is a pronounced feature of the armed forces. It is not evident why a country where a general rules the country autocratically, holding the military apparatus in his firm grips, should not qualify as military rule whereas a country where the members of the military junta continuously compete for power should. Instead, it could very well be argued that the military regime is more organized, stable, and effective in the former case than in the latter. Following Nordlinger (1977, 2), Hadenius and Teorell define military regimes as instances ‘in which military officers are major or predominant actors by virtue of their actual or threatened use of force’, and where ‘the armed forces control politics directly or indirectly (i.e. by directing civilian leaders behind the scene)’ (Wahman et al., 2013, 25). This definition is more inclusive than the one proposed by Geddes, Wright, and Frantz,

26

C. ANCKAR

as it includes among the military regimes also countries where power is concentrated into the hands of a single military officer. At the same time, it is less inclusive than the definition used by Cheibub, Gandhi, and Vreeland, since it does not automatically regard all regimes led by a former military officer as military regimes. There is, however, one drawback with Hadenius’ and Teorell’s definition. Although it makes sense to classify a country in which the military controls the political sphere indirectly rather than directly as a country ruled by a military regime, it is quite difficult to stipulate exactly how much control the military should exercise in order for the regime to be regarded as a military regime. Furthermore, military regimes are often temporary; the military takeover is frequently motivated by a need to restore order and stability in a turbulent political system. The underlying expectation is that the military authorities will return powers to civilian authorities as soon as order has been restored. The military is of course aware of such expectations and consequently, in situations where it is reluctant to give up its powers, instead tends to transform itself to a civilian government. Such a transformation normally involves substituting uniforms with suits, the restoration of the central political institutions (notably the legislature), the creation of a political party which supports the leader, and, finally, legitimizing the rule of the former military leader by holding elections. The tricky question, then, is to determine when a military regime is turned into something else. The fact that the military establishes a party which prevails in a fraud election does not turn a military regime into a single-party regime and neither does the fact that a general puts on a suit and assumes the office of president turn the system into personalist rule. These problems are of course aggravated by the fact that the study covers an extensive period of time. Here, the basic rule has been to consider a regime as a military regime during the whole period it has been uninterruptedly ruled by the same person who came to power in a military coup if the country has remained autocratic during this period. This rule, however, also has exceptions. There are cases where a military leader has transformed himself into a civilian leader and then stayed in power for many years. In such cases, the power base of the leader can have switched from the military to another institution. Admittedly, determining when such a switch has occurred is not always easy. The case of Paraguay provides an illustrative example in this respect. Following, the ‘basic rule’ described above, I consider Paraguay to have

2

CLASSIFYING POLITICAL REGIMES

27

had a military regime during the whole reign of Alfredo Stroessner, who came to power in a military coup in 1954 and was ousted of office in a military coup in 1989. However, elections (presidential and parliamentary) were held regularly during the whole period of Stroessner’s rule. In each of these elections the Colorado Party, which was effectively in the hands of Stroessner, received an absolute majority of the votes. Hadenius and Teorell classify Paraguay as a multi-party authoritarian system (or, rather, a limited multipartism hybrid) during Stroessner’s rule. Cheibub, Gandhi, and Vreeland regard the regime as a military regime whereas Geddes, Wright, and Frantz consider it to be a party regime (or, rather, a party-personal-military hybrid). Finally, we have to tackle the issue of defining ‘military’. A number of leaders have come to power as a result of a civil war and the question, then, concerns whether or not we should regard leaders of rebel movements as military leaders or not. Cheibub et al. (2009: 6–7) ‘do not consider rulers who come to power as head of guerilla movements as military’. Hadenius and Teorell apply a different strategy and include among military regimes ‘cases where a rebel movement (one not formed out of the regular armed forces) has taken power by military means, and the regime has not as yet been reconstituted as another kind of regime’ (Hadenius et al., 2012, 6). Geddes, Wright, and Frantz do not explicitly state how they deal with rebel forces who take power but an examination of their dataset support the view that rebel regimes are not counted as military regimes.1 In the present work, I follow the strategy applied by Hadenius and Teorell and consider a rebel regime a variant of a military regime. In both cases, the regime is made up of the armed forces in control of the country and dependent on the support of this group. However, regarding rebel regimes, the regime is not necessarily regarded as a military regime during the whole period the leader of the rebellion stays in power, since leaders of armed rebellions are far from always deeply rooted in the military institution. Geddes (2003, 51) defines one-party rule as systems where ‘access to political office and control over policy are dominated by one party, though other parties may exist and compete as minor players in elections’. This definition makes perfect sense if we want to separate systems where power is concentrated to the leading party from other authoritarian systems. At the same time, most countries that do not meet the requirements of democracy have political parties and even tend to conduct

28

C. ANCKAR

elections on a regular basis. However, this category of countries is actually quite heterogeneous in terms of party formation and possibilities of minor parties to participate in elections. In the most extreme cases, such as the Soviet Union and currently Laos only one party is legally allowed. Another subcategory consists of cases where some satellite parties, which can participate in elections and gain representation in parliament, exist but power is concentrated into the hands of a dominant ruling party. This was the situation in the Eastern European countries during the Cold War and currently the case in China. A final subcategory is made up of countries where real opposition parties are allowed to exist and participate in elections but the where the dominant party ensures that it always can rely on the support of a majority of the members in parliament. In order to account for such variations, Hadenius and Teorell operate with three (originally four) regime categories for states with elected assemblies, namely no-party, one-party, and multi-party regimes (Wahman et al., 2013). As it turns out, no-party regimes are extremely difficult to find (the authors note that the only country which fits into this category is Maldives until 2005). One-party regimes are ‘authoritarian regimes with only one legal party (formally or de facto)’ (Wahman et al., 2013, 26) but the category also includes systems where ‘satellite parties’ i.e. parties ‘that are independent in name but [that] do not take oppositional positions’ exist. Multi-party regimes, again, occur where ‘at least a minimal level of competition is allowed and some opposition candidates (although not necessarily all) are allowed to participate in national elections’ (Wahman et al., 2013, 27). The present work applies a subcategorization of party-based regimes which follows Hadenius’ and Teorell’s line of reasoning, making a distinction between single-party regimes and multi-party authoritarian regimes. Finally, it is worth pointing out that the mere existence of a political party does not mean that an authoritarian regime is classified as a party-based regime. Political parties are, or have been, allowed in many countries ruled by military regimes, for instance in Algeria, Brazil, and Myanmar, and also in many monarchies, such as Jordan, Morocco, and Tonga. In other words, the existence of a political party is a necessary but not sufficient criterion for inclusion in the categories single-party or multi-party authoritarian regimes. The category personalist rule is the most controversial authoritarian regime form. With reference to Brooker (2000), Hadenius and Teorell dismiss the notion of personalist regimes altogether, claiming that ‘[t]he

2

CLASSIFYING POLITICAL REGIMES

29

focus on personalism as a trait of dictatorship implies a conflation of regime and leader characteristics’ (Wahman et al., 2013, 30). Personalism, they argue, is instead a feature that denotes all authoritarian systems to varying degrees (Wahman et al., 2013, 31). Geddes et al. (2014, 319) take a different view, arguing that personalist regimes are ‘autocracies in which discretion over policy and personnel are concentrated in the hands of one man, military or civilian’. This definition comes close to the one proposed by Huntington (1991, 111) who defined personal dictatorship as a system where ‘the individual leader is the source of authority and [where] power depends on access to, closeness to, dependence on, and support from the leader’. In line with Geddes et al. (2014) as well as Kailitz (2013), I recognize personalist regimes as a separate authoritarian regime type. The most important reason for this choice is that the classification of regime types is not exhaustive unless we account for such systems. Recall that Hadenius and Teorell base their classification on three modes of power access and maintenance: hereditary succession (monarchies), military force (military regime), and popular election (party-based regimes). However, such a classification does not account for systems where the power holder lacks a strong institutional base for his exercise of power, and where he or she must rely on other means in order to remain in power. Admittedly, defining and operationalizing personalism is not uncomplicated. However, Juan Linz’ (2000, 151–155) definition of sultanism (which is based on Max Weber’s [1968] notion of the concept in question) constitutes an excellent point of departure for determining the limits of personalist rule (see also Geddes 1999, 125). According to Linz, sultanism is characterized by a fusion of the personal and public sphere, where ‘[t]he ruler exercises his power without restraint at his own discretion and above all unencumbered by rules or by any commitment to an ideology or value system’ (Linz, 2000, 151–152). The lack of an ideology is crucial for the separation of personalist regimes and party-based regimes. Although ruthless leaders in single-party regimes can behave much in the same respect as leaders in personalist regimes, their power is exercised with reference to an ideological totality. In a sultanistic society, on the other hand, if anything resembling an ideology exists, it has emanated from, and is consistent with, the person in power. In other words, as Linz and Stepan (1996, 53) argue, ‘there could be questions raised as to whether Stalin’s practices and statements were consistent with

30

C. ANCKAR

Marxism-Leninism, but there would be no reason for anyone to debate whether Trujillo’s statements were consistent with Trujilloism’. The other central distinctive feature of sultanism (and personalist rule) is a low degree of institutionalization and organization. In sultanistic systems, public office holders receive their positions from the ruler in person and since their careers are completely dependent on the capriciousness of the ruler, they are loyal to the ruler in person, not the office (Linz, 2000, 152–153). Furthermore, the bureaucracy is not meritocratic and the persons who receive the highest positions in the regime ‘are neither “disciples” nor old fighters of a movement party or conspiratorial group’, but rather ‘men who would not enjoy any prestige or esteem in the society on their own account … [and] whose power is derived exclusively from the ruler’ (Linz, 2000, 151–152). Admittedly, drawing the line between party-based rule and personalist rule is sometimes marred with difficulties. Following Geddes et al., (2014), a guiding principle is to make a distinction between situations where the leader is dependent on a political party (as is the case in most communist systems) and situations where the political party is merely a tool for the leader in his exercise of power (as in Russia or Belarus). In the first case, the political system is a single-party regime, and in the second case, a personalist regime. Of course it is often quite difficult to determine to what extent a political leader controls the party apparatus or vice versa. We therefore need some further guidelines in order to separate personalist regimes from one-party regimes. Although Linz’ criteria of sultanism are very helpful for separating personalist regimes from party-based regimes we cannot escape the fact that many cases where the leader has concentrated a great deal of power into his own hands fall in a gray area between the two categories. In such cases, it is of particular importance to try to establish the level of importance of the ruling party in the state apparatus. Following Geddes et al. (2014, 329) focus is here on the extent to which party organization ‘penetrates’ the whole country and how deeply rooted the party is in the bureaucracy. The V-Dem dataset (Coppedge et al., 2020) also provides information that is very useful for making a distinction between regime types, especially for distinguishing between single-party regimes and personalist regimes.2 Today, most of the independent autocratic countries of the world can be placed into the four categories discussed above. However, in a historical perspective, there are many cases that are difficult to fit into

2

CLASSIFYING POLITICAL REGIMES

31

the categories since countries have experienced very difficult processes of democratization. In some cases, such as Japan, democracy has been installed abruptly by one or several external powers, whereas in countries like Britain or Sweden, democratization was a piecemeal process, extending over many decades. Dahl (1971) has argued that democracy is made up of the components public contestation and inclusiveness, and further made the claim that the prospects for democracy could look very different depending on whether contestation preceded inclusiveness or vice versa. For the purpose of the present work, the classification of the countries is particularly troublesome in periods when only one of these democratic components has been in place. A necessary requirement for a country to be considered democratic by Boix et al. (2013) is that a majority of the adult male population has the right to vote. In other words, a country which meets the criteria of democracy with regard to public contestation (i.e. with free and fair elections and with an executive responsible either to the legislature or the people) remains an autocracy as long as less than half of the adult male population has the right to vote. Such systems cannot be placed into any of the four authoritarian systems categories discussed so far. In line with Geddes et al. (2014, 18) such systems are classified as oligarchies. However, in contrast to Geddes, Wright, and Frantz, oligarchies are not collapsed into the party-based category. The reason for this is that countries classified as oligarchies were not necessarily dominated by one single party (although this was indeed the case in South Africa from 1948 to 1994, with the Nationalist Party in power). Instead, two or more parties competed for power in fair elections and election results determined the composition of government. The category of oligarchies also comprises a handful of other types of political systems, where power is shared by a small group of people. At present, this is the case in Iran, where power effectively resides in the hand of the religious clergy. Within oligarchies, there is also a subcategory which is of particular interest for the purpose of the present study. In many of the European constitutional monarchies power gradually shifted from the monarch to the parliament and government. At some point in history, power was divided between the monarch and the parliament and/or government, and often it was far from clear which of the institutions that prevailed over the other. In a number of cases, the periods with a division of powers between the monarch and the prime minister extended over many years. In this respect, Britain constitutes an extreme example; its process of

32

C. ANCKAR

democratization started with the Magna Carta in 1215 and, according to Boix, Miller, and Rosato, the country met the requirements of democracy in 1885, after the passing of the Third Reform Act and the Redistribution of Seats Act. The question, then, is how a country like Britain should be classified during the time period 1800–1884. It is evident that it contains elements of both oligarchy (the House of Commons was elected in free and fair elections but the right to vote was restricted) and monarchy. However, the division of powers between the monarch and the prime ministers was obscure (the fact that Britain did not have a written constitution no doubt added to this uncertainty). Already during the reign of George III (1810–1820) the principle that the person in command of the majority of the votes in the House of Commons should be prime minister became established (Bagehot, 1872, 247). However, it is equally evident that the monarch continued to possess quite extensive executive powers after that point in time; as late as 1834, William IV, on his own authority, dismissed the Prime Minister Lord Melbourne and replaced him by Sir Robert Peel. The powers of the monarch then steadily declined during Queen Victoria’s long reign (1837–1901) but as long as Britain did not meet the requirements of democracy her powers in the executive sphere were significantly greater than those of a purely ceremonial head of state. The political system of Britain, therefore, was a special variant of oligarchy and the term monarchic oligarchy is consequently used in order to describe all systems where a hereditary monarch shares power with a parliament and government in a system with free and fair elections but a restricted right to vote.

2.3

Classifying Democratic Regimes

There is wide agreement in the literature that a presidential system presupposes three necessary criteria: (1) the president (or rather the chief executive) is elected by popular vote; (2) the government cannot be dismissed by a parliamentary vote of no confidence; and (3) the president appoints and directs the government (e.g. Sartori, 1997, 83–84; Shugart & Carey, 1992, 19). There is less agreement on whether or not additional criteria are necessary in order for a system to be labelled presidential. It is often argued that the position of head of state and head of government must coincide, but this argument is refuted by Lijphart (1992, 4–5), who argues that

2

CLASSIFYING POLITICAL REGIMES

33

the existence of a mere ceremonial head of state does not make a system something else than presidential as long as all the other criteria of presidentialism are met (also Sartori 1997, 84). Shugart and Carey (1992, 19), on their part, argue that it is necessary to introduce the criterion that ‘the president has some constitutionally granted lawmaking authority’. Their argument is that otherwise ‘[chief executives] execute laws the creation of which they had no way of influencing’. Sartori (1997, 97), however, omits this criterion, arguing that it is vague and unnecessary; the fact that the president heads and directs the government by necessity implies that he or she has some law-making authority. In contrast to Shugart and Carey, I would not argue that granting the president law-making authority constitutes a necessary condition of presidentialism. In fact, one could even claim the opposite. As Shugart and Carey (1992, 18) point out themselves, ‘the central defining characteristic of presidentialism has been the separation of legislative from executive powers’. Granting the president law-making authority blurs the difference between the legislative and the executive sphere and law-making powers of the president should therefore not be included among the defining characteristics of presidentialism. Another difference between Shugart and Carey, on the one hand, and Sartori, on the other, is that the former authors argue that the requirement that the chief executive cannot be dismissed by parliament, is insufficient. Instead, they claim that ‘the terms of the chief executive and assembly…are not contingent on mutual confidence’ (Shugart & Carey, 1992, 19 (my italics). Sartori (1997, 87) again takes a different stand and does not consider the power to dissolve parliament a factor that alone would transform a presidential system into something else. However, in agreement with Karvonen (2008, 52), I would regard the requirement that the president is unable to dissolve the legislature a central element of presidentialism. Again, the central doctrine underlying the presidential form of government is the one of separation of powers; the executive sphere is to be separated from the legislative sphere and it is therefore essential that none of the spheres have the power to eliminate the other. The popular election of the president is routinely included among the defining features of presidentialism. In Sartori’s (1997, 83) words the popular election of the president ‘is doubtless a necessary defining condition [for a presidential system]’. On the one hand, it is easy to understand the logic behind this way of reasoning; the chief executive derives his or her legitimacy from the people and is responsible directly to the people.

34

C. ANCKAR

On the other hand, however, most authors seem to be quite willing to budge from a strict definition of popular elections when they discuss presidential systems. The most obvious exception from the rule is the United States, where the president is elected by an electoral college (or, in cases where no candidate receives a majority vote in the electoral college, by the House of Representatives). Another dubious case was Bolivia until 2009, where the president was elected by parliament in cases where no candidate received a majority of the popular vote. Nevertheless, both USA and Bolivia have generally been classified as presidential systems, and the former one has even been regarded as the prototype of presidentialism. Sartori’s (1997, 85) rather flippant argument for including countries with electoral colleges in the presidential category is that ‘the established practice … is to elect the candidate that has obtained the popular relative majority’. The case of Bolivia is more problematic in this respect since its parliament often chose another candidate than the one who received a plurality of the popular vote, and Sartori (1997, 85) accordingly described it as ‘a dubious direct-like case’. Fitting USA (and other historical cases with electoral colleges) and Bolivia into the category of presidential systems is probably not a controversial decision. However, things get more complicated when we consider countries where a president elected by parliament leads and directs the executive at the same time as the legislative and executive spheres are separated. These systems, sometimes labelled assembly-independent (Samuels & Shugart 2010, 27–28), are presidential in all other respects than with regard to the popular election of the head of government. It is important to point out that such systems exist not only in theory; currently, the Federated States of Micronesia, Suriname, and Switzerland make use of exactly this constitutional arrangement. Cheibub et al. (2010, 80) offer a simple ‘solution’ to the dilemma as they define as presidential all democratic systems in which the government cannot be removed from office by parliament. However, as we shall see momentarily, their decision to completely disregard the popular election when defining presidentialism stands in sharp contrast to the criteria they operate with when defining semi-presidentialism. Parliamentaris m is often regarded as the opposite of presidentialism. A classic definition by Verney (1959) listed not less than eleven criteria of parliamentarism, but Verney himself admitted that some of these were characteristics of British parliamentarism and not parliamentarism per se

2

CLASSIFYING POLITICAL REGIMES

35

(Lijphart, 1992, 5). A better point of departure is Lijphart’s (1992, 2– 4) definition of parliamentarism which encompasses only three central characteristics: The head of government…and his or her cabinet are dependent on the confidence of the legislature and can be dismissed from office by a legislative vote of no confidence or censure; [P]rime ministers are selected by the legislature; [P]arliamentary systems have collective or collegial executives

Evidently, the first of these criteria is the most important one. The fact that parliament can oust the government of power is the sine qua non of parliamentarism. However, although it is a necessary criterion of the parliamentarism it is not a sufficient one, as the subordinate position of government to parliament is also a defining characteristic of the semi-presidential regime. The two other criteria are somewhat vague, and arguably carry less weight. The second criterion emphasizes the fact that the prime minister and his or her government ‘emerges’ from the assembly. However, the process of selecting a prime minister varies substantially among parliamentary democracies. Some countries, like Germany, apply positive parliamentarism, meaning that the government needs to win a vote of investiture before it can take office, whereas other countries, for instance the United Kingdom make use of negative parliamentarism, meaning that no vote of investiture is needed and the government is deemed to have the support of parliament until it loses a vote of no confidence. The third criterion, again, stresses the principle that the government is a collegial body and no actor is completely dominating. However, here too, the position of the prime minister varies heavily across the parliamentary systems and is, for instance, generally stronger in one-party governments than in coalition governments. It is, however, necessary to complement the list of criteria with one additional feature of parliamentarism, originally suggested by Budge, Newton et al. (1997, 238), namely that there is ‘no popularly elected president with real political powers’. This criterion is important as it separates parliamentary systems from semi-presidential ones. At the same time, any criterion which aims to separate semi-presidentialism from parliamentarism is likely to be controversial, as scholars have had a very difficult time to reach consensus on how semi-presidentialism should be defined. I shall return to this question below, but suffice it here to say that the bulk

36

C. ANCKAR

of authors writing in the field of semi-presidentialism would probably be of the opinion that a popularly elected president alone would make a parliamentary system semi-presidential, regardless of how much powers the president possessed. Such a definition is not, however, unproblematic. The third democratic regime form is semi-presidentialism. There is far from any consensus on how semi-presidentialism should be defined and some authors even reject the concept altogether. Scholars are in agreement of some criteria, notably that (1) the head of state does not inherit his or her position, (2) the position of head of state is separated from the position of head of government, and (3) the latter (along with the government) can be removed by a parliamentary vote of no confidence. Most, but not all, authors would also add the requirement that the head of state is popularly elected, directly or indirectly through an electoral college (e.g. Duverger, 1980, 166; Elgie, 2011, 3; Hague et al., 1998, 212; Pasquino, 1997, 130; Sartori, 1997, 131; Schleiter & MorganJones, 2010, 1418; Sedelius, 2006, 36; Shugart, 2005; Shugart & Carey, 1992, 23–27; Suleiman, 1994; Wu, 2011, 21). Other defining characteristics have been more controversial. Sartori (1997, 131) has suggested that in order for a system to be semipresidential there should be a ‘dual authority structure’, which ‘allows for different balances and also for shifting prevalences of power within the executive, under the strict condition that the “autonomy potential” of each component unit of the executive does subsist’. This criterion, however, is somewhat ambiguous; it could very well be argued that different balances and shifting prevalences are natural consequences of executive power sharing rather than defining elements of semipresidentialism. In fact, different balances and shifting prevalences are possible in any parliamentary system with coalition governments, where the prime minister must cooperate with leaders of the coalition partners (also Shugart 2005, 327). It is nevertheless fair to say that the difficulties in finding consensus on how semi-presidentialism should be defined have concerned two central aspects, namely how the president is elected and how much powers the president possesses. In the following part, I shall dwell more deeply on the these two criteria, both with regard to their importance for defining semi-presidentialism in general, but also how they can be used in order to classify dual executives in a broader sense.

2

2.3.1

CLASSIFYING POLITICAL REGIMES

37

Semi-Presidentialism and Presidential Powers

The biggest controversy when defining semi-presidenitalism has to do with the extent to which power is shared between the president and the prime minister. Maurice Duverger proposed three defining elements of semi-presidentialism: ‘1) the president… is elected by universal suffrage; 2) he possesses quite considerable powers; 3) he has opposite him …a prime minister and ministers who possess executive and governmental power and can stay in office only if the parliament does not show its opposition to them…’. The problem with this definition is that it is a highly problematic to determine when the powers of the president are ‘quite considerable’, or, as Elgie (2011, 3) expressed it: ‘[i]f the concept of semi-presidentialism is based on a subjective judgment about the power of the president and/or prime minister, then each scholar has free rein to decide what constitutes a fairly powerful president’. Succeeding authors had little success in reaching an understanding of how much power the president must possess in order for a system to be semi-presidential. It is therefore not surprising that the population of semi-presidential countries could differ considerably between authors. To take an extreme example: whereas Elgie (2011, 18–19) found no less than 51 countries which met his requirements for semi-presidentialism, Stepan and Skach (1993, 5, 9), at the time of their writing, were able to identify only two, namely France and Portugal. As a ‘solution’ to the problem, Shugart and Carey (1992, 23– 25) proposed that semi-presidentialism should be substituted by the categories, premier-presidentialism and president-parliamentarism. The common denominator for the two categories was the popular election of the head of state, which distinguished the two variants from parliamentary systems. The difference between premier-presidentialism and presidentparliamentarism lies in the power balance between the prime minister and the president. The former category implicates ‘the primacy of the premier’ whereas the latter ‘the primacy of the president’ (Shugart & Carey, 1992, 24), the crucial thing being that in the latter category, the president can dismiss the cabinet (Shugart & Carey, 1992, 24). A decade later, Alan Siaroff (2003, 305–309) reasoned along the same lines, suggesting that the concept of semi-presidentialism should be rejected altogether. Instead, he argued, we should stick to the categories presidentialism and parliamentarism, where the latter category is split up into the following subcategories: parliamentary systems with presidential

38

C. ANCKAR

dominance, parliamentary systems with a presidential corrective, parliamentary systems with figurehead presidents, and parliamentary systems with figurehead monarchs. His main motivation was that presidential powers vary to a very high extent among the countries which meet the generally accepted criteria of semi-presidentialism. Another way of dealing with the problem is to disregard the power dimension altogether and focus exclusively on whether the president is popularly elected or not. The ‘father’ of this strategy, Robert Elgie (2011, 3) defined semi-presidentialism as ‘the situation where there is both a directly elected fixed-term president and a prime minister and cabinet who are collectively responsible to the legislature’. Cheibub et al. (2010, 81) stick to this definition in their well-known dataset, where democratic countries where the government is responsible to parliament are coded as semi-presidential if the head of state is popularly elected for a fixed term in office and parliamentary if this is not the case. In all, Elgie’s conception of semi-presidentialism has become increasingly popular in recent years and is no doubt the dominating one in the field. Disregarding the power dimension altogether undoubtedly makes classifications easier; as we operate with indicators that are easily interpretable, the task of reaching consensus on the population of semi-presidential countries is facilitated. There are, however, drawbacks with Elgie’s definition. Since he does not pay regard to the power dimension, the population of semi-presidential countries becomes rather heterogeneous in terms of presidential prerogatives (e.g. Åberg & Sedelius, 2020, 1112). As a consequence hereof, the distinction between parliamentary and semipresidential systems becomes blurred. The crucial question is whether or not it makes sense to separate two democratic forms of government from each other simply on the basis on how a powerless head of state is (s)elected. The case of Slovakia is illustrative in this respect. In 1999, the country introduced the popular election of the president, thus removing it from the category parliamentary systems to semi-presidential ones, if we were to apply Elgie’s definition. However, at the same time the powers of the president were reduced and ‘his ability to intervene in the everyday workings of parliament and government … decreased significantly’ (Malová & Rybáˇr, 2000, 184). In a similar vein, both the Czech Republic (2012) and Moldova (2016) have recently introduced the popular election of the president without increasing the powers of the president. It is therefore worth pointing out that there are, indeed,

2

CLASSIFYING POLITICAL REGIMES

39

authors who do not share Elgie’s view that the popular election alone is crucial for separating semi-presidentialism from parliamentarism (e.g. Anckar, 2019; Brunclík & Kubát, 2016, 14–16; 2019; Hloušek, 2014). In a corresponding manner we can ask ourselves what happens in situations where the powers of the president are very broad in comparison with the powers of the prime minister. Does this turn a semi-presidential system into a presidential? Some authors, notably Tavits (2009, 24–25), appear to answer in the affirmative. She classifies Peru and Sri Lanka as presidential despite the fact that the prime minister and government can be brought down by a parliamentary vote of no confidence. The decision is based on the fact that the president, not the prime minister, functions as the head of government. I would, however, take a different view on the matter and answer the question in the negative. As long as there is a prime minister and a government which can be removed from office by a parliamentary vote of no confidence we have not crossed the line between semi-presidentialism and presidentialism. Presidentialism is characterized not only (or even primarily) by a powerful president, but also by a strict application of the separation of powers doctrine; in a presidential system the legislature cannot bring down the government, and the government cannot dissolve the legislature. Once we give the legislature the power to bring down the government, the system is no longer presidential. The fact that the president him/herself cannot be brought down is of subordinate relevance. 2.3.2

Semi-Presidentialism and the Popular Election of the President

The popular election of the president is routinely included among the defining features of semi-presidentialism. However, not all authors agree that it constitutes a defining element of semi-presidentialism and therefore include among semi-presidential countries also regimes where the president is elected by the legislature (e.g. O’Neil, 1993; Anckar, 1999, 256–257 and, somewhat less explicitly, Müller et al., 2003 as well as Beliaev, 2006). The main argument for elevating the popular election of the president to the position of a necessary fundament of semi-presidentialism is based on a principal-agent framework. Samuels and Shugart (2010, 25–34; also Shugart & Carey, 1992) classify democratic regimes according to two dimensions, executive origin and executive survival. Countries where the executive originates from the assembly and is dependent on the assembly

40

C. ANCKAR

for its survival are parliamentary, whereas countries where the executive originates from the electorate (i.e. by popular election) and cannot be removed by the assembly are presidential. In countries with shared executives, the situation is more complicated. For Samuels and Shugart (2010, 29), the central aspect, or, as they put it ‘[t]he key’, of semi-presidentialism is that the president and prime minister originate from separate sources, the president from the electorate and the prime minister from the assembly. Moreover, the two actors are dependent on different sources for their survival in office; the president is elected for a fixed term and cannot be removed by a parliamentary vote of no confidence, whereas the prime minister can be removed by the assembly at any time. As noted above, the two dimensions also make it possible to divide semi-presidential systems into two subtypes, premier-presidentialism and president-parliamentarism. In both categories, the prime minister, but not the president, originates from the assembly. However, in countries situated in the former category, the prime minister and cabinet are collectively responsible exclusively to the assembly. In other words, the president cannot dismiss the prime minister and the government. In countries placed in the latter category, however, the president has the ability to dismiss the prime minister and the government, and they are thus responsible to both the assembly and the president (Shugart & Carey, 1992). Many authors seem to agree with this principal-agent framework (e.g. Åberg & Sedelius, 2020; Schleiter & Morgan-Jones, 2009a, 2009b; Sedelius & Linde, 2018). There is one unassailable advantage with the approach: focusing on the separate origin and/or survival of the executive provides a justification for regarding semi-presidentialism as a unique regime form, rather than a hybrid, situated somewhere along a continuum from parliamentarism to presidentialism. The counterargument is that although the distinction based on origin and survival of the executive makes sense from a theoretical point of view, it is less clear if it has practical implications. In line with Elgie, Samuels and Shugart (2010) disregard the power dimension completely when they identify the population of semi-presidential systems; the inclusion in the category is based only on whether the president is popularly elected or not. The power dimension then becomes relevant for conferring the cases to the categories premier-presidentialism and president-parliamentarism. The consequence of this strategy is that a country with an indirectly (not popularly) elected president who has powers to appoint or dismiss

2

CLASSIFYING POLITICAL REGIMES

41

the prime minister, is in total control of foreign policy, can dismiss the parliament, and veto legislation would be placed in the category parliamentary system. In comparison, a country where the president is popularly elected, but only possesses ceremonial powers comparable to those of the more or less completely powerless Swedish King, is considered semi-presidential. Certainly, such a definition of semi-presidentialism is very far from Duverger’s original conception of the term. Furthermore, it should also be pointed out that although Duverger included the popular election of the president as a defining characteristic of semipresidentialism, he did not apply the criterion rigorously; very little in his article suggests that the period between 1958 and 1962 (i.e. prior to the constitutional amendment that introduced the popular election of the president) in France should be regarded as a completely different regime form than the one in place after 1962. Some authors, therefore, take the view that when classifying systems as semi-presidential it is irrelevant whether the president is elected by parliament, by an electoral college, or by popular vote; the crucial aspect is that the president, in contrast to the prime minister, cannot be brought down by parliament. The case of Finland is illustrative in this respect. There is more or less consensus in the literature that Finland had a semi-presidential form of government during the period 1919–2000 (e.g. Arter, 1985; Duverger 1980; Elgie, 2011, 82–90; Karvonen, 2008, 57– 58; Paloheimo, 2003; Sartori 1997). However, when conferring Finland to the category of semi-presidential systems the authors overlook the fact that four times during the period in question Finnish presidents were put in their position not by popular vote but by a vote in the Finnish parliament. The case of Urho Kekkonen is of particular interest. In 1973, his term as president was extended for a period of four years by a parliamentary law of exception. During the following four years, the Finnish president stood at the height of his career, being in complete control of the executive sphere and changing the government on no less than four occasions. One could hardly think of a more fitting example of a semipresidential system, except, of course, for the fact that the president had not been chosen by popular vote. The rigid application of the popular election of the president as a defining characteristic of semi-presidentialism becomes particularly strange when we take a look at how purely presidential systems are classified. As we have seen, although the popular election of the president is routinely included among the defining features of presidentialism

42

C. ANCKAR

many authors seem to be quite willing to budge from a strict definition of popular elections when they define presidential systems. Few would claim that USA and Bolivia until 2009, where presidents occasionally have been elected by parliament, were not presidential during the years a nonpopularly elected president was in office. In Switzerland, Micronesia, and Suriname the parliament elects the president, who leads and directs the executive, and where the legislative and executive spheres are separated. It is not overly controversial to refer these countries to the category of presidential systems (e.g. Cheibub et al. [2010, 80]. Anyway, it is a quite dubious strategy to apply a stricter application of popular election of the president for semi-presidential systems than for presidential systems (also Anckar, 2019). 2.3.3

Semi-constitutional monarchism

Although the classification of democratic regimes into the categories presidentialism, parliamentarism, and semi-presidentialism is standard in the comparative politics literature, it overlooks an important regime variant, namely systems where a hereditary monarch shares executive powers with a prime minister, accountable to parliament. Such systems function much in the same way as semi-presidential systems, except for the fact that the head of state inherits his or her position. These systems are routinely included among parliamentary systems, but such a classification does not pay justice to the fact that the powers of the head of state are far from always ceremonial only. In line with the terminology suggested by Corbett et al. (2017), I label the category semi-constitutional monarchies. The authors define the category as ‘systems in which the actions of monarchs are circumscribed by a constitution, but in which monarchs, as independent and autonomous political actors, nonetheless have the capacity to exert a large measure of political influence’ Corbett et al. (2017, 691). It is necessary to point out that within the framework of the present book, this category refers exclusively to democratic countries. In other words, it does not include autocratic countries where a monarch shares power with, for instance, a military regime or the ruling party in a single-party state. Drawing a line between semi-constitutional monarchies and parliamentary systems with a monarch as head of state is not easy. According to the definition by Corbett et al. (2017, 691), the monarchs should ‘exert a large measure of political influence’, but establishing how much

2

CLASSIFYING POLITICAL REGIMES

43

influence they should have is, of course, difficult. In this respect, we encounter the same problem as with regard to Duverger’s definition of semi-presidentialism discussed above. However, within the category of semi-constitutional monarchies the powers of the monarch cannot be very large since, in contrast to a president in a semi-presidential system, the monarch has no democratic legitimacy. System with very powerful monarchs would therefore not qualify as democracies, whereas this restriction does not apply to countries with presidents with a democratic basis of legitimacy. Corbett et al. (2017) do not provide a list of countries that meet the criteria of inclusion in the category of semi-constitutional monarchies, but they discuss three such cases, namely Bhutan, Liechtenstein, and Tonga. Whereas Bhutan and Tonga have democratized quite recently, Liechtenstein has been classified as a democracy for many decades by most reputable sources and could perhaps, much like France with regard to semi-presidentialism, be regarded as the prototype of semi-constitutional monarchism. According to the constitution of Liechtenstein, the Prince Regnant appoints the government, which must enjoy the confidence of both the legislature and the Prince Regnant (art. 79, 80). The hereditary monarch also has the power to dissolve parliament (art. 41) and veto powers (art. 65). The powers of the monarch of Liechtenstein are quite extensive, though, and the threshold for inclusion in the category of semiconstitutional monarchies should evidently be set at a lower value along the power scale. Elsewhere (Anckar, 2021) I have suggested that a country could be included in the category if the monarch holds either any executive powers, any legislative powers, powers over domestic policy, or powers to dissolve the legislature. If such a low threshold is applied, approximately fifteen countries have met the criteria of semiconstitutional monarchism over the last century (Anckar, 2021). Nevertheless, no matter how you slice it, regarding both semipresidential and semi-constitutional monarchic regimes, we confront the problem of drawing a line between systems with weak and strong heads of state. As we recall, with regard to semi-presidentialism, an increasingly popular strategy has become to disregard the power dimension completely and only focus on whether the president is popularly elected or not. In principle, the same solution could be applicable for separating parliamentary systems from semi-monarchic systems.

44

C. ANCKAR

Recall that the motivation for defining semi-presidentialism only with reference to the mode of selection of the head of state, was based on a principle-agent perspective, emphasizing the fact that the two branches of the executive had different sources of origin and survival. The discussion of the separate sources of origin and survival of the executive conducted in conjunction with the semi-presidential regime form is of relevance also with regard to semi-constitutional monarchies. Like in semi-presidential systems, the prime minister emerges from the parliament and can be ousted of office by the same institution. The monarch, again, inherits his or her position as head of state and occupies the position until his or her death, meaning that the legitimacy of his or her rule ‘relies on a God-given natural/historical law’ (Kailitz, 2013, 43). If we were to follow this logic, we would disregard the powers of the monarchs as well, and classify as semi-constitutional monarchies all parliamentary democracies with a hereditary ruler. The term parliamentary systems would thus be reserved for democratic systems in which the head of state emerges from (and, in some cases, also is responsible to) the legislature, i.e. in practice systems with a president elected by parliament. Yet, to my knowledge, nobody has made the suggestion that such a classification would make any sense.

2.4 Conclusion: The Difficult Art of Classifying Regimes So where are we left? As is evident from the present chapter, there is far from any consensus on how political regimes should be classified. This is particularly the case with authoritarian regimes—where there is disagreement not only on which categories to apply but also regarding the definitional aspects of ‘regime’. Here, I have introduced a classification that I consider reasonable, but I am equally aware of the fact that far from all other scholars are likely to agree. On the other hand, one should not exaggerate the problems of defining and classifying autocratic regimes. For one thing, not all regime forms are controversial. Accordingly, different datasets tend to agree strongly on the population of monarchic regimes, admittedly the least controversial one to reach consensus on (Gerring et al., 2021: 22–23). Furthermore, if similar patterns of regime transformations can be observed for several controversial regime categories, the criteria for separating these regime forms from

2

CLASSIFYING POLITICAL REGIMES

45

each become less relevant, at least with reference to the present research effort. Among countries that meet the criteria of democracy, again, authors have traditionally differed widely on the criteria of inclusion in the categories parliamentarism and semi-presidentialism. The present contribution does not aim to solve the issue of how these regimes should be classified. The question whether the current regimes of Portugal, Austria, Iceland, and the Czech Republic (with weak, popularly elected presidents), or Mauritius, the historical cases of Finland 1919–1925, Germany 1919–1925, and France 1958–1962 (with strong, non-popularly elected presidents), are semi-presidential or parliamentary is dependent on the weight authors prescribe to two central features, namely the mode of selection and the powers of the head of state. Although authors can disagree on which of the dimensions that carry more importance for regime classifications, hardly anybody would dispute the fact that both the mode of selection and the powers of the head of state are essential for the functioning of the regimes in question. The present work, aims to disentangle such patterns in a broad, historical comparative perspective. Luckily, such an endeavour can be accomplished without taking a stand on the issue where the line between semi-presidentialism and parliamentarism is situated.

Notes 1. For instance, in comparable years Angola, Cuba, Eritrea, Ethiopia, and Mozambique are coded as rebel regimes by Hadenius et al. but not classified as military regimes by Geddes et al. 2. In this respect, particularly the following questions have been used: HOS/HOG appointment in practice (v2expathhs, v2expathhg): How did the head of state reach office/How did the head of government gain access to office?; HOS/HOG removal by other in practice (v2exrmhsol, v2exrmhgnp): Which of the following bodies would be likely to succeed in removing the head of state/head of government if it took actions (short of military force) to do so?; and HOS/HOG control over (c) (v2exctlhs, v2exctlhg): In practice, from which of the following bodies must/does the head of state/head of government customarily seek approval prior to making important decisions on domestic policy?

46

C. ANCKAR

References Åberg, J., & Sedelius, T. (2020). A structured review of semi-presidential studies: Debates, results and missing pieces. British Journal of Political Science, 50(3), 1111–1136. Alvarez, M., Cheibub, J. A., Limongi, F., & Przeworski, A. (1996). Classifying political regimes. Studies in Comparative International Development, 31(2), 3–36. Anckar, C. (2019). Are powerful presidents always popularly elected? Implications for separating semi-presidentialism from parliamentarism. Democratization, 26(2), 327–342. Anckar, C. (2021). Constitutional monarchies and semi-constitutional monarchies: A global historical study, 1800–2017. Contemporary Politics, 27 (1), 23–40. Anckar, C., & Fredriksson, C. (2019). Classifying political regimes 1800–2016: A typology and a new dataset. European Political Science, 18(1), 84–96. Anckar, D. (1999). Finland inför millenniumskiftet: en regimbestämning. Statsvetenskaplig tidskrift, 102(3), 241–261. Arter, D. (1985). Government in Finland: A “semi-presidential system”? Parliamentary Affairs, 38(4), 472–495. Bagehot, W. (1872). The English Constitution (new 2). H. S. King. Barros, R. (2016). On the outside looking in: Secrecy and the study of authoritarian regimes. Social Science Quarterly, 97 (4), 953–973. Beliaev, M. V. (2006). Presidential powers and consolidation of new Postcommunist democracies. Comparative Political Studies, 39(3), 375–398. Bell, C. (2016). The Rulers, Elections, and Irregular Governance Dataset (REIGN). OEF Research. Available at oefresearch.org Boix, C., Miller, M., & Rosato, S. (2013). A complete data set of political regimes 1800–2007. Comparative Political Studies, 46, 1523–1554. Brooker, P. (2000). Non-democratic regimes: Theory, government and politics. St. Martin’s. Brooker, P. (2013). Non-democratic regimes (3rd ed.). Bloomsbury Publishing. Brunclík, M., & Kubát, M. (2016). The Czech Parliamentary regime after 1989: Origins, Developments and Challenges. Acta Politologica, 8(2), 5–29. Brunclík, M., & Kubát, M. (2019). Semi-presidentialism, parliamentarism and presidents: Presidential politics in Central Europe. Routledge. Budge, I. & Newton, K., et al. (1997). The politics of the New Europe: Atlantic to Urals. Longman. Cheibub, J. A., Gandhi, J. & Vreeland, J. R. (2009). Democracy and dictatorship revisited codebook. http://sites.google.com/site/joseantoniocheibub/dat asets/democracy-and-dictatorship-revisited Cheibub, J. A., Gandhi, J., & Vreeland, J. R. (2010). Democracy and dictatorship revisited. Public Choice, 143(1), 67–101.

2

CLASSIFYING POLITICAL REGIMES

47

Coppedge, M., Gerring, J., Knutsen, C. H., Lindberg, Teorell, J., Altman, D., Bernhard, M., Fish, M. S., Glynn, A., Hicken, A., Lührmann, A., Marquardt, McMann, K., Paxton, P., Pemstein, D., Seim, B., Sigman, R., Skaaning, S.-E., Staton, J., Cornell, A., Gastaldi, L., Gjerløw, H., Mechkova, V., von Römer, J., Sundström, A., Tzelgov, E., Uberti, L., Wang, Y-t., Wig, T., & Ziblatt, D. (2020). V-Dem codebook v10. Varieties of Democracy (VDem) Project. Corbett, J., Veenendaal, W., & Ugyel, L. (2017). Why monarchy persists in small states: The cases of Tonga, Bhutan, and Liechtenstein. Democratization, 24(4), 689–706. Dahl, R. (1971). Polyarchy: Participation and opposition. Yale University Press. Diamond, L. (2002). Thinking about hybrid regimes. Journal of Democracy, 13(2), 21–35. Duverger, M. (1980). A new political system model: Semi-presidential government. European Journal of Political Research, 8(2), 165–187. Elgie, R. (2011). Semi-presidentialism: An increasingly common constitutional choice. In R. Elgie, S. Moestrup, & Y. S. Wu (Eds.), Semi-presidentialism and democracy (pp. 1–20). Palgrave Macmillan. Geddes, B. (1999). What do we know about democratization after twenty years? Annual Review of Political Science, 2(1), 115–144. Geddes, B. (2003). Paradigms and sand castles: Theory building and research design in comparative politics. University of Michigan Press. Geddes, B., Wright, J., & Frantz, E. (2014). Autocratic regimes and transitions. Perspectives on Politics, 12(2), 313–331. Gerring, J., Wig, T., Veenendaal, W., Weitzel, D., Teorell, J., & Kikuta, K. (2021). Why monarchy? The rise and demise of a regime type. Comparative Political Studies, 54(3–4), 585–622. Hadenius, A. (1992). Democracy and development. Cambridge University Press. Hadenius, A., & Teorell, J. (2007). Pathways from authoritarianism. Journal of Democracy, 18(1), 143–157. Hadenius, A., Teorell, J., & Wahman, M. (2012). Authoritarian regimes dataset, version 5.0. Lund University. http://www2.srv.svet.lu.se/ARD Hague, R., Harrop, M., & Breslin, S. (1998). Comparative government and politics. An introduction (4th ed.). Macmillan. Herb, M. (1999). All in the family: Absolutism, revolution, and democracy in middle eastern monarchies. Suny Press. Hloušek, V. (2014). Is the Czech Republic on its way to semi-presidentialism? Baltic Journal of Law & Politics, 7 (2), 95–118. Huntington, S. P. (1991). The third wave: Democratization in the late twentieth century. University of Oklahoma Press. Kailitz, S. (2013). Classifying political regimes revisited: Legitimation and durability. Democratization, 20(1), 39–60.

48

C. ANCKAR

Karvonen, L. (2008). Statsskick: Att bygga demokrati, andra upplagan. SNS förlag. Knutsen, C. H., Møller, J., & Skaaning, S.-E. (2016). Going historical: Measuring democraticness before the age of mass democracy. International Political Science Review, 37 (5), 679–689. Lijphart, A. (1992). Introduction. In A. Lijphart (Ed.), Parliamentary versus presidential government. Oxford University Press. Linz, J. (2000). Totalitarian and authoritarian regimes. Lynne Rienner. Linz, J., & Stepan, A. (1996). Problems of democratic transition and consolidation: Southern Europe, South America, and Post-Communist Europe. Johns Hopkins University Press. Malová, D., & Rybáˇr, M. (2000). Slovakia’s Presidency: Consolidating democracy by curbing ambiguous powers. In R. Elgie & S. Moestrup (Eds.), Semipresidentialism in Central and Eastern Europe (pp. 180–200). Manchester University Press. Müller, W. C., Bergman, T., & Strøm, K. (2003). Parliamentary democracy: Promise and problems. In K. Strøm, W. C. Müller, & T. Bergman (Eds.), Delegation and accountability in parliamentary democracies. Oxford University Press. Nordlinger, E. (1977). Soldiers in politics: Military coups and government. Prentice-Hall. O’Neil, P. (1993). Presidential power in Post-Communist Europe: The Hungarian case in comparative perspective. Journal of Communist Studies, 9(3), 177–201. Paloheimo, H. (2003). The rising power of the Prime Minister in Finland. Scandinavian Political Studies, 26, 219–243. Pasquino, G. (1997). Semi-presidentialism: A political model at work. European Journal of Political Research, 31(1), 128–137. Roller, E. (2013). Comparing the performance of autocracies: Issues in measuring types of autocratic regimes and performance. Contemporary Politics, 19(1), 35–54. Samuels, D. J., & Shugart, M. S. (2010). Presidents, parties, and prime ministers. Cambridge University Press. Sartori, G. (1997). Comparative constitutional engineering: An inquiry into structures, incentives and outcomes (2nd ed.). Macmillan. Schleiter, P., & Morgan-Jones, E. (2009a). Citizens, presidents, and assemblies: The study of semi-presidentialism beyond Duverger and Linz. British Journal of Political Science, 39(4), 871–892. Schleiter, P., & Morgan-Jones, E. (2009b). Party government in Europe? Parliamentary and semi-presidential democracies compared. European Journal of Political Research, 48(5), 665–693.

2

CLASSIFYING POLITICAL REGIMES

49

Schleiter, P., & Morgan-Jones, E. (2010). Who’s in charge? Presidents, assemblies, and the political control of semipresidential cabinets. Comparative Political Studies, 43(11), 1415–1411. Sedelius, T. (2006). The Tug-of-War between presidents and prime ministers: Semi presidentialism in Central and Eastern Europe. Örebro University Press. Sedelius, T., & Linde, J. (2018). Unravelling semi-presidentialism: Democracy and government performance under four distinct regime types. Democratization, 25(1), 136–157. Shugart, M. S. (2005). Semi-presidential systems: Dual executive and mixed authority patterns. French Politics, 3(3), 323–351. Shugart, M. S., & Carey, J. M. (1992). Presidents and assemblies: Constitutional design and electoral dynamics. Cambridge University Press. Siaroff, A. (2003). Comparative presidencies: The inadequacy of the presidential, semi presidential and parliamentary distinction. European Journal of Political Research, 42(3), 287–312. Stepan, A., & Skach, C. (1993). Constitutional frameworks and democratic consolidation: Parlamentarism versus presidentialism. World Politics, 46(1), 1–22. Suleiman, E. N. (1994). Presidentialism and political stability in France. In J. J. Linz & A. Valenzuela (Eds.), The failure of presidential democracy: Comparative perspectives (Vol. 1, pp. 137–162). Johns Hopkins University Press. Tavits, M. (2009). Presidents with prime ministers: Do direct elections matter? Oxford University Press. Vanhanen, T. (2013). Measures of democracy, 1810–2012 (FSD1289, version 6.0). Finnish Social Science Data Archive. Verney, D. (1959). The analysis of political systems. Routledge & Kegan Paul. Wahman, M., Teorell, J., & Hadenius, A. (2013). Authoritarian regime types revisited: Updated data in comparative perspective. Contemporary Politics, 19(1), 19–34. Wu, Y. S. (2011). Clustering of semi-presidentialism: A first cut. In R. Elgie, S. Moestrup, & Y. S. Wu (Eds.), Semi-presidentialism and democracy (pp. 21– 41). Palgrave Macmillan.

CHAPTER 3

The Research Population

3.1

Separating Democracies from Autocracies

The focus of the book is on shared executives in democratic countries. When defining the research population, two central terms need clarification, namely ‘shared’ and ‘democratic’. As the first step in the process of defining the research population requires us to separate authoritarian countries from democratic ones, let me start with the latter one. Here, my point of departure is the dataset by Boix, Miller, and Rosato (2013 ), who make use of a dichotomous qualitative scale, classifying the countries of the world as either democracies or autocracies on a yearly basis for the time period 1800–2015. Following Dahl (1971), their measure encompasses two dimensions, contestation and participation. In terms of the first dimension, Boix et al. (2013, 1530) consider a country to be democratic if two conditions are met: ‘The executive is directly or indirectly elected in popular elections and is responsible either directly to voters or to the legislature [and t]he legislature (or the executive if elected directly) is chosen in free and fair elections’. With regard to the second dimension a country meets the criterion of democracy if ‘[a] majority of adult men has the right to vote’ (Boix et al., 2013, 1530). In the dataset by Boix, Miller, and Rosato the conditions prevailing on the date December 31 determine the classification each year, although in cases where ‘a country becomes autocratic © The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 C. Anckar, Presidents, Monarchs, and Prime Ministers, Palgrave Studies in Presidential Politics, https://doi.org/10.1007/978-3-031-03960-7_3

51

52

C. ANCKAR

and then democratizes within a single year, the year is coded as autocratic’ (Boix et al., 2013, 11). Since the dataset by Boix, Miller, and Rosato does not cover the period 2016–2019, I have, for these four years, confined the countries of the world to the categories democracy or autocracy, applying the following basic strategy: For each country, I have compared scores for the polyarchy-variable in the V-dem dataset as well as scores for Freedom House’s political rights dimension to the values the country in question received in 2015, the last year for which data for Boix, Miller, and Rosato is available, on these same scales. Thereafter I have checked how Boix, Miller, and Rosato have classified the countries during the same point in time. If, for instance, Boix, Miller, and Rosato classified a country as a democracy in 2015 and its value on V-dem’s polyarchy-variable in 2017 equalled or exceeded the corresponding polyarchy-score for 2015, I have classified the country as a democracy for the year 2017. In cases where polyarchy-scores and political rights scores differ, I have also compared the values with the Polity 2 scale in the Polity IV-dataset (Marshall et al., 2014). In cases where countries have received values on Freedom House’s political rights dimension and V-dem’s polyarchy-variable indicating that they have moved away from their initial position during the time period in question, I have made additional comparisons with the values the countries received on the two scales during the time period 2010 to 2015 and thereafter checked how Boix, Miller, and Rosao classified the countries in question during these years. For a comprehensive exemplification of the strategies applied in such cases, the reader is referred to Anckar and Fredriksson (2020). Relying on only one measure of democracy is somewhat problematic, however. We cannot escape the fact that when we confer countries to the categories autocracy and democracy, some classifications tend to be controversial. A particular case in point is Namibia, which Boix, Miller, and Rosato classify as an autocracy during the whole period of its independence. Freedom House, again, has classified it as ‘free’ ever since 1990, and it has received the value 2 on Freedom House’s political rights dimension in each Freedom House survey but one since the same year. On the Polity 2 scale, again, Namibia has received the value 6 ever since 1990. If we were to simply rely on Boix, Miller, and Rosato, Namibia would not be included in the study.

3

THE RESEARCH POPULATION

53

A reasonable strategy is therefore to complement the classification by Boix, Miller, and Rosato with an alternative source. However, the general idea is not so much to introduce another, competing, source of measurement but rather to ensure that no relevant case is left out from the analysis. Moreover, for the purpose of the present study, the dataset by Boix, Miller, and Rosato has a very big advantage compared to quantitative indicators in the sense that it is insensitive to small variations in democracy scores. In comparison, if we apply threshold points along a continuous scale, there are many countries balancing on the threshold between democracy and autocracy. This means that their position as democracy or autocracy can vary from on year to another as a consequence of extremely marginal changes in the indicators. I have therefore opted for strategy where cases that meet quite strict criteria of democracy in an alternative dataset can be added to the research population. Here, I have chosen to complement the index by Boix, Miller, and Rosato with V-dem’s polyarchy and liberal democracy measurements. The polyarchy dimension corresponds roughly with Boix, Miller, and Rosato’s conception of democracy, reflecting the electoral dimension, whereas the liberal democracy index is more comprehensive, adding the dimension of individual and minority rights to the electoral dimension. In order to be included in the population, I demand that a country scores the value 0.4 not only on the polyarchy measure but also on the liberal democracy index.1 This procedure adds some 60 cases to the final research population.2,3

3.2

Separating Dual Executives from Single Executives

Having established which of the countries that meet the criterion of democracy, we are confronted with the task of determining when an executive is shared. The point of departure is to operate with a generous definition of sharing; in essence, all countries where there is a potential for real executive power sharing are eligible for inclusion. This is the case in all countries where the position of head of state is different than that of head of government. We can therefore eliminate all countries where the position of head of state coincides with the position as head of government. This is the case notably in countries with a presidential form of government. In addition to the presidential countries, there are a few cases where the positions as head of state and head of government coincide but where the system is parliamentary. For instance, in both South

54

C. ANCKAR

Africa and Botswana, the president is elected by parliament and can be removed from office subject to a parliamentary vote of no confidence.4 In principle, all remaining systems are of interest for the study. It is evident that by combining a number of important institutional features we can come up with many different regime features, (e.g. Siaroff 2003). We can, for instance, pay regard to if the head of state is elected by plurality or majority vote, if both the head of government and the head of state must enjoy the confidence of parliament, if there is one head of state or more (as is the case in San Marino), if the prime minister is popularly elected (the Israeli experiment in 1996) or not, etc. It is, however, evident that many of these combinations are illogical and/or have few, if any, empirical instances. Instead, it is important to focus on those aspects that are central with regard to the distribution of power between the prime minister and the head of state. One important feature of the dual executive is that the actors who share powers have an independent basis for their exercise of power. This is clearly the case in dual executives where the president is popularly elected. Here the independence of the president vis-à-vis the prime minister and vice versa is primarily manifested by the fact that the two actors base their legitimacy on different institutions (the president on the people) and the prime minister on the parliament (Samuels & Shugart, 2010; Shugart 2005; Shugart & Carey, 1992). However, although the popular election of the president is likely to enhance his or her level of independence in relation to the prime minister, it is not a necessary criterion for inclusion in the population. As pointed out by Tavits (2009, 34) presidents elected by parliaments do not necessarily act as agents of parliament. The central aspects are that the legislature cannot remove the president, who, in his/her turn, has to take into account the partisan constellation in parliament when appointing (and dismissing) the prime minister. Thus, the independence of the actors is achieved when the prime minister is accountable to the legislature at the same time as the president cannot be removed by parliament (except under special circumstances, for instance as a result of an impeachment process). Things get more complicated when we turn to systems where there is a government headed by a prime minister, subject to a parliamentary vote of no confidence and a head of state elected by parliament and removable from office by a parliamentary vote of no confidence. Here, the crucial issue is whether the fate of the head of state is tied to the fate of the

3

THE RESEARCH POPULATION

55

prime minister, and vice versa. If this is indeed the case, the executive is not shared even if very mild criteria of sharing are applied. Such cases are therefore not included in the population. However, there are also cases where a parliamentary vote of no confidence to either the head of state or the head of government does not automatically lead to the resignation of the other part of the dual executive. Such cases are included in the study. Needless to say, these systems are extremely rare, but they have existed for short periods of time notably in former communist systems while the new democratic constitutions were being drafted. The case of Belarus is illustrative. The country surpassed the threshold of democracy in 1991 and at that time, the Chairman of the Supreme Council of Belarus acted as head of state. The position of head of state was held by Stanislav Shushkevich, who had been elected to this position by the parliament whereas the government was led by Prime Minister Vyacheslav Kebich. On 26 January 1994, the parliament gave a vote of no confidence to Shushkevich, who was replaced by Myechyslaw Hryb. However, Prime Minister Kebich continued in office until the first President of Belarus, Alexander Lukashenko took office in July the same year. Finally, a large part of the democracies where the position of prime minister is separated from the position of head of state are monarchies. These countries resemble countries with popularly elected presidents in the sense that the origin and survival of the two actors are based on different institutions. In most of the cases, the prime minister emerges from and is responsible only to the legislature. However, in some of the cases, the prime minister is also accountable to the monarch. The legitimacy of the monarch, again, is based on the divine right of kingship; the monarch inherits his or her position as head of state and retains the position for life. All dual executives with a monarch as head of state are therefore eligible for inclusion in the population. Let us further consider systems where a prime minister, who is not responsible to parliament, exists alongside the president, which is the case for instance in South Korea. These countries are excluded from the population. Since the prime minister’s position is completely dependent on the president, the executive is not dual and the system functions like a presidential system. There are also a number of systems which, in essence, are presidential but apply collegial presidencies. These cases are trickier. I have, however, chosen to exclude them from the population. For one thing, the systems are highly heterogeneous, meaning that they

56

C. ANCKAR

do not constitute a natural category for comparisons with other variants of dual executives. Second, as all other countries in the population function on the basis of parliamentarism, including a very small and heterogeneous group of countries where the parliamentary mechanism is absent makes little sense if the ambition is to come up with generalizations. These considerations disqualify Switzerland, with its rotating presidency and Uruguay’s collective executive (1919–1933) from inclusion in the population. However, collegial presidencies are included in the population provided that the prime minister/central ministers are subject to a parliamentary vote of no confidence. Such an arrangement is currently in use in Bosnia and Hercegovina, but have also occurred in other parts for short periods, for instance in Sudan 1956–1958. In the following part, I split up the population of dual executives with regard to the two central components discussed in Chapter 2, i.e. the mode of selection of the head of state and the powers of the head of state. As will become evident, the first task is not a difficult one, whereas the second is somewhat more challenging.

3.3

The Selection of the Head of State

In democracies, there are three ways a person can reach the position as head of state, namely by popular election (directly or indirectly), by nonpopular election or by inheritance. At first sight, the third category is the most uncontroversial one; hereditary succession is generally regarded as the defining element of a monarchic system. Yet, the matter is not completely unproblematic; as we have noted, quite a few authoritarian countries currently regarded as monarchies (for instance Kuwait, Oman, and Saudi Arabia) do not practice primogeniture. Instead, the process of selecting a monarch is in the hands of the royal family. There are also leaders, for instance Napoleon Bonaparte in France (1804), Ahmet Zogu in Albania (1928), and Bokassa in Central African Republic (1976), who proclaim themselves monarchs and introduce hereditary succession to the throne (although none of the above-mentioned ‘monarchs’ were succeeded by their offspring). In the study at hand, where only democracies are included, we escape most of these difficulties of delimitation. The only somewhat dubious cases concern countries where a new monarchical regime or a monarch has been installed by parliament, or, as was the case of Greece in 1935, by popular referendum. Such cases are here included among the other

3

THE RESEARCH POPULATION

57

monarchies provided that two conditions are met: (1) The persons who have received their position as head of state by means of such measures have belonged to a royal family (which means that the group of plausible candidates is extremely limited) and (2) The instalment of a new monarch establishes a line of inheritance, where descendants of the monarch are expected to inherit the crown. Drawing a line between popularly and non-popularly elected presidents is somewhat more problematic. In line with most of the authors in the field, I find it reasonable to include presidents elected by an electoral college among the popularly elected presidents. However, it is important that the electoral college is elected specifically for the purpose of electing a president (Samuels and Shugart 2005, 4; Sartori 1997). Conveniently, this classification is also applied in the V-dem dataset, which contains a question if the head of state is popularly elected or not. This question has been used as the main source for the classification of the cases into the separate categories of selection mode of the head of state.

3.4

The Powers of the Head of State

Let me begin by admitting the obvious: Measuring presidential or monarchical powers is a difficult task, which can be accomplished in many ways (Beliaev, 2006; Doyle & Elgie 2016; Duverger 1978; Fortin, 2013; Frye, 1997; Metcalf, 2000; Roper, 2002; Shugart & Carey, 1992, 148–158). Some years ago, Doyle and Elgie (2016, 732–733) identified nineteen different original measures of presidential powers and made the observations that correlations between the measures often were quite low and that data tended to be compiled for limited regional settings and rarely updated. Furthermore, with reference to (Fortin, 2013), they noted that there was a problem of validity, as indices of presidential power tended to be neither unimodal nor unidimensional. When constructing indices of presidential powers, the most pertinent question is whether focus should be on constitutional powers or practical, ‘real’ powers. Often, these coalesce, but there are also instances where the discrepancy can be substantial. In his seminal article, Duverger (1980) noted that among the western European semi-presidential systems, the French president was constitutionally weak (in fact, weaker than the Austrian president) but in practice more powerful than the other presidents of the seven countries examined. Correspondingly, the president of

58

C. ANCKAR

Iceland possessed strong constitutional powers but in reality his or her powers were the weakest in the group of countries. Both ways of measuring powers have their advantages and drawbacks. Essentially, the main advantage of focusing solely on constitutional provisions is that these are easily interpretable and comparable across time and space. Assessing the actual powers of the head of state involves subjective judgments of the extent to which the head of state possesses powers. Moreover, the actual powers of the president can fluctuate quite a lot from year to year, depending on, for instance, partisan constellations in parliament (Doyle & Elgie, 2016, 733–735; Elgie 2011, 2–3). The main disadvantage in focusing on constitutional provisions only is that, as the examples of France and Iceland illustrate, constitutional provisions do not necessarily reflect actual political practice. Iceland is, indeed, an extreme case; Shugart and Carey (1992, 155–156) note that the president of Iceland constitutionally has more powers in both the legislative and the executive sphere than his or her French counterpart. However, throughout Iceland’s history, executive power has in practice been vested in the prime minister, and in reality, the powers of the president are merely ceremonial and comparable to those of Scandinavian monarchs (Petersson, 2000, 83–84). To some extent, the choice between focusing on constitutional provisions or political practice is one between Scylla and Charybdis. Let it be noted, though, that the majority of the indices pay regard to constitutional provisions; Doyle and Elgie (2016) identified only five out of nineteen indices that made use of behavioural power aspects of the presidents. Within the framework of the present study, I have nevertheless opted for this alternative. For one thing, in contrast to most of the other empirical studies conducted in the field, the present research population includes not only republics, but also monarchies. The discrepancy between de jure powers and de facto powers is especially problematic in this category. In most cases, existing constitutional monarchies have arisen from previous absolute monarchies or, at least, systems in which the monarch has possessed rather strong powers. Especially in constitutions written in the late nineteenth or early twentieth century, the monarch was often granted wide-reaching powers in the executive sphere. However, these powers were granted to the monarch ‘with the expectation that [they]… would usually be exercised according to … accepted norms of constitutional propriety that, although not written down in the constitution, were nevertheless regarded as binding

3

THE RESEARCH POPULATION

59

on political actors…’ (Bulmer, 2017, 7). Accordingly, from 1809 to 1974, the Swedish constitution granted executive powers to the monarch whereas legislative powers were shared by the monarch and the legislature. In reality, however, the Swedish monarchs had been gradually stripped of their powers during the nineteenth century and lost all their influence on government formation already in 1918. In neighbouring Norway, article 3 of the constitution still proclaims that ‘[t]he executive powers is vested in the King…’ and article 12 that ‘[t]he King himself chooses a Council…’, i.e. a government led by a prime minister. In reality, however, the Norwegian monarch has not had any influence in the government formation process since 1928 (Narud & Strøm, 2000, 172). In Belgium, article 37 of the constitution bluntly states that ‘[t]he federal executive power, as regulated by the Constitution, belongs to the King’, whereas article 36 proclaims that ‘[t]he federal legislative power is exercised jointly by the King, the House of Representatives and the Senate’. The clash between constitutional powers and practical powers was driven to extremes in 1990, when the Belgian parliament adopted a law permitting abortion. Although the Belgian monarch does not have explicit veto powers, the Belgian constitution stipulates that laws require the signature of the monarch in order to come into force. A devoted Catholic, King Baudouin, proclaimed that he was unable to sign the law. The crisis was solved by a compromise; the government, invoking article 82 of the constitution, declared that the king was unable to govern, then signed the bill into force, after which the king was restored as monarch by the parliament the next day. Another reason for relying on political practice spells data-availability. In recent years, it has become much easier to make cross-country comparisons on the basis of behavioural aspects. The mastodontic V-dem dataset (Coppedge et al., 2020) is particularly useful for the purpose of the present research effort as it contains a number of variables that relate the powers of the head of state to those of the head of government based explicitly on political practice. Within the framework of the present study, I shall measure the powers of the head of state with reference to seven questions in the V-dem dataset. Three of these refer explicitly to powers in the executive sphere and two to legislative powers. With regard to executive powers, the first question concerns to what extent the head of state appoints cabinet in

60

C. ANCKAR

practice, the second one whether the head of state dismisses cabinet ministers, and the third if the head of state can remove the prime minister. The dataset also contains a question which accounts for if the prime minister is chosen by the head of state. However, an examination of the cases, reveals that many of the V-dem experts have interpreted the question rather loosely, and included many countries, for instance current day Belgium and Denmark, where the head of state is not involved in choosing the prime minister, but only makes the final, formal, appointment. The question is therefore left out from further analysis. With regard to legislative prerogatives, I include a question that accounts for to what extent the head of state can propose legislation, and another one which concerns the veto powers of the head of state. Finally, I also include two questions that do not fall explicitly in either of the aforementioned categories. First, there is a question that has to do with the ability of the head of state to dissolve the legislature. This prerogative affects directly the legislative sphere but also indirectly the executive sphere, since the dissolution of parliament also leads to the resignation of the government (e.g. Neto & Lobo 2009; Shugart 2005, 334–335). Second, there is a question that accounts for the influence the head of state exercises over domestic policy. Some of the questions are measured on a dichotomous scale whereas others are measured using an ordinal one. The ordinal ones are converted, so that they take values between 0 and 1. The values can then be added, whereby an index measuring the powers of the head of state emerges. The index varies between 0.0 and 7.0, where low values indicate that the powers of the head of state are weak, and high values that they are strong. Values along the ordinal scale are, per definition, not equidistant. In some cases, the distance between two points along the scales is very low, and in these cases I have, when transforming the values of the ordinal scale to the interval 0 to 1, merged some values. For a detailed explanation of how the index has been constructed, the reader is referred to the Appendix. Like most other indices of presidential prerogatives (Frye 1997; Shugart & Carey, 1992; Siaroff, 2003) the index used here is additive, resting on the assumption that each prerogative is equally important when assessing the overall strength of the head of state. This, of course, is not necessarily true. On the other hand, it is far from self-evident how a weighting should be conducted. One could, of course, argue that the ability to dissolve the legislature is a far more important prerogative than having the power to veto a bill. On the other hand, since dissolving the

3

THE RESEARCH POPULATION

61

legislature has far-reaching consequences (premature elections and formation of a new government) the threshold for the head of state to exercise this power in reality can be expected to be very high and definitely much higher than vetoing a bill. In one respect, there is a weighting of the dimensions as three of the prerogatives concern the executive sphere, two the legislative sphere and two none of the spheres exclusively. This makes sense, as the focus of the present study is on executive power sharing. Sartori (1997, 131–135), has argued that the ‘dual authority structure…is the distinctive feature of [semi-presidentialism]’, wherefore the crucial powers are the ones that the head of state competes for or shares with the prime minister in the executive sphere. To a considerable extent, this assessment is valid also for dual executives in general.

3.5

Dual Executives 1850–2019

When classifying the cases along the two central dimensions, we identify a large number of dual executives in the world. Unfortunately, some of these cases are not included in the V-dem dataset. For some of the countries not included in the V-dem dataset, I have calculated power scores relying on other sources than V-dem.5 However, a handful of countries, almost all of which small island states in the Caribbean or the Pacific, are excluded due to limited data-availability. In all, 4,110 cases of dual executives are included in the study and their distribution among the three selection mode categories is presented in Table 3.1. As we can see, the cases are fairly evenly distributed among the three categories. There are 1,894 cases of dual executives with a monarch as head of state, 1,160 instances of dual executives with a popularly elected president, and 1,056 cases with a non-popularly elected one. Geographically, the cases are widely dispersed over the five continents, although the bulk of the cases is found in Europe. With regard to the time period, we note that there are very few examples of dual executives in the nineteenth century. Indeed, very few countries surpass the thresholds of democracy applied in this study. From the year 1800 to 1847, the only country classified as a democracy is the United States. In 1848, Switzerland and France are added to the list. None of the above-mentioned cases meet the criteria of dual executives as applied within the framework of the present study. In the French second republic (1848–1952), again, executive powers where conferred to the president who was also responsible

62

C. ANCKAR

Table 3.1 Dual executives according to mode of selection of head of state in democracies 1850–2019 Non-popularly elected president

Popularly elected president

Monarchy

Albania 1992–1995, 1997–2019 Austria 1920–1932, 46–50 Bangladesh 1991–2006, 2009–2013 Belarus 1991–1993 Bulgaria 1990–1991 Czechoslovakia 1918–1939, 45–46, 90–92 Czech Republic 1993–2011 Estonia 1990–2019 Finland 1918–1924 Fiji 1992–1999, 2002–2005, 2014–2019 France 1875–1939, 46–64 Germany 1919–1924, 1990–2019 Ghana 1970–1971 Greece 1927–1934, 1974–2019 Hungary 1990–2019 Iceland 1944–1951 India 1950–2019 Indonesia 1955–1956 Ireland 1937 Israel 1948–2019 Italy 1946–2019 Kosovo 2012–2019 Latvia 1920–1933, 1990–2019 Lebanon 1970–1975 Lithuania 1920–1925, 1990–1992 Malta 1974–2019 Mauritius 1992–2019 Moldova 2001–2015 Mongolia 1990–1992 Myanmar 1948–1957, 60–61 Namibia 1990–1993 Nepal 2008–2019 Nigeria 1963–1965 North Macedonia 1991–1993 Pakistan 1972–1976, 88–98, 2008–2017 Poland 1918–1925, 1989 Portugal 1911–1925

Armenia 2019 Austria 1951–2019 Bosnia & Hercegovina 1996–2019 Brazil 1961–1962 Bulgaria 1992–2019 Burkina Faso 2007–2019 Cape Verde 1991–2019 Central African Republic 1993–2002 Croatia 2000–2019 Cuba 1940–1951 Czech Republic 2012–2019 East Timor 2002–2019 Finland 1925–2019 France 1965–2019 Germany 1925–1932 Georgia 2004–2019 Greece 1926 Guinea-Bissau 1993–1997, Iceland 1952–2019 Ireland 1938–2019 Kenya 2008–2012 Lithuania 1993–2019 Madagascar 1993–2008, 2016–2019 Mali 1991–2011 Mongolia 1993–2019 Montenegro 2006–2019 Moldova 1991–2000, 2016–2019 Mozambique 1994–2003 Namibia 1994–2019 Niger 1993–1995, 1999–2008, 2011–2019 North Macedonia 1994–2019 Peru 1956–1961, 63–67, 80–91, 2001–2019 Poland 1990–2019 Portugal 1976–2019 Romania 1991–2019 Russia 1992–98 São Tomé & Príncipe 1991–2019 Senegal 1993–2019 Serbia 2006–2019

Australia 1901–2019 Barbados 1966–2019 Belgium 1894–1913, 1918–1939, 1944–2019 Bhutan 2009–2019 Canada 1867–2019 Denmark 1901–1942, 1945–2019 Fiji 1970–1986 Greece 1864–1914, 35, 46–66 Ireland 1922–1936 Italy 1919–1921 Jamaica 1962–2019 Japan 1952–2019 Laos 1954–1958 Lesotho 2002–2019 Liechtenstein 1921–2019 Luxembourg 1900–1939, 1944–2019 Malta 1964–1973 Mauritius 1968–1991 Nepal 1991–2001 Netherlands 1896–1939, 1945–2019 New Zealand 1907–2019 Nigeria 1960–1962 Norway 1905–1939, 1945–2019 Pakistan 1950–1955 Papua New Guinea 1975–2019 Sierra Leone 1961–1966 Solomon Islands 1978–1999, 2006–2019 Spain 1977–2019 Sri Lanka 1948–1971 Sweden 1911–2019 Thailand 1975, 83–90, 1992–2005, 2011–2013 Tonga 2012–2019 Trinidad & Tobago 1962–1975 United Kingdom 1885–2019 Yugoslavia 1921–1928

(continued)

3

THE RESEARCH POPULATION

63

Table 3.1 (continued) Non-popularly elected president

Popularly elected president

Slovakia 1993–1998 Somalia 1960–1968 Spain 1931–1936 Sri Lanka 1972–1977 Sudan 1956–1957, 86–88 Suriname 1975–1979 Trinidad & Tobago 1976–2019 Tunisia 2012–2013 Turkey 1961–1979, 89–2013 Vanuatu 1980–2019 West Germany 1949–1989

Slovakia 1999–2019 Slovenia 1991–2019 South Korea 1960 Sri Lanka 1991–2009, 2016–2019 Taiwan 1996–2019 Tanzania 1996–2000, 2006–2015 Turkey 1983–1988, 2014–2015 Tunisia 2014–2019 Ukraine 1991–2019 Yugoslavia 2000–2005

Monarchy

to parliament.6 The aforesaid means that although the book accounts for dual executives during the whole period 1800–2019, in practice the time period covered ranges from the mid-nineteenth century to date. Many of the oldest variants of dual executives are found in the category of monarchies. The oldest example of a dual executive is Greece, which passed the threshold of democracy in 1864, at that time with a monarch as head of state. There are also four other nineteenth-century monarchies, all of which have retained the regime form to date, namely Canada, which has been classified as a democracy ever since it became independent in 1867, United Kingdom since 1885, Belgium since 1894, and the Netherlands since 1896.7 Republican dual executives did not emerge until the twentieth century. The only exception is the French third republic. During its first four years (1871–1874), France had, effectively, a single executive, where the president led the government and could be brought down by the parliament (which occurred in 1873, when President Adolphe Thiers was released from office). However, with the adoption of the constitution in 1875, France became the first democratic country in which executive power was divided between a prime minister, who could be brought down by parliament, and a president not subject to a parliamentary vote of no confidence. The second republican dual executive emerged in Portugal in 1911, after the monarchy had fallen the previous year. After the First World War, dual executives emerged in the former Central powers

64

C. ANCKAR

Germany and Austria but also in many newly independent states, such as Finland, Poland, Latvia, and Lithuania. In Table 3.2, the power dimension is added to the dimension of the selection of head of state. The cases are split up into five categories in terms of how powerful the heads of state are. Needless to say, imposing cut-off points on the power scale is highly problematic and, to a certain extent, arbitrary. However, one category is uncontroversial, namely the one consisting of countries where the head of state is completely powerless, i.e. cases where the index of powers is 0.0. In these systems, the position of the head of state is separated from the head of government, but the head of state has only ceremonial powers. Such systems, in themselves are of lesser interest for the purpose of the present study. However, as will be discussed momentarily they are of high interest when we discuss the stability and transformation patterns of the dual executives. A large part of these countries are made up of constitutional democracies with a monarch as head of state and a prime minister as head of government. However, similar arrangements occur in republics as well; there are many instances where the president has very few prerogatives, regardless of whether he or she is popularly elected or not. The rest of the cases are far from evenly distributed along the power dimension. In the vast majority of the cases, the head of state has very weak powers, almost always referring to veto powers or dissolution powers. In such cases, the head of state is almost, but not completely, powerless. It therefore makes sense to confine these numerous cases to Table 3.2 Executive power sharing patterns in the world 1850–2019. V-dem figures Selection mode of head of state Powers of head of state 5.00–7.00 3.00–4.99 1.00–2.99 > 0.00–0.99 0.00 Total

Non-popular election

Popular election

Inheritance

Total

7 (2) 36 (8) 311 (24) 542 (33) 160 (14) 1,056 (81)

240 (17) 179 (15) 215 (20) 507 (23) 19 (4) 1,160 (79)

99 (1) 72 (9) 332 (20) 832 (17) 559 (10) 1,894 (57)

346 (20) 287 (32) 858 (64) 1,881 (73) 738 (28) 4,110 (217)

Each cell returns the number of cases with the number of countries in brackets

3

THE RESEARCH POPULATION

65

a separate category, consisting of cases where the power index exceeds 0, but is less than 1.00. In terms of mode of selection of the head of state, all three categories are well represented here. Countries with popularly elected presidents include Austria (1946–2019) and Bulgaria (1990–2019), countries with non-popularly elected presidents Estonia (1992–2019) and Italy (1946–2019), and monarchies Belgium (1900– 1913, 1918–1939) and the Netherlands (1900–1939, 1945–1980). The rest of the cases are conferred to three categories. If the index of powers equals or exceeds 1.0 but is below 3.0, the head of states are weak. For instance, Portugal (1976–2004), Mauritius (1992–2012), and Sweden (1911–1917) fall within this power category. Countries in the interval 3.0 to 4.99 have moderately strong heads of state and include, for instance, France during most parts of the post WWII-period, Finland (1917–1980), Greece (1900–1910), and Thailand during the years the country has met the criteria of democracy. Finally, countries situated in the span 5.0 to 7.0 have powerful heads of state. Examples include the popularly elected presidents of Sri Lanka (1991–2009) and Ukraine (1996–2004), the non-popularly elected president of Pakistan (1988, 2008) and the monarch of Liechtenstein. A quick glance at the distribution of cases in Table 3.2 shows that the overall trend is hardly surprising; generally, powerful presidents occur in countries where the head of state is popularly elected. At the same time, there is no unequivocal association between the phenomena; in more than 45 per cent of the cases with popularly elected presidents, the powers of the head of state are insignificant at best. In addition, there are 43 cases and nine countries where a non-popularly elected president has had at least moderately strong prerogatives.8 What is even more surprising is the large representation of monarchies in the three highest categories along the scale. In all, this bird’s eye view of the combination of mode of selection and powers of the head of state provides an interesting point of departure for more comprehensive assessments of executive power sharing.

Appendix: Measuring the Powers of the Head of State Based on the V-dem Dataset 1. Head of State appoints cabinet in practice (C) (v2exdfcbhs, *_osp, *_ord). ‘In practice, does the head of state have the power to appoint –or is the approval of the head of state necessary for the appointment of—cabinet ministers?’ Recoding procedure: 0–2 = 0, 3–4 = 1.

66

C. ANCKAR

2. Head of State dismisses ministers in practice (C) (v2exdfdmhs, *_osp, *_ord). ‘If the head of state took actions to dismiss cabinet ministers, would he/she be likely to succeed?’ Recoding procedure: 0–1 = 0.0, 2 = 0.5, 3 = 1.0. 3. Head of Government removal by other in practice(C)(v2exrmhgnp). ‘Which of the following bodies would be likely to succeed in removing the head of government if it took actions (short of military force) to do so?’ Recoding procedure: 5 = 1, others = 0. 4. Head of State proposes legislation in practice (C) (v2exdfpphs, *_osp, *_ord). Recoding procedure: 0 = 1.0, 1 = 0.5, 2 = 0.0. 5. Head of State veto power in practice (C) (v2exdfvths, *_osp, *_ord). Recoding procedure: 0 = 0.00, 1 = 0.25, 2 = 0.50, 3 = 0.75, 4 = 1.00. 6. Head of State dissolution in practice (C) (v2exdfdshs, *_osp, *_ord). ‘If the head of state took actions to dissolve the legislature, would he/she be likely to succeed?’ Recoding procedure: 0 = 0.00, 1 = 0.33, 2 = 0.67, 3 = 1.00. 7. HOG control over (C) (v2exctlhg). ‘In practice, from which of the following bodies does the head of government customarily seek approval prior to making important decisions on domestic policy?’ Recoding procedure: 5 = 1, others = 0.

Notes 1. An alternative strategy would have been to focus only on V-dem’s polyarchy index, which corresponds with Boix, Miller, and Rosato’s conception of democracy. However, operating with thresholds solely along this scale is problematic; regardless of where along the scale the threshold is set, some democratically dubious cases are likely to be included and/or important cases of dual executives to be excluded. For instance, even applying a threshold of 0.5 on the polyarchy scale would mean that Armenia 1990– 1995, Belarus 1996, and Burkina Faso 2000–2006, all of which Boix, Miller, and Rosato classify as autocracies, would have been eligible for inclusion. Focusing on the polyarchy scale alone also means that some countries can reach the threshold of democracy a given year, only to be excluded the following year after a marginal change in the score. A telling example is Moldova, which Boix, Miller, and Rosato classify as a democracy from the year 1991 and onwards. If we were to classify Moldova solely based on the polyarchy scale, applying the threshold 0.5, Moldova would be classified as authoritarian during the year 1991, as a democracy in 1992, as an autocracy

3

THE RESEARCH POPULATION

67

in 1993, as a democracy 1994–2004, as an autocracy 2005, as a democracy 2006, as an autocracy 2007, as a democracy 2008, as an autocracy 2009, and as a democracy from the year 2010 and onwards. As it turned out this problem was reduced, although certainly not eliminated by applying a threshold of 0.4 on the polyarchy scale with the additional criterion that countries also receive the value 0.4 on the liberal democracy scale. What is crucial in this context is that the alternative criteria of democracy are stricter than the ones Boix, Miller, and Rosato operate with. This reduces the risk that a country meets the criterion of democracy based on V-dem’s scales but not with reference to Boix, Miller, and Rosato a given year. 2. Bhutan 2009–2010, Burkina Faso 2007–2008, 2010–2014, Fiji 1993– 1999, 2002–2005, Latvia 1991–1992, Lithuania 1990–1991, Namibia 1990–1993, 1995–2019, Netherlands 1896, Peru 1991, Senegal 1993– 1999, Solomon Islands 2004, Sri Lanka 1977, Tanzania 1996–2000, 2006–2010. Tonga is not included in the V-dem. dataset. Contrary to Boix, Miller, and Rosato, I include the country among the democracies from the year 2012, well after the constitutional reform of 2010, which reduced the powers of the monarch. Boix, Miller, and Rosato classify Cuba as a democracy in 1952. Presidential elections were to occur that year but three months before the elections, one of the contenders, Fulgencio Batista, launched a coup whereby he rose to power. I therefore do not include Cuba among the democracies for the year 1952. 3. In order to avoid allowing a marginal fluctuation on the V-dem scores during a single year count as a regime change, Namibia is included in the population also for the year 1993, Tanzania for the year 2013, and Burkina Faso for the years 2009 and 2015 despite the fact that their scores on Vdem’s lib.dem index failed to reach the threshold by a narrow margin. For the same reason, Solomon Islands is excluded for the year 2004, in spite of the fact that the country managed to barely surpass the threshold. 4. A very difficult case in this respect is Guyana. The country incorporates elements of parliamentarism, semi-presidentialism, and presidentialism. Much of the confusion is can be attributed to the constitution, which is extremely vague with regard to the relations between the executive and the legislature. Articles 177–183 do not contain any provisions which would suggest that the president and/or the ministers of his or her government could be removed from their positions by a parliamentary vote of no confidence. These articles stand in sharp contrast to article 106, which stipulates that there shall be a Cabinet, composed of the President, the Prime ministers and other ministers the president sees fit to appoint. Article 106 (2) states that ‘[t]he Cabinet shall aid and advise the President in the general direction and control of the Government of Guyana and shall be collectively responsible therefor to Parliament’. The wording of the article thus

68

C. ANCKAR

5. 6.

7.

8.

suggests that the Cabinet, including the president, can be removed from office by a parliamentary vote of no confidence. However, this is the only reference to the principle of parliamentarism in the constitution. To complicate matters further, the Cabinet is not the same as government; article 99 (1) bluntly states that ‘[t]he executive authority of Guyana shall be vested in the president…’ and many ministers are not included in the Cabinet. It is therefore not surprising that authors have had a difficult time classifying the political system of Guyana. In some categorizations, the country falls closely to a presidential system in the South Korean variant, where there is a prime minister alongside the president, but who is not responsible to parliament (e.g. Siaroff 2003, 295). For a long time, the political practice gave support to this classification. The V-dem dataset contains a question (HOS removal by legislature in practice, v2exremhsp), which asks if the legislature would be likely to succeed if it attempted to remove the head of state. During the period 1980-2016, the V-dem experts have answered ‘No, under no circumstances’ to the question. However, in 2014 a motion of no confidence was raised for the first time in Guyana, but the president chose to dissolve parliament before it could be debated (Nagamootoo, 2018). Finally, in 2018, a motion of no confidence was raised and approved by parliament. As a consequence hereof, the government and the president were forced to step down. These developments suggest that Guyana should be classified as parliamentary rather than presidential. For the purpose of the present study, it matters little if we regard Guyana as presidential or parliamentary. In either case, the country fails to meet the criteria of a dual executive, since the prime minister merely functions as an aide to the president and his or her political survival is completely tied to the fate of the president. Notably Veenendaal (2013), Beattie (2004), and Quintino (2018); Matangi Tonga (https://matangitonga.to/). In the beginning of the Second Republic, one of the ministers held a position that bears resemblance to the position as a prime minister. The minister bore the title ‘Présidant le conseil des ministres en l’absence du Président de la République (Muel 1890, 251–253). New Zealand is included as a democracy in Boix, Miller, and Rosato’s dataset from the year 1857. However, for the purpose of the present work, the country is regarded as independent from the year 1907, when it became a dominion. The added number of countries in the two highest categories is ten, but Pakistan occurs in both categories.

3

THE RESEARCH POPULATION

69

References Anckar, C., & Fredriksson, C. (2020). Coding rules for the political regimes of the world dataset. Political Regimes of the World Database, https://doi.org/ 10.7910/DVN/GWQDQ3/SQ407. Harvard Dataverse, V1Coding Rules for Political Regimes of the World Dataset. Harvard Dataverse. Beattie, D. (2004). Liechtenstein: A modern history. I. B. Tauris. Beliaev, M. V. (2006). Presidential powers and consolidation of new postcommunist democracies. Comparative Political Studies, 39(3), 375–398. Boix, C., Miller, M., & Rosato, S. (2013). A complete data set of political regimes 1800–2007. Comparative Political Studies, 46, 1523–1554. Bulmer, E. (2017). Constitutional Monarchs in Parliamentary democracies. International IDEA Constitution-Building Primer 7. Cheibub, J. A., Gandhi, J., & Vreeland, J. R. (2010). Democracy and dictatorship revisited. Public Choice, 143(1), 67–101. Coppedge, M., Gerring, J., Lindberg, S. I., Skaaning, S.E., Teorell, J., … Miri, F. (2020). V-Dem Codebook v5. Varieties of Democracy (VDem) Project. Doyle, D., & Elgie, R. (2016). Maximising the reliability of cross-national measures of presidential power. British Journal of Political Science, 46(4), 731–741. Duverger, M. (1978). Échec au roi. Albin Michel. Duverger, M. (1980). A new political system model: Semi-presidential government. European Journal of Political Research, 8(2), 165–187. Elgie, R. (2011). Semi-Presidentialism: An increasingly common constitutional choice. In R. Elgie, S. Moestrup, & Y. S. Wu (Eds.), Semi-presidentialism and democracy (pp. 1–20). Palgrave Macmillan. Fortin, J. (2013). Measuring presidential powers: Some pitfalls of aggregate measurement. International Political Science Review, 34(1), 91–112. Freedom House. Online. Available www.freedomhouse.org. Frye, T. A. (1997). Politics of institutional choice: Post-communist presidencies. Comparative Political Studies, 30(5), 523–552. Marshall, M. G., Gurr, T. R., & Jaggers, K. (2014). Polity IV project: Political regime characteristics and transitions, 1800–2013. George Mason University. Matangi Tonga. https://matangitonga.to/. Metcalf, L. K. (2000). Measuring presidential power. Comparative Political Studies, 33(5), 660–685. Muel, L. (1890). Gouvernements, ministères et constitutions de la France depuis cent ans. Mouillot. Nagamootoo, M. (2018, December 23). Historic no-confidence vote. Guyana Chronicle. Narud, H. M., & Strøm, K. (2000). Norway: A fragile constitutional order. In Müller, W., & Strøm, K. (Eds.), Coalition governments in Western Europe. Oxford University Press.

70

C. ANCKAR

Neto, O. A., & Lobo, M. C. (2009). Portugal’s Semi-presidentialism (re)considered: An assessment of the president’s role in the policy process, 1976–2006. European Journal of Political Research, 48(2), 234–255. Paloheimo, H. (2003). The rising power of the Prime Minister in Finland. Scandinavian Political Studies, 26(3), 219–243. Pasquino, G. (1997). Semi-presidentialism: A political model at work. European Journal of Political Research, 31(1), 128–37. Petersson, O. (2000). Nordisk politik. Femte upplagan. Norstedts Juridik AB. Roper, S. D. (2002). Are all semipresidential regimes the same? A comparison of premier-presidential comparison regimes. Comparative Politics, 34(3), 253– 272. Quintino, K. (2018). The kingdom of Bhutan: A democracy by obligation. Portuguese Journal of Political Science, 10, 65–76. Samuels, D. J., & Shugart, M. S. (2010). Presidents, parties, and Prime Ministers. Cambridge University Press. Sartori, G. (1997). Comparative constitutional engineering: An inquiry into structures, incentives and outcomes (2nd ed.) Macmillan. Shugart, M. S. (2005). Semi-presidential systems: Dual executive and mixed authority patterns. French Politics, 3(3), 323–351. Shugart, M. S., & Carey, J. M. (1992). Presidents and assemblies: Constitutional design and electoral dynamics. Cambridge University Press. Siaroff, A. (2003). Comparative presidencies: The inadequacy of the presidential, Semi presidential and parliamentary distinction. European Journal of Political Research, 42(3), 287. Tavits, M. (2009). Presidents with Prime Ministers: Do direct elections matter? Oxford University Press. Veenendaal, W. P. (2013). Politics and democracy in microstates. A comparative analysis of the effects of size on contestation and inclusiveness (Dissertation). University of Leiden.

CHAPTER 4

How Dual Executives Emerge

4.1

Introduction

Whereas authors have spent a great deal of time and effort explaining effects of different regime forms, the literature on the determinants of democratic regime choice is surprisingly scarce. A study by Jung and Deering (2015) constitutes a welcome exception. They introduce a wide set of variables in order to explain the occurrence of presidential, parliamentary, and ‘mixed’ regime forms in democratizing countries. Although the three different regime forms are partly explained by different independent variables, the results show that historical or constitutional legacy along with geographical proximity tend to be the most important explanations for all three regime forms. Similar results were reported by Åberg and Denk (2020, 93), who found that ‘the establishment of parliamentary, presidential and semi-presidential regimes stems from geographical, cultural, and post-colonial networks’ (see also Persson & Tabellini, 2003, 103). For the purpose of the present study, it is particularly noteworthy that the category ‘mixed regimes’, or semi-presidential regimes conceived of in Elgie’s terms, is completely immune to other explanations than geography and colonial heritage (Jung & Deering, 2015, 69–72). These findings are well in line with other observations; when tracing the determinants of semi-presidentialism, authors have consistently identified constitutional © The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 C. Anckar, Presidents, Monarchs, and Prime Ministers, Palgrave Studies in Presidential Politics, https://doi.org/10.1007/978-3-031-03960-7_4

71

72

C. ANCKAR

heritage and diffusion patterns as the most important ones (e.g. Åberg & Denk, 2020; Elgie et al., 2011; Neto & Lobo, 2014; Wu, 2011). In line with the aforesaid, it is natural to put emphasis on the role diffusion plays when countries introduce regimes with dual executives. Within this tradition, the historical legacy is widely seen as important; when countries gain independence, they tend to copy or inherit the regime form of the former mother country. Another pattern of diffusion highlights the role of geographic proximity, suggesting that countries choose dual executives because the regime form is used in countries in their neighbourhood. A second perspective is to trace the origin of existing dual executives from their previous regime forms. The basic idea is that certain regime forms are more likely than others to transform into specific regime types. Regime transformations have received a growing scholarly interest during the last two decades. Although the main purpose of most of the studies have been to answer the question if certain authoritarian regime types are more likely than others to democratize, some of these studies have also tried to detect patterns of transformation between different types of authoritarian regime forms (Anckar, 2018; Geddes, 1999; Geddes et al., 2014; Hadenius & Teorell, 2007; Magaloni, 2008). For the purpose of the study at hand, the expectation is that some regime forms are more likely than others to generate dual executives. This approach has been somewhat neglected in the literature on mixed regime forms. Although authors have linked the semi-presidential regimes in Eastern Europe to their previous party-based systems, there has, to my knowledge, been no systematic comparison of if and how various forms of dual executives have emerged from different types of authoritarian regime forms. In the following section, I argue that three of the authoritarian regime forms discussed above, namely absolute monarchies, monarchic oligarchies, and party-based systems, are more likely than other autocratic regime forms to transform into dual executives. I also discuss the possibility that executive power sharing arrangements grow out of former democratic presidential systems. The two sets of explanations cannot be completely separated from each other. Countries that gain their independence do not lack a ‘previous’ regime form, as they have been administered in accordance with the regime form of their mother countries and/or other constitutionlike arrangements prior to their independence. I shall give this issue due attention, theoretically and empirically towards the end of the chapter.

4

HOW DUAL EXECUTIVES EMERGE

73

Suffice it here to say, that the interrelatedness between the explanatory factors means that conducting separate bivariate analyses of their respective impact on how dual executives are chosen leads to partly overlapping results and conclusions. To avoid repeating findings that mainly reflect the interrelatedness of the previous regime form and colonial heritage, I choose to direct more space and attention to the relationship between the previous regime form and regime choice than to the association between colonial heritage and regime choice in the analyses. This choice is motivated primarily by the fact that the link between colonial heritage and regime choice (semi-presidentialism in particular) already has been treated quite extensively in the literature. It is thereby also natural to begin with assessing the relationship between the previous regime form and regime choice and then continue by linking diffusion patterns to regime choice.

4.2

Previous Regime Form and Regime Choice 4.2.1

The Monarchic Legacy

The theoretical justification for a relationship between the monarchic regime form and dual executives goes back half a century in time, more precisely to the year 1968, when Huntington (1968, 177–191), coined the term ‘the King’s dilemma’ for describing the challenges monarchs faced when choosing the best strategy for defending their position in a modernizing world. This ‘dilemma’, he argued, could be solved in three different ways, by transformation, coexistence, and maintenance. The strategy of transformation implied the transfer of power from the monarch to another institution, such as the people, a bureaucratic elite, or the military. Coexistence, again, referred to a strategy of combining monarchic powers with a democratic form of government, or more specifically, ‘to institutionalize competitive coexistence in the polity of two independent sources of power’ (Huntington, 1968, 180). Maintenance, finally, encompassed various means by which the monarch attempted to maintain his or her powers. Such means included allowing persons with a middleclass background to obtain high-level positions in government, fighting modernization, or intensifying repression. For the purpose of the present research effort, we are explicitly concerned with those strategies where the ambition is to democratize the country, i.e. the options transformation and coexistence. However, the two strategies are far from mutually exclusive; in fact, they overlap to

74

C. ANCKAR

considerable extents, merely reflecting differences in degree rather than kind. When choosing the strategy of transformation, the powers of the monarch are, in essence, completely transformed to the people, resulting in a constitutional monarchy where the monarch, echoing Macaulay, ‘reigns but does not rule’ (Huntington, 1968, 177). When the option coexistence is chosen, the monarch preserves some (or a substantial part) of his or her powers, coexisting with democratic institutions, particularly with a prime minster emerging from and responsible to parliament. It is evident that in their refined versions, the option transformation fits well with constitutional monarchy and coexistence with semi-constitutional monarchy. However, as pointed out above, the choice between the two strategies is not necessarily an ‘either/or’ choice but rather reflect a spectrum. What is crucial is the fact that the monarch is compelled to give up some, a lot, or all of his or her prerogatives. It is, however, reasonable to suspect that in most cases the monarch is reluctant to give up his or her powers. Especially, in cases where the monarch enjoys widespread popularity and/or has reigned for extensive periods of time, a handy compromise in a democratizing country would be to let the monarch continue to exercise influence along with the prime minister rather than to abruptly strip him or her of all powers. Indeed, Huntington (1968, 178) expected that a radical shift from absolute monarchy to constitutional monarchy was highly unlikely, and would require ‘either time or revolution’. Perhaps even more importantly, he was very pessimistic regarding the long-term prospects for power sharing between monarchic rule and party government, arguing that ‘[t]he pressures in such a system are either for the monarch to become only a symbol or for him to attempt to limit the expansion of the political system…’ (Huntington, 1968, 180–181). Particularly the death of the monarch provides a good opportunity to strip the monarch of his or her powers (see also Anckar, 2021; Huntington, 1968, 180). Based on the aforesaid, we can expect that dual executives with a monarch as head of state grow out of former absolute monarchies. We can further make the assumption that the powers of the monarch in a democratic country are likely to decrease as time passes by. This is important when we are linking the birth of dual executives to the former regime types, as we can expect the category monarchic oligarchy to occupy an intermediate position in a transformation process from absolute monarchy to semi-constitutional and constitutional monarchy. In other words, the

4

HOW DUAL EXECUTIVES EMERGE

75

dual executives with a monarch as head of state are expected to trace their origin to absolute monarchies, either directly or indirectly via the category monarchic oligarchy. 4.2.2

The Party-Based Legacy

The other authoritarian regime form likely to give rise to executive power sharing arrangements is the party-based regime. In a ‘pure’ single-party regime, one party is in complete control of the political sphere. In the classic communist variant, in use in the Soviet Union, the tentacles of the party usurped every level of society. At the lowest level were the party cells, comprising at least three members and functioning in widely diverse organizations but particularly in workplaces, e.g. in factories or schools. The intermediate level was the party committee, which existed in many variants and at different levels in society; the youth organizations of the communist party were particularly important for indoctrination purposes. At the highest level, power resided formally in the Congress of the Communist party, an organ that convened every five years. However, in reality the party’s supreme governing body was the Central Committee, comprising a few hundred members, elected by the party congress. Since the Central Committee rarely met, power was largely exercised within a much narrower organ, the Politburo, made up of around 30 members, elected by the Central Committee from among its members. The highestranking official within the party was the General Secretary of the Central Committee, who was thereby also de facto the leader of the Soviet Union. In parallel with the internal party hierarchy, the Soviet Union also had a legislature, a government, and a head of state. These organs, however, did not possess any real powers. At times, the position as head of state (in the Soviet case, Chairman of the Presidium of the Supreme Soviet) coincided with the position as General Secretary, but this was far from always the case. There was also a prime minister, who headed the government. Here, too, the position of prime minister could be held by the General Secretary of the Communist party (as was the case with Nikita Khrushchev during the years 1958–1964) but in reality, the government was an administrative organ rather than a political one. The single-party model described above was, with some modifications, adopted in the Soviet satellites states in Eastern Europe and in communist systems around the world, including China.

76

C. ANCKAR

The reason for why we expect former single-party regimes to develop into systems with dual executives when they democratize lies in their experiences with the dual power structure outlined above. After the fall of the Berlin Wall, the communist regimes in Eastern Europe had to transform themselves into democracies. At the same time, as there were popular pressures for democratic reforms, the power holders were eager to keep as much power as possible in their own hands. This could be achieved by transforming the position of General Secretary of the Communist party into the position of President, while giving in to popular demands for parliamentary democracy (Frison-Roche, 2007, 56–58). For the opposition forces, too, the shared executive appeared acceptable, notably because it made it less likely that one single party could receive total control by securing a majority in parliament. A semi-presidential form of government could be seen as a handy compromise as ‘it appeared to concomitantly guarantee democracy and political efficacy, while ensuring the political survival of each of the chief actors’ (Frison-Roche, 2007, 59; see also Blondel, 1992; Wu, 2011, 24–25). The link between a communist single-party legacy and semipresidentialism is well-established. However, it makes sense to assume that the same line of arguments is valid also when linking other variants of single-party rule to dual executives. Party-based authoritarian regime forms were, and are, found in many countries in Sub-Saharan Africa. In comparison with the communist party-based regimes, they tend to be much less organized; many times, the parties have to considerable extents merely functioned as a tool for the leader in his or her exercise of power, which means that the systems in question have fallen in a grey area between party-based rule and personalist rule. Nevertheless, in a transformation process from autocracy to democracy, we can expect that the central actors follow a similar line of reasoning as the actors in communist systems. In other words, while the power holder reluctantly has to accept democratic reforms, his or her main ambition is to keep as much power as possible in his or her own hands. The opposition, again, is likely to be sceptic towards a pure parliamentary form of government as this provides a complete control of power to the leader of the largest party in cases where the party receives a majority of the seats in parliament. In fact, the aversion against such a system should be particularly strong in the democratizing African single-party states, as the total dominance of a single party in parliament is, indeed, what has characterized the previous authoritarian regime form. There is, of course, also

4

HOW DUAL EXECUTIVES EMERGE

77

the option to choose a presidential form of government. This might be a tempting alternative for the authoritarian leader in power, but hardly acceptable for the opposition, as there is the risk that the president would try to consolidate his powers and, perhaps, eventually return the country to an authoritarian regime form (Linz, 1990; Riggs, 1997, 268). As we recall, the category party-based democracies was decomposed into single-party and multi-party authoritarian regimes. The discussion conducted above refer to single-party systems, i.e. systems where one party is in complete dominance in the political sphere. To some extent, the logic behind the reasoning is valid also for multi-party authoritarian systems, i.e. systems where one party rules but other parties than the dominating one are allowed to participate in elections and gain representation in parliament. After all, the distinction between single-party regimes and multi-party authoritarian regimes merely reflects the fact that the degree of autocracy varies in non-democratic systems. In a strict sense, the concept ‘multi-party authoritarianism’ is an oxymoron, since a true multi-party system cannot operate within a non-democratic setting. However, although we cannot escape the fact that the difference between single-party regimes and multi-party authoritarian regimes is a difference in degree rather than kind, it is equally evident that as we move along the scale from single-partyism to multi-partyism, the formal political institutions become increasingly important and the internal party organization become decreasingly important for gaining access to power (Levitsky & Way, 2010, 3–32; Barkan, 2013). This would suggest that the dual power structure (internal party hierarchy vs. formal institutions) is less manifest in systems with multi-party elections. At the same time, it is far from unimportant. When a multi-party authoritarian country is about to transform into a full-fledged democracy, the power holders are likely to reason along the same lines as power holders in single-party states, thereby opting for a system that allows the former power holder to continue in power. Whereas neither a presidential system nor a purely parliamentary system might be acceptable for both negotiating partners, an arrangement where executive power is shared between two actors is likely to be an acceptable solution, both to the existing power holders and to the leaders of the opposition.

78

C. ANCKAR

4.2.3

The Legacy of Other Regimes

With regard to other authoritarian regime forms, it is more difficult to see how they should naturally evolve into democracies with shared executives. In contrast to the case with monarchic and party-based regimes, there is less room for smooth institutional arrangements that would allow the former power holders to retain a substantial part of their powers within a democratic framework. Instead, regime transformations from military and personalist regimes to democracy are more likely to incorporate elements of abruptness and radicalism. In a military regime, parliaments tend to be suspended and legislative and executive powers concentrated in the hands of the military junta. At the same time, military officers do not necessarily seek political power, and, in cases where they do come to power, their primary goal is not necessarily to hold on to power but rather to ensure that the military remains hierarchical, well-functioning, and adequately funded (Geddes, 1999). Accordingly, military regimes tend to be more short-lived than other authoritarian regime forms (Anckar, 2018; Brownlee, 2009; Gandhi, 2008, 175–177; Geddes, 1999; Magaloni, 2008, 734–736; Smith, 2005). Personalist rulers, again, play an ‘all or nothing-game’ for their existence. As the power of the personalist ruler is not based on an institution but rather on personal connections, his or her support base is fragile, and the room for reaching compromises when faced with demands for democratic reforms is rather limited. Thus, when a personalist ruler falls, he or she should fall heavily. With regard to oligarchies I have already pointed out that the sub-category monarchical oligarchies occupy a natural transitory stage when a country democratizes from absolute monarchy to a semi-constitutional or constitutional monarchy. With regard to the remaining oligarchies, it is more difficult to make predictions about their propensity to develop into systems with shared executives, mostly because the group is quite heterogeneous. The bulk of the countries in this group consists of Latin American countries in the nineteenth and early twentieth centuries, where suffrage was restricted but elections were generally free and fair. It is difficult to see why these systems should be particularly prone to adopt a dual executive once suffrage was extended to the point where the countries surpassed the threshold of democracy. Finally, we should acknowledge the fact that dual executives also can emerge from existing democratic regime forms. Transition patterns within countries with a separate head of state and head of government will be

4

HOW DUAL EXECUTIVES EMERGE

79

assessed in Chapter 5. Within the framework of the present chapter, the question concerns to what extent it makes sense to assume that different regime forms with a single head of state and government should switch to a dual executive. The overwhelming majority of systems where the position as head of government coincides with that of head of state are presidential. However, as we recall, there are also a handful of parliamentary systems, notably South Africa and Botswana, where a president acts as both head of state and head of government while being subject to a parliamentary vote of no confidence. Within the framework of the present study, the crucial question is if there are any convincing theoretical arguments that would lend support to an assumption that presidential systems are likely to transform to systems with shared executives, and, if this is indeed the case, if they are more likely to choose semi-presidential or parliamentary arrangements. Perhaps one could argue that a shift from presidentialism into semi-presidentialism is a less radical, and thus more preferable, reform than a shift from presidentialism to parliamentarism. The counterargument, however, would be that if a country is dissatisfied with the presidential system, it is more likely to choose a new system that is radically different from the previous one, or, in other words, to choose a parliamentary form of government. With the lack of theoretical arguments pointing in either directions, it is difficult to escape the conclusion that the question to what extent presidential systems transform into shared executives is empirical rather than theoretical. 4.2.4

Operationalizing Regime Transformations

Let me then turn to the question how regime changes should be defined. Most datasets on regime characteristics use yearly observations for the countries included. Since regime changes can occur at any time during a year, countries that experience regime changes usually have more than one regime form during the year a regime change has occurred. Authors generally solve this problem by focusing on the situation at a specific date (e.g. December 31 or January 1) a given year (e.g. Boix et al., 2018; Geddes et al., 2014; Wahman et al., 2013). In the present contribution, I adopt the same strategy and account for the situation on December 31 a given year. Another problem has to do with the fact that some transitions are slow and/or complicated, involving exceptional measures and, often,

80

C. ANCKAR

provisional leaders with limited powers. When we are interested in transformation patterns, it makes sense to disregard most of these transitionary phases. For one thing, they are difficult to classify, as they often do not fit into the regime categories we operate with. Another reason is that they do not necessarily reflect the heritage left by the previous regime. Among such transitionary phases I also, as a rule, include periods of occupation during World Wars I and II. In some cases, the occupying force established puppet regimes, with the task of administering the occupied country. Such regimes, for instance the Vichy-regime in France, are generally disregarded when accounting for transformation patterns. The case of Greece provides a perhaps overly explicit illustration of the problem of classifying such transitionary phases in times of wars. During the years 1936 to 1941, the country was effectively ruled by a military regime under General Ioannix Metaxas, although the King remained the head of state. During World War II, King George II was in exile while Greece was occupied by the Axis powers, which had established a puppet regime led by a prime minister. After the end of the Greek occupation, the question of the future regime form of Greece remained open as large portions of the Greek population were critical towards the King and accused him of having collaborated with the Metaxas regime. However, in a referendum held in 1946, a large majority of the voters voted in favour of restoring the monarchy. Between the liberation and the referendum, Greece was ruled by Archbishop Damaskinos, who bore the title Regent, and functioned as both head of state and head of government. In the Greek, case, then I disregard the period of occupation as well as regime under Archbishop Damaskinos, and stipulate that the Greek dual executive with a hereditary monarch as head of state emerged from the military regime of Metaxas during the years 1936–1941. Moving on, many countries included in the study have become independent within the time span of the study. For these countries, the regime form in use in the country from which they emerged is regarded as their previous regime form. For instance, East Timor is included in the study from the year 2002, when it received its independence and opted for a semi-presidential form of government. East Timor emerged as an independent state from Indonesia, which at that point in time, like now, had a presidential form of government. Thus, the regime transformation in East Timor is deemed to have occurred from presidentialism to semi-presidentialism.

4

HOW DUAL EXECUTIVES EMERGE

81

Admittedly, sometimes the regime heritage of emergent independent states is difficult to establish. Finland (1917), Latvia (1918), and Lithuania (1918) evolved as independent nations from Soviet Russia in the aftermath of World War I. At that point, the political situation in Soviet Russia was chaotic; the Bolsheviks had risen to power in 1917, but the country was immediately thrown into a civil war. For Finland, Latvia, and Lithuania the dual executives are deemed to have emerged from the regime form of Tsarist Russia, i.e. absolute monarchy. The same conclusion is valid for Poland, who became independent in 1918 and included territories from the former Russian, German, and Austro-Hungarian Empires.

4.3

Previous Regime Form and Introduction of Dual Executives---Empirical Patterns

Let us then turn from theory to empirics. Table 4.1 gives a bird’s eye view of the relationship between the previous regime form and regime choices countries have made at the time they received their independence or transited from autocracy to democracy. For the sake of comparability, I also include a category of single executives, the bulk of which is made up of presidential systems but also includes a handful of parliamentary systems with a single head of state and government. A substantial number of countries inherited a dual executive regime form from their mother country when they received their independence. For the sake of consistency, these cases are also listed here. Finally, since the power dimension is not in focus here, the table also includes countries for which data on the power dimension is lacking in the V-dem dataset. In other words, the population of dual executives is slightly larger than the one presented in Table 3.1. All added cases are small states in the Caribbean or the Pacific.1 Based on the distribution of cases, we note that the strongest association between the former regime category and the ensuing regime type is the one between monarchic oligarchies and dual executives with a monarch as head of state. Eleven of the thirteen former monarchic oligarchies transformed into dual executives headed by a hereditary monarch when they democratized or gained independence. The contrast to former absolute monarchies is striking; more than half of the cases in this category developed into dual executives with a non-popularly elected president.

82

C. ANCKAR

Table 4.1 Previous regime form and type of executive at time of independence or at transition to democracy Head of state popularly elected

Head of state inherits his/her position

Single head of state and government

Total

Previous regime form ( t − 1)

Head of state not popularly elected

Parliamentarism, dual HOS/HOG Parliamentarism, single HOS/HOG Presidentialism Single-party system Multi-party authoritarian rule Military rule Personalism Absolute monarchy Monarchic oligarchy Other oligarchy Total

12 (30.0)

2 (5.0)

21 (52.5)

5 (12.5)

40 (100.0)

1 (100.0)

0 (0.0)

0 (0.0)

0 (0.0)

1 (100.0)

0 (0.0) 12 (54.5) 0 (0.0)

3 (50.0) 8 (36.4) 7 (38.9)

0 (0.0) 1 (4.5) 2 (11.1)

3 (50.0) 1 (4.5) 9 (50.0)

6 (100.0) 22 (100.0) 18 (100.0)

11 (19.3) 6 (14.0) 8 (53.3)

14 (24.6) 8 (18.6) 0 (0.0)

2 (3.5) 6 (14.0) 5 (33.3)

30 (52.6) 23 (53.5) 2 (13.3)

57 (100.0) 43 (100.0) 15 (100.0)

2 (28.6)

0 (0.0)

11 (84.6)

0 (0.0)

13 (100.0)

3 (37.5) 55 (24.7)

0 (0.0) 42 (18.8)

0 (0.0) 48 (21.5)

5 (62.5) 78 (35.0)

8 (100.0) 223 (100.0)

Each cell contains the absolute number of cases as well as the row percentage of cases (in parentheses) Chi-square 146.700 (sig. 0.000)

Another interesting discrepancy exists between single-party systems and multi-party authoritarian systems. Whereas countries situated in the former category have tended to transform into dual executives with a president as head of state, half of the countries in the latter category transformed into executives with a single head of state and government. This is in line with the expectation that rigid single-party states, with their dual power structure, would be more likely to transform into dual executives than systems where the dominating party has to confront some opposition on the electoral arena. As we recall, this line of reasoning was especially relevant for former communist single-party states. Military and personalist regimes also tend to opt for single executives and when they develop into dual executives, the variant with a popularly elected president is slightly more popular than the non-popularly elected one. Furthermore, we note that dual executives rarely develop from democratic single executive regime forms; only four such cases are

4

HOW DUAL EXECUTIVES EMERGE

83

found in the material. Finally, there are 40 cases in which dual executives have developed from other dual executives, but these are of little interest as they refer to the regime form of the mother country at the time the countries in question gained their independence. The same conclusion is valid for the three cases where a presidential form of government has given rise to a single-head executive. Let us then move on to a more detailed examination of the countries with dual executives included in the V-dem dataset. Tables 4.2–4.4 show the distribution of cases along the power dimension in the three various forms of dual executives. A significant part of the dual executives have emerged from military or personalist regimes. Findings in Table 4.2 show, that among countries with non-popularly elected presidents, military regimes were substituted by parliamentary regimes with presidents that held ceremonial powers only in Fiji (2014) and Turkey (1961). In nine cases and six countries (Bangladesh 1991, 2009, Ghana 1970, Greece 1974, Myanmar 1960, Pakistan 1972, 1988, 2008, and Sudan Table 4.2 Previous regime type and powers of president in dual executives with a non-popularly elected president Regime type at t−1 Parliamentarism, dual HOS/HOG Parliamentarism, single HOS/HOG Single-party system Military rule Personalism Absolute monarchy Monarchic oligarchy Other oligarchy Total

Powers of head of state at t 0.00

>0.00–0.99

1.00–2.99

3.00–4.99

5.00–7.00

Total

2 (3.7)

5 (9.3)

4 (7.4)

0 (0.0)

0 (0.0)

11 (20.4)

0 (0.0)

0 (0.0)

1 (1.9)

0 (0.0)

0 (0.0)

1 (1.9)

1 (1.9)

5 (9.3)

5 (9.3)

1 (1.9)

0 (0.0)

12 (22.2)

2 (3.7) 0 (0.0) 0 (0.0)

3 (5.6) 4 (7.4) 2 (3.7)

2 (3.7) 2 (3.7) 3 (5.6)

1 (1.9) 0 (0.0) 3 (5.6)

3 (5.6) 0 (0.0) 0 (0.0)

11 (20.4) 6 (11.1) 8 (14.8)

1 (1.9)

0 (0.0)

1 (1.9)

0 (0.0)

0 (0.0)

2 (3.7)

0 (0.0) 6 (11.1)

2 (3.7) 21 (38.9)

0 (0.0) 18 (33.3)

0 (0.0) 5 (9.3)

1 (1.9) 4 (7.4)

3 (5.6) 54 (100.0)

Each cell contains the absolute number of cases along with total percentages in parentheses

84

C. ANCKAR

1986), military regimes transformed into dual executives with a president that held at least some powers. The bulk of the countries where a military regime has been substituted with a dual executive have opted for a system with a popularly elected president. As indicated in Table 4.3, this category encompasses fourteen cases (Burkina Faso 2007, Central African Republic 1993, Cuba 1940, Greece 1926, Guinea-Bissau 1993, Mali 1991, Niger 1993, 1999, 2011, Peru 1956, 1963, 1980, Portugal 1976, and Turkey 1983). As shown in Table 4.4, there are also two cases where dual executives with a monarch as head of state have materialized from military regimes, namely Greece 1946 and Thailand 1983. The Greek case has been discussed above and is far from an ideal example of a transformation from a military regime to a dual executive. On the one hand, the regime classification Metaxas’ authoritarian regime is difficult as it incorporated elements of military rule, personalism and party-based rule. Second, the transformation from military rule to a constitutional monarchy occurred during a long period of time. Between 1941 and 1944, Greece was under Axis occupation and between 1944 and 1946, the Archbishop of Greece was the head of state, bearing the title Regent. Table 4.3 Previous regime type and powers of president in dual executives with a popularly elected president Regime type at t−1 Parliamentarism, dual HOS/HOG Presidentialism Single-party system Multi-party authoritarianism Military rule Personalism Total

Powers of head of state at t 0.00

>0.00–0.99 1.00–2.99

3.00–4.99

5.00–7.00

Total

0 (0.0)

2 (4.8)

0 (0.0)

0 (0.0)

0 (0.0)

2 (4.8)

0 (0.0) 1 (2.4)

0 (0.0) 3 (7.1)

1 (2.4) 2 (4.8)

0 (0.0) 0 (0.0)

2 (4.8) 2 (4.8)

3 (7.1) 8 (19.0)

0 (0.0)

1 (2.4)

0 (0.0)

1 (2.4)

5 (11.9)

7 (16.7)

0 (0.0) 0 (0.0) 1 (2.4)

1 (2.4) 2 (4.8) 9 (21.4)

3 (7.1) 1 (2.4) 7 (16.7)

4 (9.5) 3 (7.1) 8 (19.0)

6 (14.3) 14 (33.3) 2 (4.8) 8 (19.0) 17 (40.5) 42 (100.0)

Each cell contains the absolute number of cases along with total percentages in parentheses

4

HOW DUAL EXECUTIVES EMERGE

85

Table 4.4 Previous regime type and powers of monarch in dual executives with a hereditary monarch as head of state Regime type at t Powers of head of state at t −1 0.00 >0.00–0.99 1.00–2.99 Parliamentarism, dual HOS/HOG Multi-party authoritarianism Military rule Personalism Absolute monarchy Monarchic oligarchy Total

0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 2 (5.1) 2 (5.1)

3.00–4.99

5.00–7.00

Total

6 (15.4)

7 (17.9)

1 (2.6)

0 (0.0)

1 (2.6)

1 (2.6)

0 (0.0)

0 (0.0)

14 (35.9) 2 (5.1)

0 (0.0)

1 (2.6)

1 (2.6)

0 (0.0)

2 (5.1)

2 (5.1)

0 (0.0)

3 (7.7)

0 (0.0)

0 (0.0)

2 (5.1)

2 (5.1)

1 (2.6)

2 (5.1)

6 (15.4)

1 (2.6)

0 (0.0)

11 (28.2)

17 (43.6)

8 (20.5)

1 (2.6)

5 (12.8) 5 (12.8) 11 (28.2) 39 (100.0)

Each cell contains the absolute number of cases along with total percentages in parentheses

Thailand, too, is a controversial case. The country has oscillated between democracy and autocracy for half a century. Thailand’s semiconstitutional monarchic form of government has been interrupted on four occasions since 1975, but essentially, the position of the monarch has not been very much affected. The unique symbiosis between the monarchy and the military has secured the position of the monarch, both in democratic and authoritarian eras. The long reign of King Bhumibol also meant that he could gradually consolidate his position as Pater Patriae. The fact that the monarch generally has exercised influence from behind the scenes is another important factor for the stability of the monarchy, as it has meant that he has not been accountable for unpopular decisions (McCargo, 2005). Although the military and the king have been the central actors in Thai politics, it is interesting to note that there has been only one regime change directly from military rule to a dual executive with a monarch as head of state, namely in 1983, when the leader of the military regime, general (and prime minister) Prem Tinsulanonda, was able to hold on to the position as prime minister after parliamentary elections. On the other occasions where democracy has been re-introduced (1975, 1992, 2011),

86

C. ANCKAR

the military first transferred powers to a civilian authoritarian government, after which democratic elections were held. As a final observation regarding the cases of former military regimes, it is worth noting that the ensuing dual executives have been very unstable. In general, the countries have returned to authoritarian rule after a few years of democracy. The only exceptions to this rule are Greece and Portugal, where the democratic regimes have managed to survive for extensive periods of time. Transition patterns from personalist rule to dual executives resemble those between military regimes and dual executives. There is one important difference, however. Whereas there were only two cases of transformations from a military regime to a dual executive with a monarch as head of state, there are five cases of transformations from a personalist regime to a dual executive regime headed by a monarch. However, a closer look at the cases reveals that it probably makes sense to play down this finding. Three of the cases refer to transitions in Thailand, and as we have seen, in these cases, the personalist rule was merely a transition phase between military rule and semi-constitutional monarchy. The other cases in the category are Solomon Islands in 2006 and Spain 1977. In the Solomon Islands, the powers of the monarch are exercised by the Governor-General, who is formally appointed by the British monarch for a period of five years, but in reality elected by the National Assembly. In essence, the system bears more resemblance to a dual executive with a non-popularly elected president than of a semi-constitutional monarchy. In Spain, again, it is questionable if the transition occurred from a military regime or a personalistic one. Before Franco died in 1975, he had appointed the then Prince Juan Carlos as his successor. Juan Carlos climbed the throne the same year, but had to endure a short power struggle with holdovers from the Franco regime before democratic elections were held in 1977, a year during which the monarch possessed considerable powers. A new constitution was adopted in 1978, and from that year, the Spanish monarch no longer possesses any significant powers. As noted in Table 4.2, six countries have moved from the category personalism to a dual executive with a non-popularly elected president. In Austria (1946), West Germany (1949), and Italy (1946), this happened after or during a period of allied occupation following the defeat of the Axis powers in World War II. The other countries in the category are Albania (1997), Indonesia (1955), and Lebanon (1970). Finally, Croatia (2000), Georgia (2004), Madagascar (2014), Peru (1991, 2001), South

4

HOW DUAL EXECUTIVES EMERGE

87

Korea (1060), Sri Lanka (2015), and Yugoslavia (2000) have substituted personalist rule with a dual executive with a popularly elected president. A closer look at the cases with a single-party heritage confirms the link between a communist heritage and the dual executive. Altogether twenty countries with a separate head of government and head of state have emerged from single-party systems. In two of these countries, Estonia and Slovenia, the president was in possession of no powers at the time of transition. Of the remaining countries, fourteen (Albania, Belarus, Bosnia-Herzegovina, Bulgaria, Czechoslovakia, Hungary, Latvia, Lithuania, Moldova, Mongolia, North Macedonia, Poland, Russia, and Ukraine) have a past as communist single-party states. The socialist singleparty regimes of Cape Verde, Mozambique, and São Tomé and Príncipe, again grew out of anti-colonial liberation movements. The three countries chose dual executives with a popularly elected president when they democratized in the 1990s, but this choice was probably more affected by their Portuguese colonial experience than with the legacy of the single-party state. The last country on the list, Tunisia, is a somewhat more complex case. Until very recently, Tunisia has been hailed as the only example of a successful transition to democracy in the aftermath of the Arab spring. Like in the other African examples, Tunisia’s dominating party, The Democratic Constitutional Rally, traced its origin to the liberation movement for independence. Furthermore, in Tunisia too, there is colonial link to the model of the shared executive as its former mother country, France, is the prototype of semi-presidentialism. Nevertheless, it is evident that these circumstances did not generate the semi-presidential system of Tunisia. Instead, the Tunisian regime form was largely a compromise between rivalling political fractions. The elections to the Constituent Assembly in 2011 were won by The Ennahda, an Islamic party. The Ennahda favoured a parliamentary form of government, whereas their opponents feared that this would lead to the complete dominance of The Ennahda and instead proposed a presidential system. Eventually, a compromise was reached and the new constitution, adopted in 2014, provided for a ‘dyarchic’ regime; executive powers were to be shared by a prime minister accountable to parliament and a popularly elected president (Pastor y Camarasa, 2018). Subsequent developments have shown that Tunisia’s experiment with semi-presidentialism has been a complete failure. In general, prime ministers have been weak and short-lived. At the

88

C. ANCKAR

time of writing, Tunisia’s democratic era appears to have come to an end as the president has dismissed the government and suspended parliament. In eight countries and nine cases, dual executives have emerged from multi-party authoritarian states. Within this category, Romania constitutes the only example of an Eastern European country where the process of democratization was gradual; in contrast to the other countries in the region, the regime change did not occur directly from a single-party regime to a fully democratic one. Instead, The National Salvation Front, with many former communist officials in its highest positions, managed to hold on to power until 1996. In two cases, Pakistan (1950) and Lesotho (2002), a monarchic dual executive emerged from a multi-party authoritarian state. However, only in Lesotho did the monarch effectively exercise any powers. In Pakistan, the British monarch was formally the head of state but the powers of the head of state were exercised by the Governor-General, Sir Khwaja Nazimuddin, a key figure in Pakistani politics, who assumed the position of Prime Minister in 1951. Among the other cases in the category, Senegal (1993) and Taiwan (1996) have remained democratic ever since the transition to democracy, whereas Madagascar (1993), Sri Lanka (1991), and Tanzania (1996 and 2006) experienced subsequent regime changes back to authoritarian rule. Finally, it is worth noting that none of the countries for which the transformation pattern is valid has chosen a non-popularly elected president as head of state. Also, the popularly elected presidents have, on average, been very powerful; five of the seven cases, namely Senegal 1993, Sri Lanka 1991, Taiwan 1996, and Tanzania 1996, 2006, are situated in the highest power category. Findings in Table 4.4 show, that in line with the expectations of the ‘King’s dilemma’, we note that the majority of the dual executives with a monarch as head of state have emerged from absolute monarchies or monarchical oligarchies. Although most of the dual executives headed by a monarch developed from monarchical oligarchies, there are five cases of transformations directly from an absolute monarchy to a dual executive with a monarch as head of state, namely Greece (1864), Liechtenstein (1921), Nepal (1991), Tonga (2012), and Yugoslavia (1921). Furthermore, we note that altogether eleven democratic monarchies have emerged from existing monarchic oligarchies. Having formed part of the United Kingdom of Sweden and Norway, Norway became independent in 1905. The parliament elected Norway’s first King, Haakon VII, who yielded significant constitutional powers but in practice only

4

HOW DUAL EXECUTIVES EMERGE

89

possessed ceremonial ones. In the remaining cases, the monarch retained at least some of his or her powers. All but two of the cases in this category refer to Western European monarchs (Belgium 1894, Canada 1867, Denmark (1901), Italy (1919), Luxembourg (1900), the Netherlands (1896), Sweden (1911), and the United Kingdom (1885). The countries in the category with no links to European royal houses is Bhutan, which surpassed the threshold of democracy in 2009 and Japan in 1952. In the case of Japan, the transitionary period between the regimes lasted for a number of years. Before and during World War II, Japan’s regime could best be described as a monarchical oligarchy in which the military yielded significant powers. During the occupation 1945–1952, Japan held parliamentary elections and the country had a national government from the year 1947. However, the authority of the government was subordinate to the American occupation forces, more specifically to General Mac Arthur. Interestingly enough, absolute monarchies do not always transform into constitutional or semi-constitutional monarchies when they democratize; in eight republics, a dual executive has arisen from an absolute monarchy. The bulk of the cases where the monarchy has survived the transition from autocracy to democracy consists of cases where countries have received their independence from a country ruled as an absolute monarchy. The cases in question refer to a number of Eastern or Central European countries which received their independence in the aftermath of World War I, namely Austria and Czechoslovakia, which emerged from the Austro-Hungarian Empire, along with Finland, Latvia, and Lithuania, all of which became independent from Russia, and, finally, Poland, which arose from the ruins of the German, Austro-Hungarian, and Russian empires. In only two instances has a similar transformation occurred within independent countries, namely in Spain 1931 and in Nepal 2008. However, regarding Spain, it is questionable if this transformation really occurred directly from an absolute monarchy or not. During the period 1923–1930, Spain had effectively been ruled by General Miguel Primo de Rivera, who enjoyed the support of the King. In 1930, de Rivera was forced to resign and King Alfonso XIII resumed his position as the central political figure in Spanish politics (albeit far from a monarch with unlimited powers) for a short period of time before he was deposed and the Second Spanish Republic installed. The other case in question, Nepal, is more unambiguous. The country had initially passed the threshold of

90

C. ANCKAR

democracy in 1991, when King Birendra, facing a rebellion, reluctantly had to accept a new constitution, which transformed Nepal from an absolute monarchy to democratic dual executive with a monarch as head of state (Nepal et al., 2011, 887). In a like manner, only two countries have transformed from a monarchic oligarchy into a dual executive with a president as head of state. The cases in question are Germany, which adopted its semi-presidential constitution in 1919, and Portugal, where the monarchy was overthrown in the 5 October 1910 revolution and substituted with a republican form of government. A final observation concerns transitions to dual executives from other democratic regime forms. The evidence at hand shows that this pattern of transformation is extremely rare. Dual executives have emerged from presidential forms of government only in Brazil in 1961 and in Kenya in 2008. In addition, the case of East Timor is included in the category, as the country has had a semi-presidential form of government (by any definition) ever since it gained its independence in 2002 from Indonesia, the form of government of which is presidentialism. There is also one example of a country that has experienced a transformation from a parliamentary system in which the head of state has also functioned as head of government, responsible to parliament. The case in question is France during the third republic, where the president was elected by, and responsible to, the National Assembly until the constitutional laws were adopted in 1875. Moving on, in terms of the powers of the head of state, there is an interesting difference between popularly elected and non-popularly elected presidents. A comparison of the figures presented in Tables 4.2 and 4.3 show, that whereas newly installed non-popularly elected presidents tend to have weak or moderate powers, popularly elected presidents tend to possess strong powers. It is remarkable that over 40% of the cases where a dual executive with a popularly elected president has emerged from another regime form are situated in the highest power category. The corresponding figure for non-popularly elected presidents is less than 8%. Similarly, whereas half of the cases with non-popularly elected presidents are found in the two weakest power categories, only less than one fourth of the cases with popularly elected presidents are situated within the same categories. Indeed, the evidence tells us that there is only one case where a democratic regime with a completely powerless, popularly elected president has been installed, namely Slovenia in 1991.

4

HOW DUAL EXECUTIVES EMERGE

91

The finding is well in line with the expectation that popularly elected presidents are in possession of more powers than presidents elected by parliaments. At the same time, this is not a rule without exceptions. In two countries, Namibia, and Pakistan (1972, 1988, 2008), democratic regimes with extremely powerful, non-popularly elected presidents have been installed. However, the case of Namibia merely strengthens the case for an association between the popular election of the president and strong powers, as only the first election of the president was conducted by parliament in 1990. Pakistan 1972, again, is an even more dubious case. Having been ruled by a military regime, the country adopted an interim constitution in 1972. The President, Zulfiqar Ali Bhutto was handed the presidency when the military regime left power in December 1971. The interim constitution gave the president wide powers. There was no prime minister, but the ministers of the government (but not the president) where all subject to a parliamentary vote of no confidence. The following year, the country adopted a new constitution where powers were largely transferred from the president to the prime minister, who was accountable to parliament. In general, Pakistan has been oscillating between democracy and military rule throughout its existence. The transfers to civilian rule in 1988 and 2008 follow the same pattern in the sense that the powers of the head of state were very large during the first year as a democracy, but were thereafter reduced quite substantially. There is an interesting difference between the authoritarian regime categories regarding the distribution of cases among the power categories in countries with popularly elected presidents. Whereas the vast majority of military- and personalist regimes transform into popularly elected presidents with wide powers, single-party systems tend to produce weak and strong presidents in fairly equal proportions. In countries with nonpopularly elected presidents, the distribution of cases along the power dimensions does not vary between the authoritarian regime categories. As shown in Table 4.3, in five instances, countries have introduced a regime with a non-popularly elected president exercising powers in the second highest power category. However, these cases too, most often refer to exceptional and turbulent periods in the history of the countries, periods during which non-elected actors have held the position as head of state exercising wide powers for a short, transitionary stage. When Belarus became independent in 1991, the Chairman of the Supreme Council, Stanislav Shushkevich, was elected head of state.

92

C. ANCKAR

However, the new constitution, adopted in 1994, provided for a popularly elected president and presidential elections were held the same year, bringing Alexander Lukashenko to power. Finland proclaimed independence from Soviet Russia in 1917. During the drafting of the constitution, the country was ruled by a head of state, titled Regent, alongside a prime minister, responsible to parliament. The constitution, adopted in 1919, provided for a semi-presidential system, with a rather powerful, popularly (but indirectly) elected president. However, as the constitution was adopted, it was also decided that the first president of the republic was to be elected by parliament. The first popular election of president in Finland occurred six years later, in 1925. Poland, again, became independent in November 1918. During the last weeks of 1918, its first head of state, Józef Pilsudski, yielded farreaching powers. However, with the adoption of the so called small constitution in 1919, his powers were significantly reduced. Spain, on its part, transited from monarchy to a republic in 1931 and during this year, its first President, Niceto Alcalá-Zamora was deemed by the V-dem experts to hold significant prerogatives not only in the legislative sphere but also with regard to cabinet composition. After 1931, however, the president was no longer considered to exercise influence over cabinet formation and termination. In Ghana, finally, military rule ended in 1970. The constitution adopted one year earlier, provided for a parliamentary system with a president as head of state, elected by parliament. The V-dem experts have deemed the president to be in possession of extensive powers in the legislative field but also with regard to cabinet appointment and dismissal of ministers. Ghana’s experiment with democracy ended already in 1972, when the military took power in a bloodless coup. A corresponding pattern is not discernable for regimes with weak, popularly elected presidents. As shown in Table 4.3, in addition to the completely powerless president of Slovenia, we find nine cases of popularly elected presidents in the second weakest power category. Bosnia-Hercegovina (1996–2019), Croatia (2000–2019), Montenegro (2006–2019), and Romania (1991–2019), are marked by exceptional stability in the sense that the head of state has continued to be popularly elected at the same time as the powers of the president have changed only marginally from the introduction of the dual executive to date. In Cape Verde (1991–2019) and Serbia (2006–2019), the powers of the president have varied somewhat more whereas Moldova is different in the sense that the country has switched between popularly and non-popularly elected

4

HOW DUAL EXECUTIVES EMERGE

93

presidencies. The only country that shows a different pattern is Turkey. The country moved from a military regime to democracy in 1983, when parliamentary elections were held. Although powers were in the hands of the prime minister, the leader of the military regime, Kenan Evren, retained the position as head of state. However, whether or not Evren was popularly elected or not is up to the assessor to decide. The democratic constitution, which contained an article stating that Evren was to remain as president until 1989, had been approved in a referendum in 1982, but any real presidential election had not been held. Finally, Armenia made the transition to democracy in 2019, but it is yet too early to speculate whether democracy will consolidate or not. With respect to the category of dual executives with a hereditary monarch as head of state, we note that most of the countries in Table 4.4 have inherited the system in question when they received their independence. Indeed, thirteen of the fourteen cases are made up of former British colonies, all of which not only inherited the monarchic form of government but also the British monarch in person as head of state. The only exception to this rule is Laos, which became independent from France in 1954. Finally, we also find that in the category of dual executives with a hereditary monarch as head of state, the distribution of the cases along the power categories follows quite nicely the normal distribution. The plurality of the cases are situated within the centremost category, whereas the smallest number of cases are found in the two most extreme categories. The transition to a democratic dual executive with the strongest monarch occurred in Liechtenstein in 1921, whereas Norway (1905) and Japan (1952) constitute the cases with completely powerless monarchs at the time of transition to democracy (or, in the case of Norway, at the time of independence).

4.4

Diffusion

The term diffusion ‘indicates the spread of a practice within a social system, where the spread denotes flow or movement from a source to an adopter typically via communication, role modeling, and/or coercion’ (Wejnert, 2014, 35). Such practices can spread through many channels and along different types of networks. Accordingly, Simmons et al. (2007) identify four mechanisms of diffusion, emulation, coercion, competition, and learning.

94

C. ANCKAR

Within the framework of the present study, focus is on two central forms of interaction, namely what Wejnert (2014, 43) refers to as ‘hegemonic imposition’ and spatial proximity. According to Wejnert (2014, 43), there are many contexts in which hegemonic imposition can occur but she explicitly mentions ‘colonial networks, military pacts … [and] economic unions’. For the purpose of the present study, the colonial networks are of particular relevance as the regime choices countries made when they received their independence often were modelled after the regimes of their mother countries. Indeed, especially in former British colonies, a democratic regime form was established in the colonies before they received their independence (e.g. Smith, 1978). Geographic settings are of importance as well. Countries are likely to be influenced by policy choices made by countries situated in the same regional contexts. In the field of comparative politics, spatial proximity is frequently used in regression models where the ambition is to explain democratization. Several studies have shown that processes of democratization occur in spatial clusters and/or temporal waves (e.g. O’Loughlin et al., 1998; Doorenspleet, 2005; Houle et al., 2016; Teorell, 2010). It is therefore natural to make the assumption that geographic diffusion is important also when countries choose what kind of democratic regime forms to adapt. In fact, such an approach is motivated simply by a quick glance at the distribution of the presidential regime form, which is largely confined to the American continent. With regard to other democratic regime forms, the evidence is perceptible. In particular, it is an open question to what extent our two dimensions, mode of selection and powers of the head of state are affected by constellations of the same features in neighbouring countries. 4.4.1

The Colonial Heritage

4.4.1.1 Colonial Heritage and Regime Choice The relationship between former colonies and their former mother countries is of crucial importance for the present study as the bulk of the countries in the population have emerged from existing state formations. Most of these countries received their independence in connection with the process of decolonization but a substantial number of states have taken other routes towards independence. In Eastern Europe, a great number of states emerged as a consequence of the disintegration of the

4

HOW DUAL EXECUTIVES EMERGE

95

Soviet Union and Yugoslavia and some 70 years earlier, a number of countries were created on the ruins of fallen empires in the aftermath of World War I. A handful of countries, such as the Czech Republic and Slovakia, Pakistan and Bangladesh, have been formed as a consequence of a split of the mother entity into two parts. Irrespective of how the countries were formed, they were immediately faced with the task of forming a constitution and, thereby, choosing a regime form. When making this choice, it is only natural that the influence of the country from which the newly formed entity emerged was substantial. Already in 1796, Adam Smith made the observation that the political institutions of the possessions overseas closely resembled the institutions of the metropolitan power (cited in Greene, 2007: 171). In Africa, the first states that received independence have been said to have ‘bequeathed political institutions patterned on those of their former colonial masters’ (Keller, 1995, 166), whereas Elgie (2012, 774), in a similar line of reasoning noted that ‘the French metropolitan model was transposed onto the African context’. Many times, emerging states had very little to say when the constitutional provisions were drafted; notably, a democratic form of government often constituted a direct requisite for the granting of independence. Many times, however, the first constitutions of countries were not long-lived. A telling example is the Republic of Congo, which had three constitutions, all of which were modelled after the French constitution, during its first four years as an independent nation (Lane, 1996, 78). Nevertheless, the legacy of the former colonial power tends to remain strong and as countries revert to democratic rule after long periods of authoritarian rule, the constitutions of the former colonial power often serve as the natural point of reference when the new democratic constitution is framed (Cabanis & Martin, 2005; du Bois de Gaudusson, 2009; Elgie et al., 2011, 271). It is therefore important to account for the colonial heritage of countries, not only with regard to the first regime form countries introduce as they gain independence, but also for subsequent regime variants. The aforesaid also means that countries that have received its independence from non-democratic countries can be expected to frame democratic constitutions modelled after their former colonial power at later stages in history, when the former colonial power has made the transition to democracy.

96

C. ANCKAR

4.4.1.2 Operationalizing Colonial Heritage It is far from self-evident how the connection to the former mother country should be measured. First, colonial heritage is a problematic concept and can be conceived of either in a narrow or in a broad sense. Here, colonialism is understood in its broadest sense, much in line with the definition provided by Merriam-Webster dictionary (https:// www.merriam-webster.com/dictionary/colonialism), i.e. ‘domination of a people or area by a foreign state or nation’. The definition includes under the umbrella term colonialism, relations between the metropolitan centre and subordinate nations in historical empires, such as the Ottoman Empire and the Austro-Hungarian Empire. Within the framework of the present study, I also include similar relations within the Soviet Union and Yugoslavia. However, in cases where an existing country has split into two components, I do not regard the newly born countries as ex-colonies of the former entity. In other words, the Czech and Slovak republics are not former colonies of Czechoslovakia, and current day Pakistan and Bangladesh are not former colonies of Pakistan prior to the split between Bangladesh and Pakistan in 1971. Another question of concern is the length of time that has elapsed since the granting of independence. Evidently, a country that gained its independence a hundred years ago is less likely to be influenced by the political system of its former mother country than a country that gains its independence today. However, based on the discussion conducted above, we also note that when countries revise their constitutions they often frame them in accordance with the constitutions of their former mother countries. When determining if a country has a colonial legacy or not, it thus makes sense to apply the criterion of heritage rather loosely. Accordingly, I stipulate, as a general point of reference, that a country’s choice of a dual executive is deemed to have been plausibly affected by its former mother country insofar the country in question had been independent for less than 60 years at the time it passed the threshold of democracy for the first time. In addition, the time during which the country spent as a colony is of importance. A country that was ruled as a colony for a hundred years is much more likely to have been influenced by its mother country than a country which only was under foreign rule for, say five years. Therefore, I stipulate that the colonial rule must have lasted for at least a decade. Finally, some countries have been under the rule of several colonial powers. For instance, the administration of the German colonies

4

HOW DUAL EXECUTIVES EMERGE

97

was transferred to other colonizing powers under a League of Nations mandate after World War I. In such cases, I have generally regarded the last colonial power as the relevant one. However, in some cases, more qualitative assessments have been made. For instance, East Timor, which was under Portuguese rule for over three centuries before it came under Indonesian occupation for 24 years, is regarded as a former Portuguese colony rather than an Indonesian one. A handful of countries have been under the rule of several colonizing powers at the same time. This is the case, for instance, with Poland, which received its independence in 1918 from the German Empire, the Austro-Hungarian Empire, and Soviet Russia. These cases are included among countries with no colonial heritage in the empirical analyses. The same strategy is applied for countries like Somalia and Vanuatu, which were under the rule of multiple colonial powers at the same time. 4.4.2

Geographical Proximity

The second pattern of diffusion puts emphasis on geography, more specifically on Tobler’s ‘First Law of Geography’, which states that ‘everything is related to everything else, but near things are more related than distant things’ (Tobler, 1970, 236). The basic assumption is that countries are particularly likely to adopt practices that are popular in neighbouring states. A number of studies operating with different datasets and applying varying research designs have shown that the democratic form of government, democratic transitions, as well as democratic breakdowns, are clustered in time and space (Beck et al., 2006; Bernhard et al., 2004; Brinks & Coppedge, 2006; Darmofal, 2015; Doorenspleet, 2005; Gleditsch & Ward, 2006; Houle & Kayser, 2019; Houle et al., 2016; Kopstein & Reilly, 2000; Li & Thompson, 1975; O’Loughlin et al., 1998; Przeworski et al., 2000; Starr & Lindborg, 2003; Wright, 2008). In other words, countries tend to adjust their regimes to match the surrounding neighbours (Brinks & Coppedge, 2006). For example, non-democratic countries whose neighbours are undergoing a transition from autocracy to democracy also tend to democratize (Teorell, 2010) and autocratic countries with strong linkages to democracies tend to become more democratic (Levitsky & Way, 2006). The fact that we are operating with two dimensions of diffusion means that it is possible to compare the importance of geographical proximity with colonial heritage. In other words, it is very possible indeed,

98

C. ANCKAR

that a country which has belonged to a certain colonial power is more influenced by political developments in countries with a similar colonial background than by political occurrences in neighbouring countries with another, or no, colonial heritage. Indeed, a common colonial heritage can establish cultural similarities and thereby strong ties between countries, which are geographically located very far from each other (Acemoglu et al., 2001; Hariri, 2012; Wejnert, 2005, 2014). The connection to the mother country is especially strong in cases where the former colonies have inherited crucial ethnocultural characteristics, such as language and religion, from their former mother country. Accordingly, by almost any measure, the British influence or legacy is much stronger in Australia and New Zealand than in its closest neighbours on the European continent. Contrary to the case with colonial heritage, operationalizing spatial proximity is a less complicated enterprise. Regional diffusion patterns are assessed by calculating the share of countries in the same geographical context with the similar type of dual executive (i.e. either popularly elected president, non-popularly elected president, or hereditary monarch) as the one introduced in the respective countries. The variable is lagged by one year. The main concern is how to define the regional contexts in question. In this respect, a recent article by Åberg and Denk (2020, 89–91) provides some guidance as it compares how different operationalizations of regions affect the relationships between regime density and the introduction of democratic regime forms. As it turns out, when explaining the introduction of both parliamentary and semi-presidential regimes, the strongest associations are detected when the countries of the world are divided into 16 regions. Conveniently, this classification corresponds quite well with the variable Region (geographic) (E) (e_regiongeo), which is included in the V-dem dataset. Here, the countries of the world are confined to nineteen geographic regions and I consequently make use of this variable when defining the regional contexts.

4.5 Diffusion and Introduction of Dual Executives---Empirical Patterns 4.5.1

Colonial Heritage and Regime Choice

Table 4.5 presents a compilation of the regime choices countries have made at the time of independence or at the time of transformation from an authoritarian to a democratic form of government. Again, in order to

4

HOW DUAL EXECUTIVES EMERGE

99

Table 4.5 Colonial heritage and democratic regime form Colonial heritage

Head of state not popularly elected

Head of state popularly elected

Head of state inherits his/her position

None Britain France Soviet Union Yugoslavia The Netherlands Portugal Soviet Russia Spain USA Austro-Hungarian Empire Japan Belgium Denmark Ottoman Empire South Africa Total

23 (24.7) 16 (24.2) 2 (13.3) 4 (40.0) 2 (25.0) 2 (40.0) 0 (0.0) 3 (75.0) 0 (0.0) 0 (0.0) 1 (0.0)

10 (10.8) 5 (7.6) 9 (60.0) 5 (50.0) 6 (75.0) 0 (0.0) 5 (100.0) 0 (0.0) 1 (20.0) 0 (0.0) 0 (0.0)

19 27 1 0 0 0 0 0 0 0 0

(20.4) (40.9) (6.7) (0.0) (0.0) (0.0) 80.0) (0.0) (0.0) (0.0) (0.0)

41 (44.1) 18 (27.3) 3 (20.0) 1 (10.0) 0 (0.0) 3 (60.0) 0 (0.0) 1 (25.0) 4 (80.0) 5 (100.0) 0 (0.0)

93 (100.0) 66 (100.0) 15 (100.0) 10 (100.0) 8 (100.0) 5 (100.0) 5 (100.0) 4 (100.0) 5 (100.0) 5 (100.0) 1 (100.0)

0 (0.0) 0 (0.0) 1 (100.0) 0 (0.0) 1 (100.0) 55 (24.7)

1 (50.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 42 (18.8)

0 (0.0) 0 (0.0) 0 (0.0) 1 (100.0) 0 (0.0) 48 (21.5)

1 (50.0) 1 (100.0) (0.0) 0 (0.0) 0 (0.0) 78 (35.0)

2 (100.0) 1 (100.0) 1 (100.0) 1 (100.0) 1(100.0) 223 (100.0)

Single head of state and government

Total

Each cell returns the absolute number of cases with row percentages in parentheses. Pearson Chi-square 129.349 (Sig. 0.000)

provide a comprehensive picture of the regime choices made, I include, in addition to the three variants of dual executive systems, a category made up of democracies with a single head of state and government. The results largely confirm findings from studies in field of semipresidentialism in the sense that countries that have been under French or Portuguese rule to very large extents adopt dual executives with popularly elected presidents as heads of state. Nevertheless, regarding countries formerly ruled by France, there are many exceptions from this rule. Benin and Comoros adopted a presidential form of government when the countries transited to democracy in 1991 and 2006, respectively. The Republic of Congo had a parliamentary system with a single head of state and government for a transitionary stage after gaining independence in 1960. The country changed to a presidential form of government in 1962 and then slid into the category of autocracies the following year. Laos, again,

100

C. ANCKAR

was a semi-constitutional monarchy during its first five years as an independent state (1954–1958), after which the democratic era ended. Both Lebanon (1970) and Tunisia (2012) had non-popularly elected presidents when they democratized for the first time, but in Tunisia this was an interim arrangement during the period of the drafting of the new constitution, which provided for a popularly elected president. The Eastern European heritage is also strong. All but one countries that have emerged from the Soviet Union or Yugoslavia have chosen dual executives with a president as head of state. The only exception is Belarus, which transformed from a dual executive to a presidential form of government in 1994. For what it is worth, Belarus returned to a dual executive in a semi-presidential variant already in 1996, but at that year, the powers had been concentrated in the hands of President Lukashenko to such an extent that the country no longer qualified as a democracy. In this group, too, the variant with the popularly elected president is more popular than the non-popularly elected one. Although six out of 18 countries had a non-popularly elected president when they gained independence three of these cases (Belarus, Lithuania, and North Macedonia) refer to interim presidents elected by the parliament or the constituent assembly. Only in Estonia, Latvia, and Kosovo did the non-popular election of the president become established in the constitution. Dual executives with a monarch as a head of state have primarily been introduced in former British colonies at the time of their independence. In most of these cases, the British monarch formally remained the head of state, although her duties were performed by a Governor-General. A substantial number of the dual monarchies have also been established in countries without, or with a mixed, colonial past. Most of these cases refer to the monarchies in continental Europe, although examples are also found in Asia, namely Bhutan, Japan, Nepal, and Thailand. Although constitutional monarchies are closely associated with a British legacy, there is a substantial number of former British colonies with a president as head of state. Most of these presidents are non-popularly elected, and in some cases, such as Botswana and Kiribati, they function as both head of state and head of government, subject to a parliamentary vote of no confidence. Altogether eight countries with a British colonial heritage have introduced dual executives with non-popularly elected presidents on various occasions, namely Bangladesh, Fiji, Ghana, India, Israel, Myanmar, Pakistan, and Sudan. Although the group of countries is quite heterogeneous, it is noteworthy that it consists of countries that tend to

4

101

HOW DUAL EXECUTIVES EMERGE

be large and/or culturally more distant to Britain than, for instance, its former colonies in the Caribbean. There are also five instances (Kenya 2008, Sri Lanka 1991, 2016, and Tanzania 1996, 2006), where systems with a popularly elected president have been introduced after periods of autocratic rule in former British colonies. Finally, the most even distribution of systems along the four democratic regime categories are found among countries without a colonial past. It is noteworthy, though, that nearly half of these cases are situated in the regime category encompassing single executives. Most of these transformations refer to transitions from autocratic rule to a presidential form of government in Latin America. Tables 4.6–4.8 provide information on how the colonial past is linked to the five power categories in the three dual executive variants. Overall, Table 4.6 Colonial heritage and powers of the president in dual executives with a non-popularly elected president Colonial heritage

Powers of head of state 0.00

None Britain France Soviet Union Soviet Russia Yugoslavia The Netherlands Austro-Hungarian Empire Denmark South Africa Total

2 (3.7) 2 (3.7) 0 (0.0) 1 (1.9) 0 (0.0) 1 (1.9) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 6 (11.1)

>0.00–0.99

1.00–2.99

3.00–4.99

5.00–7.00

Total

10 (18.5)

9 (16.7)

2 (3.7)

0 (0.0)

6 (11.1)

3 (5.6)

1 (1.9)

3 (5.6)

2 (3.7)

0 (0.0)

0 (0.0)

0 (0.0)

23 (42.6) 15 (27.8) 2 (3.7)

1 (1.9)

1 (1.9)

1 (1.9)

0 (0.0)

4 (7.4)

0 (0.0)

2 (3.7)

1 (1.9)

0 (0.0)

3 (5.6)

1 (1.9)

0 (0.0)

0 (0.0)

0 (0.0)

2 (3.7)

1 (1.9)

1 (1.9)

0 (0.0)

0 (0.0)

2 (3.7)

0 (0.0)

1 (1.9)

0 (0.0)

0 (0.0)

1 (1.9)

0 (0.0)

1 (1.9)

0 (0.0)

0 (0.0)

1 (1.9)

0 (0.0)

0 (0.0)

0 (0.0)

1 (1.9)

1 (1.9)

21 (38.9)

18 (33.3)

5 (9.3)

4 (7.4)

54 (100.0)

Each cell contains the absolute number of cases along with total percentages in parentheses

102

C. ANCKAR

the findings provide little additive information. Table 4.6 show, that among countries with non-popularly elected presidents, more than 70 per cent of the cases have a British colonial heritage or lack a colonial heritage altogether and here, all power categories are represented. Cases are also fairly well distributed along the different power categories in entities with a Soviet or Soviet Russian heritage. Among the countries with a popularly elected president (Table 4.7), cases are concentrated towards the highest power category. This pattern is consistent for countries that have a former French, British, or no colonial legacy. Former Portuguese colonies are mainly situated in the middle categories whereas the presidents of the countries that have emerged from Yugoslavia tend to be located in the weaker power categories. For ex-Soviet republics, we detect the same pattern that was observed with regard to non-popularly elected presidents, i.e. the countries are fairly evenly distributed along the different power categories. Finally, regarding monarchies (Table 4.8), the number of cases is fairly low. On the whole, however, the cases follow the normal distribution, with a concentration of cases in the middle, and with declining numbers as we move towards the endpoints. Table 4.7 Colonial heritage and powers of the president in dual executives with a popularly elected president Colonial heritage

Powers of head of state

None France Britain Portugal Soviet Union Yugoslavia Japan Spain Total

0 0 0 0 0

(0.0) (0.0) (0.0) (0.0) (0.0)

2 0 0 1 2

1 0 0 1

(2.4) (0.0) (0.0) (2.4)

4 (9.5) 0 (0.0) 0 (0.0) 9 (21.4)

0.00

>0.00–0.99 (4.8) (0.0) (0.0) (2.4) (4.8)

1.00–2.99 1 1 0 3 1

(2.4) (2.4) (0.0) (7.1) (2.4)

1 (2.4) 0 (0.0) 0 (0.0) 7 (16.7)

3.00–4.99 2 2 1 0 1

(4.8) (4.8) (2.4) (0.0) (2.4)

0 (0.0) 1 (2.4) 1 (2.4) 8 (19.0)

5.00–70.0

Total

5 (11.9) 6 (14.3) 4 (9.5) 1 (2.4) 1 (2.4)

10 (23.8) 9 (21.4) 5 (11.9) 5 (11.9) 5 (11.9)

0 0 0 17

4 (9.5) 1 (2.4) 1 (2.4) 42 (100.0)

(0.0) (0.0) (0.0) (40.5)

Each cell contains the absolute number of cases along with total percentages in parentheses

4

HOW DUAL EXECUTIVES EMERGE

103

Table 4.8 Colonial heritage and powers of the monarch in dual executives with a monarch as head of state Colonial heritage None Britain Ottoman Empire France Total

Powers of head of state 0.00

>0.00–0.99

1.00–2.99

30.0–4.99

5.00–7.00

Total

2 (5.1) 0 (0.0) 0 (0.0)

2 (5.1) 9 (23.1) 0 (0.0)

8 (20.5) 7 (17.9) 0 (0.0)

6 (15.4) 2 (5.1) 1 (2.6)

1 (2.6) 0 (0.0) 0 (0.0)

19 (48.7) 18 (46.2) 1 (2.6)

0 (0.0) 2 (5.1)

0 (0.0) 11 (28.2)

1 (2.6) 17 (43.6)

0 (0.0) 8 (20.5)

0 (0.0) 1 (2.6)

1 (2.6) 39 (100.0)

Each cell contains the absolute number of cases along with total percentages in parentheses

4.5.2

Spatial Proximity and Regime Choice

Table 4.9 returns results from three separate logistic analyses, where the choice of the respective dual executive variants are regressed on the regional density of the same form of dual executive one year prior to the adoption of the regime form in question. Results show that dual executives with popularly elected presidents and dual executives with a monarch as head of state are statistically related to the prevalence of the respective regime variants in the geographical context where they are located. In countries with non-popularly elected presidents, on the other hand, the association barely fails to reach the level of statistical significance (p = 0.059). For all regime variants, the directions of the associations are in line with our assumptions, i.e. the likelihood of choosing a particular dual executive type increases with the popularity of the said variant in neighbouring countries. However, the odds ratios are very close to 1, indicating that the effect is very small.

4.6 4.6.1

The Comprehensive Picture

The Interrelatedness of Regime Form and Diffusion

The bivariate analyses have shown that both the previous regime form and diffusion mechanisms are relevant for the introduction of dual executives. The final step of the empirical analyses is to assess to what extent these relationships are altered when we include all the independent variables in multivariate analyses. However, before we proceed to this part of

0.038** (0.012) 1.038 −1.756*** (0.206) 0.173 0.071** 223

Head of state popularly elected

In each cell, regression coefficients are listed first, with standard errors in parentheses, followed by odds ratios ***p < 0.001, **p < 0.01, *p < 0.05

Nagelkerke R 2 N

Constant

0.021 (0.011) 1.021 −1.296*** (0.187) 0.274 0.023 223

Head of state not popularly elected

0.028*** (0.007) 1.028 −1.716*** (0.210) 0.180 0.105*** 223

Hereditary monarch as head of state

Similar regime density and choice of dual executives. Logistic regression analyses

Similar regime density (lag 1)

Table 4.9

104 C. ANCKAR

4

HOW DUAL EXECUTIVES EMERGE

105

the analyses, it is necessary to reflect on to what extent previous regime form, colonial heritage, and geographic proximity can be separated for the purpose of conducting statistical analyses. Although our conception of dual executive is far broader than that of semi-presidentialism, the theoretical propositions that have been put forward with respect to the introduction of semi-presidentialism is highly illustrative of the problem of isolating the effects of the independent variables. The spread of semi-presidentialism occurred in two phases: one extremely slow phase, which began in 1919 with the framing of the constitutions of Finland and Weimar Germany, and another, rapid one, which began 70 years later with the fall of the Berlin Wall and continues to date. The first two forms of semi-presidentialism had widely different fates. The Weimar-republic was short-lived, unstable, and conflict-ridden, and eventually had to succumb to Hitler’s Machtergreifung in 1933. The Finnish variant, on the other hand, proved to work rather well. Its vaguely formulated constitution has been likened with a ‘smorgasbord’, from which the Finnish President could choose those prerogatives he wished to use (Anckar, 2000, 11). However, after President Urho Kekkonen’s long term in office, (1956–1981), during which he gradually (and ruthlessly) concentrated more and more power into his own hands, it was widely agreed that it was necessary to reduce the powers of the president. The path towards a parliamentary system started in the 1980s and continues to date. Today, the Finnish President has merely ceremonial powers, although he or she continues to exercise influence in the field of foreign policy. During the subsequent decades, parliamentary systems with a popularly elected president were adopted in various parts of Europe (Austria, Ireland, Iceland, and France). However, similar systems were also introduced for short periods of time in the Americas, namely in Cuba (1940–1952) and in Brazil (1961–1962). A striking feature is that these early variants of semi-presidentialism were chosen independently from each other and for different reasons. Nothing suggests that the countries that adopted semi-presidentialism at this early stage in history were influenced by other semi-presidential countries, regime transitions, geographic proximity, or colonial heritage; instead, these early variants of semipresidentialism were ‘dispersed and seemingly unrelated’ (Elgie 2011, 7). With regard to the semi-presidential surge that began in the 1990s, the picture is completely different. In his exposition of the global distribution

106

C. ANCKAR

of semi-presidentialism, Wu (2011: 22–24) identifies two main clusters and several subclusters of semi-presidential regimes installed during the time period in question. The first cluster consists of post-Leninist states and the second one of postcolonial states. On a theoretical level, the distinction between post-Leninist states and postcolonial states is important as it suggests that different explanations underlie the choice of the semi-presidential regime in the two categories of countries. For postLeninist states, the assumption is that semi-presidential regimes emerge as a consequence of the parallel power structures that characterize partybased authoritarian regimes. For post-colonial states, on the other hand, we assume that the semi-presidential constitution has been adopted more or less by copying the constitution of the former mother country, i.e. by a process of diffusion. Recall that the bivariate analyses indicated that both a former singleparty regime and a Soviet/Yugoslavian heritage were strongly linked to the adoption of dual executives. However, since all the former Soviet and Yugoslav republics have emerged from single-party systems, the colonial heritage and a previous single-party regime coincide. It is not easy to make an assessment of whether the colonial heritage or previous regime form carries more weight for explaining dual executives in these units. On the one hand, there is reason to assume that the Soviet/Yugoslavian heritage carries less weight than the previous regime form. None of the two mother countries were democratic at the time the former republics proclaimed their independence. In other words, there was no democratic constitution with a dual executive to ‘import’ from either the Soviet Union or Yugoslavia. In addition, in many of the newly established countries, the former mother country was not in high esteem and thus hardly a natural model to take examples from when the constitutions were to be framed. On the other hand, however, the Soviet/Yugoslavian heritage is of a different kind, than the ones existing between the European colonial powers and their colonies overseas. Countries that have belonged to a large political entity, which has later on disrupted do not, in a strict sense, have a ‘mother country’, from which policies or practices are transported to its former possessions. In these cases, the colonial legacy supposedly takes another form. The countries share a common history and thus form a natural ‘cluster’. In this cluster, they do not look for one source of inspiration but are affected by policy choices made by all the other entities in the cluster.

4

HOW DUAL EXECUTIVES EMERGE

107

One way of coming to grips with the problem is to rerun the analyses conducted in Table 4.1 by excluding the Soviet or Yugoslavian countries. If the relationship between single-party legacy and choice of dual executive persists, it is an indication that colonial heritage carries less weight than regime type for explaining why dual executives emerge. Indeed, results show that with regard to former single-party states, the patterns remain largely unaltered when the ex-Soviet and Yugoslavian republics are excluded: 58% of the cases have dual executives with non-popularly elected presidents, 25% have popularly elected presidents, 8% have a dual executive with a monarch as head of state, and 8% have a single head of state and government. Nevertheless, we cannot completely ignore the possibility that networks between former Soviet or Yugoslavian states are important for explaining regime choices in these particular settings. To complicate matters further, we also need to ponder on the interrelatedness between a single-party legacy and regional diffusion in the Eastern European context. Studies where the previous regime form is contrasted with diffusion mechanisms are rare, but not completely absent. In a recent study, Goldring and Greitens (2020, 344) found that ‘including regime type [in statistical models] significantly reduces the effect of geographic proximity, although regimes may be influenced more by breakdowns or democratizations in similar regimes that are geographically proximate than they are by transitions in similar regimes that occur further away’. In Eastern Europe, regime transformations occurred from single-party systems to dual executives with a president as head of state. At the same time, the countries were geographically located close to each other. Consequently, we can ask if the choice of a new regime form in an Eastern European country was more affected by the country’s previous singleparty system or by the fact that a neighbouring country had recently adopted that particular system. The problem should not be overstated, though. For one thing, not all Eastern European countries adopted the same type of dual executives; both systems with popularly and nonpopularly elected presidents were introduced. Second, the countries did not pass the threshold of democracy at the same time, which means that the density of the dual executive forms varies across time. Nevertheless, we cannot completely disregard the fact that in an Eastern European context, geographic proximity and a single-party heritage are closely connected.

108

C. ANCKAR

4.6.2

Previous Regime Form, Diffusion, and Dual Executives—Multivariate Findings

Bearing in mind the potential problems of multi-collinearity discussed in the previous section, I shall now proceed with some multivariate analyses. The analyses are conducted by means of logistic regression analyses, conducted separately for the three dual executive variants. The first task is to determine how the variables should be categorized for the purpose of the multivariate analyses. Regarding the previous regime form, bivariate analyses showed that some authoritarian regime forms were more clearly connected to ensuing variants of dual executives than others. First, it makes sense to operate with a category consisting of former monarchies, i.e. absolute monarchies and monarchic oligarchies. As we recall, monarchic oligarchies in particular were likely to transform into dual executives with a hereditary head of state. At the same time, it must be admitted that this association between the previous and the ensuing regime forms is of somewhat limited interest as the difference between an autocratic monarchic oligarchy and a democratic constitutional or semi-constitutional monarchy is not profound. Second, in line with our theoretical expectations, single-party regimes gave rise to dual executives with a president as head of state. The empirical evidence showed that multi-party authoritarian systems were more likely to develop into single executives than dual executives. On the other hand, in cases where multi-party authoritarian systems did develop into dual executives, they tended to transform into systems with a popularly elected president. I therefore operate with a broad category labelled party-based rule, consisting of former single-party states and multi-party authoritarian states. This strategy is also motivated by the fact that drawing the line between the two said categories is not always an easy task. In the regressions, I include the two categories monarchies and party-based rule, which means that the reference category is made up of the remaining categories. With reference to the colonial heritage, the bivariate analyses showed that countries with a French or Portuguese colonial heritage were particularly likely to transform into dual executives with a popularly elected president as head of state. It therefore makes sense to make use of a category which accounts for Portuguese or French legacy. In addition, a British colonial heritage was strongly associated with hereditary heads of state and also, to a lesser extent, to a non-popularly elected president.

4

HOW DUAL EXECUTIVES EMERGE

109

I consequently include a category which accounts for a British colonial past. Finally, all countries that have emerged from the Soviet Union or Yugoslavia have dual executives with a president as head of state. In the ex-Yugoslavian countries in particular, the presidents have tended to be popularly elected. Consequently, a third category, which accounts for a Yugoslavian or Soviet legacy is included. The reference category consists of countries without a colonial heritage and countries that have gained their independence from other countries than the ones mentioned above. Results are presented in Tables 4.10–4.12. For each dual executive form, four separate regression analyses are presented. I add, in separate regressions, the regime categories and the colonial heritage categories to the variable measuring the effect of geographic diffusion. Finally, in the fourth columns, all variables are entered into the models. The main reason for also running separate regressions with the respective multi-nominal variables is the fact that colonial heritage and regime heritage partly coincide. As we have seen above, the problem is manifest particularly with regard to the categories Soviet/Yugoslav heritage and party-based rule. Finally, since there are no cases where monarchy has given rise to a dual executive with a popularly elected president and no cases where a Soviet/Yugoslav heritage has produced a dual executive with a popularly elected president, these variables are excluded from the respective regression analyses. The most palpable result of the regressions tells us that whereas the regression models are quite successful in predicting why countries choose monarchic regimes and regimes with popularly elected presidents, they do not perform well in predicting the choice of regimes with non-popularly elected presidents. Indeed, as shown in Table 4.10, for countries with non-popularly elected presidents no variable reaches the level of statistical significance in any of the regressions carried out. With respect to countries with popularly elected presidents (Table 4.11) and monarchs (Table 4.12), we find that with reference to the findings in Table 4.9, regional diffusion loses much of its predictive value when included in regressions along with the other independent variables. The variable fails to reach the level of statistical significance in all regressions that include all independent variables. Results from Table 4.11 further reveal that colonial heritage outweighs previous regime form when predicting why regimes with popularly elected presidents emerge. The single most important predictor of a dual executive with a popularly elected president is a French/Portuguese colonial past. A past as a Soviet

−1.426*** (0.213) 0.240 0.038 223

−0.108 (0.357) 0.898 −1.091 (0.776) 0.336 −0.171 0.651 0.843 −1.173*** (0.229) 0.309 0.039 223

0.020 (0.013) 1.021

Reg. 2

In each cell, regression coefficients are listed first, with standard errors in parentheses, followed by odds ratios ***p < 0.001, **p < 0.01, *p < 0.05

Nagelkerke R 2 N

Constant

Soviet Union/Yugoslavia

France/ Portugal

Colonial heritage Britain

Monarchy

0.313 (0.414) 1.368 0.655 (0.444) 1.925

0.018 (0.011) 1.018

Similar regime density (lag 1)

Regime (lag 1) Party-based

Reg.1

Independent variables

0.025 (0.372) 1.026 −1.088 (0.790) 0.337 0.310 (0.585) 1.363 −1.263*** (0.254) 0.283 0.042 223

0.453 (0.424) 1.573 0.673 (0.461) 1.961

Reg. 3

0.002 (0.373) 1.002 −1.040 (0.791) 0.354 −0.120 (0.686) 0.887 −1.344*** (0.265) 0.261 0.052 223

0.397 (0.430) 1.487 0.584 (0.468) 1.793

0.017 (0.013) 1.017

Reg. 4

Table 4.10 Explaining choice of a dual executive with a non-popularly elected president as head of state 1850–2019. Logistic regression analyses

110 C. ANCKAR

−1.985*** (0.235) 0.137 0.120*** 223

−0.312 (0.556) 0.732 3.021*** (0.577) 20.508 2.508** (0.739) 12.275 −2.208*** (0.314) 0.110 0.348*** 223

0.005 (0.018) 1.005

Reg. 2

In each cell, regression coefficients are listed first, with standard errors in parentheses, followed by odds ratios ***p < 0.001, **p < 0.01, *p < 0.05

Nagelkerke R 2 N

Constant

Soviet Union/Yugoslavia

France/ Portugal

Colonial heritage Britain

1.095** (0.398) 2.543

0.033** (0.012) 1.034

Similar regime density (lag 1)

Regime (lag 1) Party-based

Reg. 1

Independent variables

−0.341 (0.558) 0.711 2.955*** (0.580) 19.201 2.389*** (0.601) 10.897 −2.282*** (0.317) 0.102 0.357*** 223

0.641 (0.489) 1.898

Reg. 3

−0.341 (0.558) 0.711 2.935*** (0.582) 18.814 2.216** (0.768) 9.167 −2.310*** (0.328) 0.099 0.357*** 223

0.650 (0.488) 1.915

0.006 (0.018) 1.006

Reg. 4

Table 4.11 Explaining choice of a dual executive with a popularly elected president as head of state 1850–2019. Logistic regression analyses

4 HOW DUAL EXECUTIVES EMERGE

111

−1.783*** (0.243) 0.168 0.210*** 223

1.316*** (0.360) 3.730 −0.840 (1.063) 0.432 −2.116*** (0.281) 0.121 0.204*** 223

0.023** 0.007 1.024

Reg. 2

In each cell, regression coefficients are listed first, with standard errors in parentheses, followed by odds ratios ***p < 0.001, **p < 0.01, *p < 0.05

Nagelkerke R 2 N

Constant

France/ Portugal

Colonial heritage Britain

Monarchy

−1.006 (0.642) 0.366 1.544** (0.454) 4.685

0.023** (0.008) 1.024

Similar regime density (lag 1)

Regime (lag 1) Party-based

Reg. 1

Independent variables

2.435*** (0.490) 11.413 0.004 (1.112) 1.004 −2.742*** (0.427) 0.064 0.354*** 223

−0.927 (0.676) 0.396 2.917*** (0.572) 18.495

Reg. 3

2.305*** (0.496) 10.022 0.123 (1.115) 1.131 −2.885*** (0.439) 0.056 0.371*** 223

−0.833 (0.677) 0.435 2.717*** (0.584) 15.132

0.014 (0.008) 1.014

Reg. 4

Table 4.12 Explaining choice of a dual executive with a hereditary monarch as head of state 1850–2019. Logistic regression analyses

112 C. ANCKAR

4

HOW DUAL EXECUTIVES EMERGE

113

or Yugoslavian republic also increases the likelihood of choosing a dual executive with a popularly elected president. Based on Table 4.12, we reach the conclusion that with respect to choosing a dual executive with a monarch as head of state, a previous monarchic regime form strikes through as the most important explanation, although a British colonial heritage is important as well.

4.7

Discussion

The present chapter set out to test the propositions that the introduction of dual executives could be attributed to the previous regime form, the colonial heritage, and geographic diffusion from neighbouring countries. Initial bivariate analyses gave support to the relevance of all three explanations. In line with the expectations, we noted that republics with dual executives tended to trace their origin to former single-party systems. This finding was particularly relevant for explaining the dual executives that emerged in Eastern Europe after the fall of the Berlin Wall. Furthermore, much in line with the scenario outlined in Huntington’s discussion of the ‘King’s dilemma’, we found that dual executives with a monarch as head of state had developed from monarchic oligarchies, i.e. systems where the monarch had shared power with other actors, most often prime ministers in countries with a restricted right to vote. However, the analyses also showed that the same pattern was less strong with regard to former absolute monarchies; among these systems only a third of the countries transformed into dual executives with a monarch as head of state whereas more than half of the cases transformed into systems with a non-popularly elected president as head of state. In line with findings from studies on semi-presidentialism, we found that a Portuguese and a French colonial heritage tended to give rise to dual executives with a popularly elected president as head of state. The same tendency was observable also for countries with a Soviet or Yugoslavian legacy. Correspondingly, dual executives with a monarch as head of state were often former British colonies. However, the relationship between British colonialism and the monarchic regime form is not straightforward; indeed, a majority of the former British colonies have transformed into republics when gaining independence or when transiting from autocracy to democracy. Finally, geographical diffusion explained the emergence of two of the three dual executive variants, namely popularly

114

C. ANCKAR

elected presidents and monarchs as heads of state. However, the introduction of dual executives with non-popularly elected presidents were not affected by the density of the said regime form in neighbouring countries. Multivariate analyses largely confirmed the findings. The most important discrepancy between the bivariate and multivariate analyses concerned spatial diffusion, which lost its predictive power for all three variants of the dual executive regime form, when included in regressions with the previous regime form and the colonial heritage. The most tangible result is the difference in the relevance of the independent variables with regard to executives with popularly elected presidents and monarchs on the one hand, and non-popularly elected presidents on the other. The multiple regression analyses showed that the introduction of the non-popularly elected president was completely insensitive to variations in the independent variables. This is perhaps natural, as the theoretical arguments were largely derived from studies on semipresidentialism—where the popular election of the president, as we have seen, is widely considered a central component—and from Huntington’s discussion on how powerful monarchs are gradually dispersed of their powers as countries democratize. In contrast, theoretical arguments in favour of why countries should adopt dual executives with presidents not elected by popular vote are hard to find. The counterargument is that it is difficult to come up with reasons for why diffusion mechanisms should not explain the emergence of nonpopularly elected presidents. This is, indeed, a valid argument. In practice, however, the category of non-popularly elected presidents is different from the other two variants in the sense that there is no colonial power which stands out as the torchbearer of the regime form in question. Whereas Britain is the prototype of the dual executive with a hereditary monarch, and France and Portugal the models of a dual executive with a popularly elected president, there is no colonial power which can be expected to transport the non-popularly elected president variant to its former dependencies. True, the leaders of the Soviet Union and Yugoslavia, were not popularly elected, but as we have seen, the colonial heritage in these contexts is of a different nature in the sense that the colonizing entity ceased to exist before it became democratic. In other words, there has never been a metropolitan democratic regime model, which could be exported to the entities of the collapsed state. Instead, the qualitative assessment of the cases conducted in conjunction with the bivariate analyses suggests that countries choose dual

4

HOW DUAL EXECUTIVES EMERGE

115

executives with non-popularly elected presidents for a variety of reasons. As we recall, many of the countries with non-popularly elected presidents were former British entities. A distinctive feature of this group was that it mainly consisted of countries that were large and/or culturally distant from Britain. Evidently, for these countries substituting the monarch with a non-popularly elected president was an important means of distancing itself from the mother country and enhancing the national identity and pride.2 In other countries, the arrangement with a president not elected by popular vote has been a temporary solution at times where the countries have gained independence or transited from autocracy to democracy. With regard to the power dimension, we found little evidence of a link between the former regime type and the powers of the head of state. The only exception occurred in the category of popularly elected presidents, where countries with presidents that had emerged from singleparty systems were distributed fairly evenly along the spectrum of powers whereas presidents in systems that had transformed from military or personalist regimes were concentrated into the two highest power categories. With regard to the colonial heritage, we also reach the conclusion that it matters little for the powers the head of state possesses when the dual executive regime form is adopted. The most obvious finding concerned the former Yugoslav republics, where the presidents consistently were very weak. The strongest powers a president in this context has possessed is 1.42, which was the score for the president of Yugoslavia in the year 2000, when the country (effectively Serbia and Montenegro) surpassed the threshold of democracy. Instead, in terms of presidential powers, the crucial dividing line is the one that separates popularly elected and non-popularly elected presidents, irrespective of their previous regime form or colonial heritage. Initial findings suggested that in countries that introduced popularly elected presidents, the powers of the head of state were generally higher than in countries where non-popularly elected presidents were introduced. A more careful analysis of the cases revealed that the pattern was even stronger than the bivariate statistical analyses suggested. Although, systems with very strong, non-popularly elected presidents have been introduced in a number of countries, these instances most often refer to exceptional and transitional periods in the history of the countries. In the category of dual executives with a monarch as head of states, results showed that there was substantial variation in terms of how much

116

C. ANCKAR

powers the monarchs possessed at the time of the introduction of the monarchic dual executive. This might seem odd, as we would intuitively expect that hereditary heads of states in democracies possess merely ceremonial powers. However, the pattern discovered makes full sense, when we consider the fact that the dual executives with a monarch as head of state almost exclusively have grown out of existing absolute monarchies or monarchic oligarchies. A regime change from these categories to a democratic constitutional or semi-constitutional monarchy does not necessarily imply that there has been a radical change in the form of government. Instead, the shift from autocracy to democracy can occur as a consequence of a rather moderate reduction of the powers of the ruling monarch. However, in line with Huntington’s proposition, we expect that such systems are not stable and as time passes by, the powers of the monarch should gradually wither away. Such tendencies, along with other patterns of transformations across the three forms of dual executives, will be scrutinized in detail in Chapter 5, to which we now turn.

Notes 1. The countries are Antigua and Barbuda, Bahamas, Belize, Dominica, Grenada, St. Kitts and Nevis, St. Lucia, St. Vincent and the Grenadines, and Tuvalu. 2. Indeed, as the manuscript is in its final stage of development, Barbados has substituted the British monarch with a president, and thus become a republic.

References Åberg, J., & Denk, T. (2020). Diffusion and the choice of democratic government system at the time of democratisation. Zeitschrift für Vergleichende Politikwissenschaft, 14(2), 75–98. Acemoglu, D., Johnson, S., & Robinson, J. A. (2001). The colonial origins of comparative development: An empirical investigation. American Economic Review, 91(5), 1369–1401. Anckar, C. (2018). Global patterns of regime change 1800–2015. Political Science, 70(3), 207–233. Anckar, C. (2021). Constitutional monarchies and semi-constitutional monarchies: A global historical study, 1800–2017. Contemporary Politics, 27 (1), 23–40. Anckar, D. (2000). Jäähyväiset Semipresidentialismille. Politiikka, 42(1), 9–14.

4

HOW DUAL EXECUTIVES EMERGE

117

Barkan, J. D. (2013). Emerging legislatures. In N. Cheeseman, A. Scheibler, & D. Anderson (Eds.), Routledge handbook of African politics (pp. 250–264). Routledge. Beck, N., Gleditsch, K. S., & Beardsley, K. (2006). Space is more than geography: Using spatial econometrics in the study of political economy. International Studies Quarterly, 50(1), 27–44. Bernhard, M., Reenock, C., & Nordstrom, T. (2004). The legacy of Western overseas colonialism on democratic survival. International Studies Quarterly, 48(1), 225–250. Blondel, J. (1992). Dual leadership in the contemporary world. In Lijphart, A. (Ed.). Parliamentary versus presidential government. Oxford University Press. Boix, C., Miller, M., & Rosato, S. (2018). Boix-Miller-Rosato dichotomous coding of democracy, 1800–2015. https://doi.org/10.7910/DVN/ FJLMKT. Harvard Dataverse, V3. UNF:6:dt+jCMIItovQfCIW0PIpOg== [fileUNF]. Brinks, D., & Coppedge, M. (2006). Diffusion is no illusion: Neighbor emulation in the third wave of democracy. Comparative Political Studies, 39(4), 463–489. Brownlee, J. (2009). Portents of pluralism: How hybrid regimes affect democratic transitions. American Journal of Political Science, 53(3), 515–532. Cabanis, A., & Martin, M. L. (2005). Un espace d’isomorphisme constitutionnel. In La Constitution et les valeurs. Mélanges en l’honneur de Dmitri Georges Lavroff (pp. 343–358). Paris: Dalloz. Darmofal, D. (2015). Spatial analysis for the social sciences. Cambridge University Press. Doorenspleet, R. (2005). Democratic transition: Exploring the structural sources of the fourth wave. Lynne Rienner. du Bois de Gaudusson, J. (2009). Le mimétisme postcolonial, et après? Pouvoirs, 129(2), 45–55. Elgie, R. (2011). Semi-presidentialism: An increasingly common constitutional choice. In R. Elgie, S. Moestrup, & Y. S. Wu (Eds.), Semi-presidentialism and democracy (pp. 1–20). Palgrave Macmillan. Elgie, R. (2012). Exogenous political institutions? Constitutional choice in postindependence francophone sub-Saharan Africa. Political Research Quarterly, 65(4), 771–783. Elgie, R., Moestrup, S., & Yu, S. W. (2011). Semi-presidentialism: What have we learned? In R. Elgie, S. Moestruop, & S. W. Yu (Eds.), Semi-presidentialism and democracy (pp. 264–274). Palgrave Macmillan. Frison-Roche, F. (2007). Semi-presidentialism in a post-communist context. In R. Elgie & S. Moestrup (Eds.), Semi-presidentialism outside Europe: A comparative study. Routledge.

118

C. ANCKAR

Gandhi, J. (2008). Political institutions under dictatorship. Cambridge University Press. Ganga, B. T., & Sharma, J. (2009). From insurgency to democracy: The challenges of peace and democracy-building in Nepal. International Political Science Review, 30(2), 205–219. Geddes, B. (1999). What do we know about democratization after twenty years? Annual Review of Political Science, 2, 115–144. Geddes, B., Wright, J., & Frantz, E. (2014). Autocratic regimes and transitions. Perspectives on Politics, 12(2), 313–331. Gleditsch, K. S., & Ward, M. D. (2006). Diffusion and the international context of democratization. International Organization, 60(4), 911–933. Goldring, E., & Greitens, S. C. (2020). Rethinking democratic diffusion: Bringing regime type back in. Comparative Political Studies, 53(2), 319–353. Greene, J. P. (2007). Traditions of consensual governance in the construction of state authority in the early modern empires in America. In M. Jansson (Ed.), Realities of representation: State building in early modern Europe and European America (pp. 171–186). Palgrave Macmillan. Hadenius, A., & Teorell, J. (2007). Pathways from authoritarianism. Journal of Democracy, 18(1), 143–156. Hariri, J. G. (2012). The autocratic legacy of early statehood. American Political Science Review, 106(3), 471–494. Houle, C., Kayser, M. A., & Xiang, J. (2016). Diffusion or confusion? Clustered shocks and the conditional diffusion of democracy. International Organization, 70(4), 687–726. Houle, C., & Kayser, M. A. (2019). The two-step model of clustered democratization. Journal of Conflict Resolution, 63(10), 2421–2423. Huntington, S. P. (1968). Political order in changing societies. Yale University Press. Jung, J. K., & Deering, C. J. (2015). Constitutional choices: Uncertainty and institutional design in democratizing nations. International Political Science Review, 36(1), 60–77. Keller, E. J. (1995). Decolonization, independence and the failure of politics. In P. Martin & P. O’Meara (Eds.), Africa (3rd ed., pp. 156–171). Indiana University Press. Kopstein, J., & Reilly, D. A. (2000). Geographic diffusion and the transformation of the postcommunist world. World Politics, 53(1), 1–37. Lane, J.-E. (1996). Constitutions and political theory. Manchester University Press. Levitsky, S., & Way, L. (2006). Linkage versus leverage: Rethinking the international dimension of regime change. Comparative Politics, 38(4), 379–400. Levitsky, S., & Way, L. (2010). Competitive authoritarianism: Hybrid regimes after the Cold War. Cambridge University Press.

4

HOW DUAL EXECUTIVES EMERGE

119

Li, R. P. Y., & Thompson, W. R. (1975). The ‘coup contagion’ hypothesis. Journal of Conflict Resolution, 19(1), 63–88. Linz, J. (1990). The perils of presidentialism. Journal of Democracy, 1(1), 51–69. Magaloni, B. (2008). Credible power-sharing and the longevity of authoritarian rule. Comparative Political Studies, 41(4/5), 715–741. McCargo, D. (2005). Network monarchy and legitimacy crises in Thailand. The Pacific Review, 18(4), 499–519. Merriam-Webster Online Dictionary. https://www.merriam-webster.com/dictio nary/colonialism Nepal, M., Bohara, A. K., & Gawande, K. (2011). More inequality, more killings: The Maoist insurgency in Nepal. American Journal of Political Science, 55(4), 885–905. Neto, O. A., & Lobo, M. C. (2014). Semi-presidentialism in lusophone countries: Diffusion and operation. Democratization, 21(3), 434–457. O’Loughlin, J., et al. (1998). The diffusion of democracy. Annals of the Association of American Geographers, 88(4), 545–574. Pastor y Camarasa, A. (2018, December 11). The challenges of semipresidentialism in Tunisia. Constitution Making. https://www.constitutionalchange.com/the-challenges-of-semi-presidentialism-in-tunisia/ Persson, T., & Tabellini, G. (2003). The economic effects of constitutions. MIT Press. Przeworski, A., Alvarez, M., Cheibub, J. A., & Limongi, F. (2000). Democracy and development: Political institutions and well being in the world, 1950–1990. Cambridge University Press. Riggs, F. (1997). Presidentialism versus parliamentarism: Implications for representativeness and legitimacy. International Political Science Review, 18(3), 253–278. Simmons, B., Garrett, G., & Dobbin, F. (2007). The global diffusion of public policies: Social construction, coercion, competition, or learning? Annual Review of Sociology, 33(1), 449–472. Smith, T. (1978). A comparative study of French and British decolonization. Comparative Studies in Society and History, 20(1), 70–102. Smith, B. B. (2005). Life of the party: The origins of regime breakdown and persistence under single-party rule. World Politics, 57 (3), 421–451. Starr, H., & Lindborg, C. (2003). Democratic dominoes revisited: The hazards of government transition, 1975–1996. Journal of Conflict Resolution, 47 (4), 490–519. Teorell, J. (2010). Determinants of democratization: Explaining regime change in the world, 1972–2006. Cambridge University Press. Thapa, G. B., & Sharma, J. (2009). From insurgency to democracy: The challenges of peace and democracy-building in Nepal. International Political Science Review, 30(2), 205–219.

120

C. ANCKAR

Tobler, W. (1970). A computer movie simulation of urban growth in the Detroit Region. Economic Geography, 46(2), 234–240. Wahman, M., Teorell, J., & Hadenius, A. (2013). Authoritarian regime types revisited: Updated data in comparative perspective. Contemporary Politics, 19(1), 19–34. Wejnert, B. (2005). Diffusion, development, and democracy, 1800–1999. American Sociological Review, 70(1), 53–81. Wejnert, B. (2014). Diffusion of democracy: The past and future of global democracy. Cambridge University Press. Wright, J. (2008). Political competition and democratic stability in new democracies. British Journal of Political Science, 38(2), 221–245. Wu, Y. S. (2011). Clustering of semi-presidentialism: A first cut. In R. Elgie, S. Moestrup, & Y. S. Wu (Eds.), Semi-presidentialism and democracy (pp. 21– 41). Palgrave Macmillan.

CHAPTER 5

Patterns of Transformation in Dual Executives

5.1

Theoretical Patterns of Transformation

Having established the patterns that explain the birth of dual executives, I shall now shift focus to the third aim of the study, which is to explain why certain forms of dual executives are more stable than others. The keyword here spells legitimacy; strong heads of state are expected to survive only in systems that are democratically legitimate. The main assumption is that the popular election of the head of state constitutes the crucial element, which legitimizes strong presidents in democratic settings. The argument is attributed in particular to Matthew Shugart (Samuels & Shugart, 2010, 25–34; Shugart, 2005; Shugart & Carey, 1992) who used a principle-agent framework for classifying democratic regimes. As discussed in Chapter 2, according to Samuels and Shugart, the relationship between the executive and the legislature is crucial for this classification. They make a distinction between the origin and the survival of the executive. In parliamentary systems, the cabinet emerges from the legislature and can be brought down, at any time, by the same organ. This means that ‘the prime minister and the cabinet are agents of the parliamentary majority’ (Samuels & Shugart, 2010, 26). A presidential system, again, is characterized by the fact that the president is popularly elected. Therefore, he or she derives his or her legitimacy

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 C. Anckar, Presidents, Monarchs, and Prime Ministers, Palgrave Studies in Presidential Politics, https://doi.org/10.1007/978-3-031-03960-7_5

121

122

C. ANCKAR

from the people, not the parliamentary majority. In addition, the government (including the president) is not dependent on the parliament for its survival. The constellation is different in dual executives where the president is popularly elected at the same time as the prime minister emerges from, and is accountable to, the legislature. To complicate matters further, the president can have substantial powers when choosing or appointing the prime minister and the president can also have the power to dismiss the prime minister and or the government. In such situations, the two branches of the executive originate from separate institutions (the president from the electorate, and the prime minister from the assembly and, in some cases, the president) and are dependent on their survival on different institutions (the president on the electorate and the prime minister on the legislature and, in some cases, the president) (Samuels & Shugart, 2010, 29–30). As we have seen in Chapter 2, Samuels’ and Shugart’s (2010) principleagent framework has been highly influential when authors have discussed how hybrid forms, essentially semi-presidentialism, should be defined. For the purpose of the present chapter, however, we are not interested in drawing a line between semi-presidentialism and parliamentarism. Instead, we expect the popular election of the president to enhance the legitimacy of the head of state as a political actor who holds real powers and actively participates in the policy process. Since the president does not depend on the parliament for his or her survival in office, the president can challenge the prime minister and other ministers without having to fear the threat of being removed from office. Particularly in situations where the president feels there is strong popular support for his or her actions, he or she should have strong possibilities to push his or her own agenda. Köker (2017, 42–43) points out that popularly elected presidents not only have a possibility to exercise powers, they also have a duty towards the electorate to play an active role in politics. Voters expect presidents to deliver on their promises made during the campaign; a president who does not even attempt to do so risks losing the re-election or reducing the bargaining weight in relation to the legislature (Köker, 2017, 42). Overall, then, the strong legitimacy of the popularly elected president is expected to strengthen the position of the president in relation to the prime minister (Baylis, 1996; Duverger, 1980; Neto & Lobo, 2009; Shugart & Carey, 1992). The situation is different for a non-popularly elected president. Most often, these presidents are elected by parliament, but other variants exist

5

PATTERNS OF TRANSFORMATION IN DUAL EXECUTIVES

123

as well. In situations where the presidents are elected by parliament, their position is similar to prime ministers in the sense that they base their origin on the legislature. Are they thereby also agents of the legislature? Köker (2017, 246) answers in the affirmative, noting that non-popularly elected presidents ‘acknowledge their dependence on the legislature and therefore use their powers less frequently as not to interfere in the work of their principals’. The view that the popular election of the president in particular strengthens the positions of the president does not stand unchallenged. The opposing view has been championed notably by Tavits (2009a, 33), who has noted that ‘the argument remains largely a recurring assumption rather than an elaborated theory’. Moreover, she points out that while non-popularly elected presidents indeed base their legitimacy on the legislature, we cannot disregard the fact that these non-popularly elected presidents are not accountable to the parliamentary majority (Tavits, 2009a, 34). In other words, the positions of a non-popularly elected president and a prime minister are quite different. Furthermore, parliamentary majorities are likely to shift, and ‘the assembly that puts an indirectly elected president into power is not the same assembly that decides on his or her reappointment’ (Tavits, 2009a, 34–35). Having conducted an extensive empirical study involving qualitative and quantitative techniques, she notes that ‘[directly elected presidents] are not more likely to exercise influence over government formation process than [indirectly elected presidents]’ and that ‘[t]here was no evidence that the method of election is responsible for inflating presidential activism in terms of initiating or blocking legislation, making appointments, engaging in foreign policy and using symbolic politics’ (Tavits, 2009a, 233). Köker (2017, 246), on the other hand, in a study of presidential vetoes in Central and Eastern Europe received results that pointed in the other direction, reaching the conclusion that the popular election of the president constituted the most important explanation of presidential activism. The discussion conducted above has strong implications for the stability of different forms of dual executives. Although the empirical evidence cited above does not provide an unequivocal answer to the question if popularly elected presidents are more powerful than non-popularly elected ones, it is evident that popularly elected presidents have a stronger legitimacy reserve than presidents not elected by popular vote. However, this legitimacy is important particularly in systems where the president

124

C. ANCKAR

has strong constitutional powers. In such systems, the powerful president is perceived of as a legitimate actor, both by the electorate and by the political establishment, notably the parliamentarians and the members of government. Things should be completely different in situations where a president has substantial powers at the same time as he or she is not popularly elected. Such presidents cannot rely on popular support and are likely to be challenged by other heavyweights in parliament and government, notably the prime minister. Such systems should therefore be hard to find, and in cases where they exist, they should not be long-lived. What about hereditary monarchs, then? For these systems, the lack of democratic legitimacy is even more profound than for systems with a nonpopularly elected president. The monarch has not been popularly elected, nor has he or she been elected by any other representative organ of the people. Instead, the legitimacy is based on tradition and the ‘monarch legitimizes him or herself by a strong divine or natural source outside the political regime’ (Kailitz, 2013, 48–49). Recall the discussion on the King’s dilemma conducted in Chapter 4. A powerful monarch who was under pressure to democratize the country had three options, transformation, coexistence, and maintenance. Unless the monarch aimed to resist the pressures for democratization (i.e. opt for the strategy maintenance or transform powers to another institution than a democratic one), he or she either had to completely transform powers to the people or to find a solution where he or she would share powers with the democratic institutions and thus ‘coexist’ with them. Faced with these two options, monarchs would, presumably in most cases, choose the latter one. After all, it is better to have some powers than no powers. However, Huntington was very pessimistic regarding the long-term prospects for power sharing between monarchic rule and party government; ‘[t]he pressures in such a system are either for the monarch to become only a symbol or for him to attempt to limit the expansion of the political system…’ (Huntington, 1968, 180–181). It is therefore reasonable to make the assumption that the time period during which a monarch is in possession of substantial powers is limited, and that pressures for reducing the powers of the monarch will grow as democracy consolidates. Furthermore, the more powers the monarch possesses, the stronger the pressures for reducing his or her powers. Table 5.1 summarizes the discussion conducted above.1 It combines the mode of selection of head of state with the power dimension. The arrows indicate expected transitions patterns within and between the three

5

PATTERNS OF TRANSFORMATION IN DUAL EXECUTIVES

125

Table 5.1 Expected patterns of transformations in democratic countries with dual executives Selection of head of state Popular election

Non-popular

Inheritance

election 7.0 Powers of head . of state

. . . . . 0.0

categories of dual executives. In countries where the president is powerful, at the same time as he or she is not popularly elected (uppermost part in the second column from the right), the issue of the legitimacy of the head of state should be central. Such systems should experience pressures for reforms that would either introduce the popular election of the president (and consequently move the countries leftward) or, alternatively, strip the head of state of his or her powers (thereby moving the country downwards in the column). In principle, a similar argument could be raised with regard to the countries situated in the lower left corner. If a powerless president is popularly elected, one would expect there to be pressures towards strengthening the position of the president (a move upwards towards the upper left corner), or, alternatively to allow the legislature to elect the president (a move towards right). However, we can expect these tendencies to be significantly less pronounced than in systems with strong, non-popularly elected presidencies, since the issue of legitimacy is less palpable in systems with weak but popularly elected presidents. Opponents of popular elections in countries with weak presidents tend to base their arguments not on the issue of legitimacy but rather on cost efficiency; since the president does not have significant powers, it is

126

C. ANCKAR

unnecessary to conduct costly and time consuming (for parties, candidates, and voters) elections. Furthermore, the holding of elections also for the position of largely ceremonial presidencies can be motivated by an ambition to strengthen democracy by enhancing the level of participation and also by the fact that the popular mode of election tends to enjoy widespread support among the electorates (Lindberg, 2006; Tavits, 2009b, 43–44). For monarchies, the expected patterns of transformations are very similar to the ones for non-popularly elected presidents. If a monarch is vested with substantial powers when a country transforms from absolute monarchy to democracy, we expect the time period during which a monarch is in possession of powers to be limited; pressures for reducing the powers of the monarch will grow as democracy consolidates. It is consequently reasonable to expect that the general trend will go from the upper right corner down towards the lower right corner. Another possibility is that monarchies with strong monarchs transform into republics with strong presidents. Such a pattern would transform systems in the upper right corner to move leftwards and directly or eventually end up in the upper left corner. In addition to the patterns laid out above, other theoretical transformation patterns exist. For instance, it is possible that a strong monarch is directly replaced by a weak, non-popularly elected president. However, pressures for transformations along these lines are less obvious than the ones outlined in the table.

5.2

Empirical Patterns of Transformations

I shall continue by an empirical assessment of the extent to which the modes of transitions discussed above are matched by the empirical reality. As we recall, both the mode of selection and the powers of the head of state were expected to affect transformation patterns. Let me start by providing an overview of the extent to which the powers of the head of state increase or decrease after a change in the selection procedure occurs. Since it is possible that a change in the mode of selection of the head of state is not immediately reflected in the powers of the head of state, the power dimension is measured during a four years interval, beginning in the year when the selection procedure changed. Results are presented in Table 5.2. A first observation tells us that selection procedures tend to be extremely stable. Altogether, there are only 27 cases where the mode of selection of the head of state has changed in the

5

PATTERNS OF TRANSFORMATION IN DUAL EXECUTIVES

127

Table 5.2 Mode of selection and changes in powers of heads of state 1850– 2019 Changes in powers of head of state during t + 3 Change in mode of selection Less of head of state at t powers No change 146 (3.7) Popular to non-popular election Non-popular election to inheritance Non-popular to popular election Inheritance to non-popular election Sum

More powers 104 (2.6)

3,942 (100.0)

1 (33.3)

No changes 3,692 (93.7) 1 (33.3)

Sum

1 (33.3)

3 (100.0)

0 (0.0)

0 (0.0)

1 (33.3)

1 (100.0)

1 (5.9)

9 (52.9)

7 (41.2)

17 (100.0)

3 (50.0)

3 (50.0)

0 (0.0)

6 (100.0)

151 (3.8)

3,705 (93.3)

113 (2.8)

3,969 (100.0)

Each cell returns the total number of cases with row percentages in parentheses

empirical material. Furthermore, nearly two thirds of these cases are found in the category where a non-popular elected president is substituted by a popularly elected. The inverse pattern occurs much more rarely, only three cases, Greece 1927, Turkey 1989, and Moldova 2001, are found in the material. In Turkey, the powers of the presidents remained the same, namely 0.58. In Greece, the powers of the presidents were reduced from 3.00 to 2.17 whereas in Moldova, the powers increased from 0.83 to 2.83. The second most frequent mode of change is when a hereditary monarch is substituted by a president elected by parliament. The cases in question are Ireland 1937, Sri Lanka 1972, Malta 1974, Nigeria 1963, Trinidad and Tobago 1976, and Mauritius 1992. Interestingly enough, in half of these cases, the powers of the head of state were reduced in connection with the reform. In Ireland, Nigeria, and Sri Lanka, the powers of the president equalled those of the previous monarch. Moving on, we note that there are no instances of direct transformations from a monarchy to a popularly elected presidency and, likewise, no examples of the reversed pattern. However, on one occasion, in Greece in 1935, a non-popularly elected president has been replaced by a hereditary monarch.

128

C. ANCKAR

Although the empirical evidence is far from overwhelming, it does provide some support for the transformation patterns laid out in Table 5.1. When countries move from a non-popularly to a popularly elected president, the powers of the head of state either remain the same (in 53% of the cases) or increase (in 41% of the cases). Only in one case, Germany in 1925 did the powers of the president decrease in the same time period. Another interesting finding shows that in the six instances, where a monarch is replaced by a non-popularly elected president, the powers of the head of state have not increased in any instances. In comparison, the powers of the head of state decreased in three countries in connection with the reform. We then turn to the next task of the research effort, which is to assess how stable the powers of the heads of state are in the three categories based on the mode of the selection of the head of state. In these analyses, I make use of the five presidential powers categories presented in Chapter 3. The ambition is to reveal the patterns by which countries move between the categories in question. The tests are conducted by means of Markov chains probability analyses. The basic assumption underlying these analyses is that the probability of a country’s movement between categories from one year to another (i.e. from t − 1 to t) is independent of all other circumstances than the category in which the country in question was placed at time t − 1. In other words, the time interval (the longevity of the country’s placement in the category) that precedes t − 1 is disregarded. Let us start by taking a look at the category of non-popularly elected presidents. Results are presented in Table 5.3, and overall, the patterns are in line with our expectations. Countries with weak, non-popularly elected presidents are more likely to remain in the same power category than countries with strong, non-popularly elected presidents. There are very few cases in the highest power category and presidents in these systems are exceptionally likely to lose their powers. In contrast, the evidence does not support the presumption that countries with powerful, nonpopularly elected presidents would be particularly prone to introduce a popular election of the president. Quite to the contrary, more than half of the countries that have introduced the popular election of the president have been placed in the two weakest power categories at the time of the reform, whereas only three countries, Finland 1925, France 1965, and Namibia 1994, have been found in the two categories with the most powerful presidents.

5

PATTERNS OF TRANSFORMATION IN DUAL EXECUTIVES

129

Table 5.3 Transition probabilities of powers of the head of state in democratic countries with presidents not elected by popular vote

Powers of HOS at t − 1 0.00 >0.00–0.99 1.00–2.99 3.00–4.99 5.00–7.00 Frequency N Powers of HOS at t − 1 0.00 >0.00–0.99 1.00–2.99 3.00–4.99 5.00–7.00 Frequency N Powers of HOS at t − 1 0.00 >0.00–0.99 1.00–2.99 3.00–4.99 5.00–7.00 Frequency N

Mode of selection of HOS at t

Powers of head of state at t

Non-popular election

0.00 >0.00– 0.99 96.0 4.0 1.6 97.1 0.0 3.8 0.0 0.0 0.0 0.0 15.4 52.1 153 516 0 1

1.00– 2.99 0.0 1.4 95.2 16.1 16.7 29.1 288 2

3.00– 4.99 0.0 0.0 1.0 83.9 33.3 3.1 31 3

5.00– 7.00 0.0 0.0 0.0 0.0 50.0 0.3 3 4

N

0.0 0.0 0.0 0.0 0.0 0.0 0 0

100.0 77.8 0.0 0.0 0.0 52.9 9 1

0.0 22.2 100.0 0.0 0.0 29.4 5 2

0.0 0.0 0.0 100.0 0.0 11.8 2 3

0.0 0.0 0.0 0.0 100.0 5.9 1 4

2 9 3 2 1 100.0 17 N

0.0 0.0 0.0 0.0 0.0 0.0 0

0.0 0.0 0.0 0.0 0.0 0.0 0

0.0 0.0 100.0 0.0 0.0 100.0 1

0.0 0.0 0.0 0.0 0.0 0.0 0

0.0 0.0 0.0 0.0 0.0 0.0 0

0 0 1 0 0 100.0 1

Popular election

Inheritance

151 514 289 31 6 100.0 991 N

Each cell presents, for each power category, the row percentage of the number of observations

Furthermore, the presentation confirms the earlier finding showing that a transition from a non-popularly elected president to a popular elected one does not enhance the powers of the president. In fourteen of the cases, the countries remained in the same power category one year after the reform. The powers of the presidents were significantly increased only in Ireland 1938 and Moldova 2016, which both moved from category 0.00 to category >0.00–0.99, as well as in Poland 1990 and Turkey 2014, both of which moved from category >0.00–0.99 to 1.00–2.99. We also note that there is no example of a country, where the introduction of the popular election coincided with a significant reduction of the powers

130

C. ANCKAR

of the president. In Greece, which remains the only country in the dataset that has switched from a republic to a monarchy, the powers of the head of state remained the same after the transition in 1935. We shall then turn to the category of countries with popularly elected presidents. Results are presented in Table 5.4. Overall, the level of stability does not vary much between the power categories. The only exception is the category where the president is completely powerless. In systems with popularly elected powerless presidents, the president has received increasing powers in nearly one-fifth of the cases. However, since this result is based on an extremely limited number of cases (only three out of seventeen cases, namely Slovenia 1992, Finland 2018, and Moldova 2019, have moved towards more powers from the lowest power category), we should not draw too far-reaching conclusions Table 5.4 Transition probabilities of powers of the head of state in democratic countries with popularly elected presidents

Powers of HOS at t − 1 0.00 >0.00–0.99 1.00–2.99 3.00–4.99 5.00–7.00 Frequency N Powers of HOS at t − 1 0.00 >0.00–0.99 1.00–2.99 3.00–4.99 5.00–7.00 Frequency N Powers of HOS at t − 1

Mode of selection of HOS at t

Powers of head of state t

Popular election

0.00 >0.00– 0.99 82.4 11.8 0.8 98.0 0.0 3.9 0.0 0.0 0.0 0.0 1.6 44.4 18 489 0 1

1.00– 2.99 5.9 1.2 92.2 3.6 0.0 18.4 203 2

3.00– 4.99 0.0 0.0 2.9 94.0 2.7 15.4 170 3

5.00– 7.00 0.0 0.0 1.0 2.4 97.3 20.1 222 4

17 489 206 168 222 100.0 1,102 N

0.0 0.0 0.0 0.0 0.0 0.0 0 No

0.0 50.0 0.0 100.0 0.0 66.7 2

0.0 0.0 0.0 0.0 0.0 0.0 0

0.0 0.0 0.0 0.0 0.0 0.0 0

0 2 0 1 0 100.0 3

Non-popular election

Inheritance

0.0 50.0 0.0 0.0 0.0 33.3 1 cases

N

Each cell presents, for each power category, the row percentage of the number of observations

5

PATTERNS OF TRANSFORMATION IN DUAL EXECUTIVES

131

based on these figures alone. Anyhow, in the three intermediate categories, there is no similar tendency in either direction; in general, the powers of the head of state do not change and in those rare cases where changes occur, they tend to occur in both directions in roughly equal proportions. We also note that radical shifts in the powers of the president are extremely rare. The largest change in the powers of the president in one year happened in Moldova between 2018 and 2019, when the president’s powers rose was from 0.0 to 1.75. In 2019, the V-dem experts considered the Moldovan president to have received the power to remove the prime minister from office. In addition, he was considered to have increased his veto powers and powers to introduce legislation. Finally, as we have already noted, three countries moved from a popularly elected to a nonpopularly elected president. In two of the countries, Greece 1927 and Moldova 2001, the change of powers was significant enough to shift the placement of the countries from one category to another. We finally turn to Table 5.5, which presents transformations patterns in countries with a monarch as head of state. In contrast, to the other categories, the powers are very stable both in systems where the monarchs are weak and in systems where they are strong. This finding clearly contradicts the expectation that powerful monarchs should not survive in democratic settings. It is nevertheless noteworthy that in cases where monarchies have moved from one power category to another, the direction has tended to be towards weakening powers. Only in six instances have the powers of the monarch increased significantly from one year to another, namely in Barbados 2013, Bhutan 2016, Greece 1874, Lesotho 2005, Luxembourg 1930, and Nepal 2000. Furthermore, although there are 98 cases where the monarch is situated in the highest power category, they all refer to one single country, namely Liechtenstein. On the other hand, the group of countries situated in power category 3.00–4.99 (at t) is much more heterogeneous, comprising eight different states. Finally, in the six instances, where a monarch has been substituted by a non-popularly elected president, the reform in question has only had a marginal impact on the powers of the head of state. It is noteworthy, though, that in no cases has the president been more powerful than the monarch he or she succeeded.

132

C. ANCKAR

Table 5.5 Transition probabilities of powers of the head of state in democratic countries with a monarch as head of state

Powers of HOS at t − 1 0.00 >0.00–0.99 1.00–2.99 3.00–4.99 5.00–7.00 Frequency N Powers of HOS at t − 1 0.00 >0.00–0.99 1.00–2.99 3.00–4.99 5.00–7.00 Frequency N Powers of HOS at t − 1

Mode of selection of HOS at t

Powers of head of state t

Inheritance

0.00 >0.00– 0.99 100.0 0.0 0.7 99.0 0.6 2.2 0.0 1.6 0.0 0.0 30.0 44.3 555 819 0 1

1.00– 2.99 0.0 0.2 96.0 6.3 0.0 17.0 314 2

3.00– 4.99 0.0 0.0 1.2 92.2 0.0 3.4 63 3

5.00– 7.00 0.0 0.0 0.0 0.0 100.0 5.3 98 4

N

100.0 0.0 0.0 100.0 0.0 25.0 0.0 0.0 0.0 0.0 16.7 33.3 1 2 No cases

0.0 0.0 75.0 0.0 0.0 50.0 3

0.0 0.0 0.0 0.0 0.0 0.0 0

0.0 0.0 0.0 0.0 0.0 0.0 0

1 1 4 0 0 100.0 6

Non-popular election

Popular election

547 819 321 64 98 100.0 1,849 N

Each cell presents, for each power category, the row percentage of the number of observations

5.3

Multivariate Patterns

5.3.1

Introducing Control Variables

So far, the discussion has centred around the relationship between the mode of selection and the powers of the head of state. It is now time to introduce a number of other potential explanations for why the powers of the head of state might vary. Within the framework of the present study, the role of control variables is particularly crucial, since the power dimension is measured with reference to behavioural traits rather than constitutional provisions. This means that the ability of the head of state, and particularly a president, to exercise power supposedly is closely tied to his or her ability to co-operate with persons in central positions in government and parliament.

5

PATTERNS OF TRANSFORMATION IN DUAL EXECUTIVES

133

This ability to exercise power is affected by a number of circumstances, which can vary extensively from year to year. Such circumstances are not irrelevant when measuring powers based on constitutional provision. However, when de facto powers are in focus they become extremely important and in the following section I shall therefore introduce a number of variables, which can be expected to affect the ability of the head of state to play an active role in politics. Some of these are found in what Tavits (2009a, 35–40) call the ‘political opportunity framework’. According to Tavits (2009a, 35), ‘this… framework—the strength of other political institutions and the constellation of political forces in government and parliament—influences the extent to which presidents choose to interfere in executive and legislative decision making’. In the subsequent analyses, I pay regard to two potential features of such constellations, namely cohabitation and party system fragmentation in government and parliament. In addition, I discuss another potential determinant of monarchical power, namely population size. 5.3.1.1 Cohabitation In comparative politics, the French term ‘cohabitation’ is commonly used to describe situations in semi-presidential systems where the prime minister and the president represent different political parties. Such situations have been referred to as ‘the Achilles heel of semi-presidentialism’ (Elgie & McMenamin, 2011, 616). It is easy to understand why cohabitation could be detrimental for semi-presidential systems. If the two actors that are supposed to share executive power represent different political camps, their relationship is likely to be marked by rivalry and conflict rather than smooth co-operation. This, in turn, could lead to severe deadlocks and general turmoil, but also to regime breakdown. For the purpose of the present study, it is evident that cohabitation can affect the ability of the head of state to exercise powers. There is, to my knowledge, little empirical research with an explicit aim to test this relationship. However, in recent years, a number of research efforts have dealt with the association between cohabitation and presidential activism. The prevailing view, expressed notably by Tavits, suggests that cohabitation is likely to increase the level of presidential activism. According to Tavits (2009a, 35), the goal of presidents is to fulfil their policy preferences, a goal that is accomplished most effortlessly ‘in a noncompetitive environment with like-minded actors in policymaking positions’. This would suggest that in periods when the president and prime minister represent

134

C. ANCKAR

the same party, there is little need for the president to intervene in the policymaking process. During periods of cohabitation, correspondingly, presidents have a stronger incentive to become more actively involved in the processes, as it is quite possible that his or her policy goals are not shared by the prime minister or the parliamentary majority (Tavits, 2009a, 35–37). This line of reasoning has received empirical support not only by Tavits (2009a, 48–50) herself, but also by Protsyk (2005a, 2006) as well as Köker (2017, 61–69). In line with the theoretical arguments and the empirical evidence laid out above it could be argued that it makes sense to assume that presidential powers are higher in periods when the president’s party is not represented in government than in periods when the president’s party is part of the cabinet. However, this is not necessarily the case, as presidential activism is not tantamount to presidential powers. In Tavits’ (2009a, 41) work, the concepts interfuse as presidential activism is defined as ‘the “actual” as opposed to constitutional power of the president’. Her dependent variable, the share of non-partisan cabinet appointments, incorporates elements of both presidential activism and presidential power exercise, but is hardly the most appropriate measure for either one of the concepts. To complicate things further, she introduces a number of presidential power indicators among her independent variables, whereby the theoretical distance between the explanandum and the explanans becomes marginal (Köker, 2017, 28). In Tavits’ defence, it must be admitted that separating the two concepts is a very difficult task when operating with behavioural indicators of presidential powers. Nevertheless, the fact that a president is in possession of powers does not necessarily mean that he or she actually uses these powers. A president can, for instance, have extensive veto powers, but in situations where the party of the president is in control of the legislature, there is little need for him or her to make use of it. Accordingly, during his first term in office, President George W. Bush never used his veto powers. In comparison, Gerald Ford, during his short term in office, vetoed no less than 48 bills. Based on this indicator alone, Ford was perhaps a more ‘active’ president than Bush. However, it would be clearly incorrect to draw the conclusion that Ford’s legislative powers were more extensive than those of Bush’. Luckily, the V-dem dataset is extremely useful for distinguishing between presidential activism and presidential powers, as its indicators do not reflect the extent to which heads of state actually have used a particular prerogative a given year. Instead,

5

PATTERNS OF TRANSFORMATION IN DUAL EXECUTIVES

135

the V-dem experts have been asked to estimate to what extent the head of state would be successful if he or she attempted to make use of the prerogative in question that year. When cohabitation is linked to presidential powers rather than presidential activism, it makes more sense to assume that the association is negative than positive. If a president shares powers with a prime minister from the opposing camp, the president’s ability to forward his or her political agenda is circumscribed as he or she has to compete with a strong political actor backed by the parliament. Especially, with respect to the appointment and dismissal of the prime minister and the government, the hands of the president are tied. If the president attempted to get rid of a prime minister he or she disliked, the parliament could always propose the same, or, from the point of view of the president, an even worse, candidate for the post. In situations where the president and prime minister represent the same party, the president should have a dominant role. This assumption appears quite intuitively, and is, perhaps too often, taken more or less for granted. The basic logic underlying the proposition is that the authority of the president, the head of state, should outweigh that of the prime minister. Apparently, the line of reasoning is heavily influenced by the French experience under the early years of the fifth republic, in which the prime minister’s subordination to the president was ‘based first on law, second on the legitimacy conferred by the election mode, and third on the majority factor’ (Vallée, referred to in Poulard, 1990, 248). Nevertheless, as the two central actors represent the same party, there should be little disagreement about the policy goals of the executive. Moreover, as the party of the president is in control of parliament the president’s role in the government formation and termination processes should be decisive. Any candidate the president proposes for the position as prime minister should receive the legislature’s blessing. This all makes sense, under the important qualification that the authority of the president is stronger than that of the prime minister within the political party they both represent. The probability that this is the case should be high particularly in systems where the president is popularly elected and where the president, constitutionally and/or traditionally, commands strong powers. The popular election is likely to enhance the position of the president, as it provides his/her leadership with legitimacy extending far beyond the party. Prime ministers presumably have a stronger position in systems where the president is elected by parliament. In these cases, both actors base

136

C. ANCKAR

their legitimacy on the legislature, and it is less evident which of them will have a stronger position in the internal party hierarchy. Wide powers, again, imply that the presidency is considered the ‘biggest prize’ in the political competition and therefore the ultimate goal of the powerseeking politician. In such systems, the authority of the president should clearly outweigh that of the prime minister and the role of the prime minister would merely be to assist the president. If the powers of the president are weak, perhaps even merely ceremonial, politicians compete for the position as prime minister, not as president. In these systems, the prime minister should emerge as the more powerful part in the executive branch, regardless of the party affiliation of the president. The aforesaid leads to a general assumption that cohabitation should reduce the powers of the president. However, the tendency should be less pronounced in systems with non-popularly elected presidents and decrease with decreasing powers of the head of state, regardless of the selection mode of the president. In the literature, cohabitation is a term associated with semipresidential forms of governments in which the president is popularly elected. This is natural because in such systems periods of cohabitation also entail a plausible conflict of democratic legitimacy; the two central actors derive their legitimacy from two separate sources, the president from the people and the prime minister from the legislature. In case where the president is elected by parliament, both actors have the same source of legitimacy. As argued above, such systems should rarely have powerful presidents and in cases where non-popularly elected powerful presidents exist, they should not be durable. The same line of argument is of relevance also when discussing effects of cohabitation. Under normal circumstances, during periods of cohabitation the prime minister is backed (or at least tolerated) by a parliamentary majority, whereas the president faces a hostile parliament. Such a constellation should make it more difficult for the president to exercise powers. In other words, both in countries where the president is popularly elected and in countries where the president is non-popularly elected, we expect the powers of the president to be more limited under periods of cohabitation than under periods where the party of the president is represented in government. As monarchs are rarely, if ever, affiliated with political parties, cohabitation is less relevant in hereditary systems. Admittedly, monarchs can have strong political preferences; intuitively, we would perhaps assume

5

PATTERNS OF TRANSFORMATION IN DUAL EXECUTIVES

137

that they tend to be more sympathetic towards conservative than liberal or left-wing prime ministers. However, since these preferences cannot be explicitly (or, indeed, even implicitly) expressed, it makes little sense to attempt to assess the ideological distance between the monarch and the prime minister. As party systems are often highly fragmented it is far from self-evident how cohabitation should be defined and operationalized. The fact that presidents and prime ministers represent different parties does not necessarily mean that their ideological differences are substantial. Indeed, it is quite possible that they represent parties that have a long history of governing together or even run in elections under the same umbrella organization. To partially remedy this problem, I have made use of Elgie and McMenamin’s (2011, 619) definition, where cohabitation is defined ‘as the situation where a president from one party holds power at the same time as a prime minister from an opposing party and where the president’s party is not represented in the cabinet’.2 They also add another clarification: ‘Whatever the party composition of the cabinet, if either the president or the prime minister is non-partisan or independent, then we do not class this as a case of cohabitation’ (Elgie & McMenamin, 2011, 619). 5.3.1.2

Parliamentary Support and Level of Fractionalization in Government and Parliament Another feature of the Tavits’ (2009a) opportunity framework takes as its point of departure the strength of the other central policymaking institutions than the head of state, i.e. the government and the parliament. One feature which is likely to affect the ability of the head of state to exercise powers is the level of support the government enjoys in the legislature. In cases where governments can rely on the support of a comfortable majority of parliamentarians, the possibilities of the head of state to intervene in the policymaking process should be limited. Instead, among the two potential leaders of the executive, the prime minister should emerge as the primus motor in the operations of government. The situation is different when the government does not command a majority in parliament. In such cases, the possibilities of the head of state to take action are radically improved, as he or she can try to rally support among oppositional forces in parliament in order to obstruct the work of government (Baylis, 1996; Köker, 2017, 46–47; Tavits, 2009a, 38).

138

C. ANCKAR

The empirical evidence at hand consistently shows that the partisan support of the government in parliament is an important determinant of presidential activism. Accordingly, Tavits (2009a, 49–50), in her population of semi-presidential and parliamentary systems, found that her measure of presidential powers increased in situations where the cabinet did not enjoy support of a parliamentary majority. In a similar vein, Protsyk (2005b) made the observation that in Eastern Europe, presidents were more likely to challenge prime ministers in situations where minority governments were in power than when majority governments were in place. Interestingly enough, this tendency was observed both in situations where the president’s party was in opposition and when it was represented in government. In addition to the extent to which the parties of the cabinet also control the parliament, we must pay attention to how homogeneous the government and the legislature are. A head of state who wants to play an active role in politics can face either a united or fragmented opponent. In situations where there is a one-party government, the position of the head of state is expected to be weak as the whole government is ideologically cohesive and should stand united behind the prime minister. The situation is different where there is a coalition government composed of several parties representing different ideological positions. In such cases, the president can strive to apply the maxim divide et impera, i.e. split the coalition and try to rally support for his or her own policy proposals from some of the parties in government (Tavits, 2009a, 38). In a like manner, party system characteristics in the parliament shape the possibilities for heads of state to play an active role in the executive sphere. A fragmented parliament is also a weak parliament. As Protsyk (2005b, 153) puts it ‘[p]arliamentary fragmentation invites presidential claims on executive leadership by lowering the president’s political costs of attacking a prime minister and cabinet that lack solid support in Parliament’. Empirically, the significance of government and parliamentary fragmentation is equivocal. In her assessment of presidential powers, Tavits (2009a, 50–51) found evidence of a weak but consistent relationship between cabinet fragmentation and her measure of presidential powers. As expected, the results showed that the more fragmented the government, the more active the president. However, she found no association between parliamentary fragmentationand her measure of presidential powers (Tavits, 2009a, 50–51). In a like manner, Köker (2017, 63– 64) found that presidents’ use of veto powers were not affected by

5

PATTERNS OF TRANSFORMATION IN DUAL EXECUTIVES

139

the level of fragmentation in the legislature. On the other hand, Elgie and Schleiter (2011, 54–57) noted that semi-presidential regimes were particularly likely to suffer a breakdown in situations where parliamentary fragmentation was moderately high. Within the framework of the present study, government fractionalization is measured by the number of parties in government whereas the level of parliamentary fragmentation is measured according to the well-known Laakso-Taagepera formula (Laakso & Taagepera, 1979). 5.3.1.3 Population Size Variables related to physical characteristics have been surprisingly neglected in political science. In recent years, this has changed somewhat and today we often run across studies where population size in particular is related to various phenomena. However, it is easy to concur with Gerring and Veenendaal’s (2020, xx) notion that ‘[population size] plays a supporting role, but hardly ever a starring role in the work of anthropologists, economists, political scientists, and sociologists’. Admittedly, this conclusion is valid also within the context of the present contribution. Although size is rarely considered to be associated with variations in presidential prerogatives, there is reason to believe that it is connected to the survival of monarchical powers. In line with Huntington’s assessment of the ‘King’s dilemma’, we expected former absolute monarchies and monarchic oligarchies to transfer into dual executives in which monarchs possessed more than ceremonial powers. The empirical analyses, conducted in Chapter 4, gave strong support to such patterns of regime transformation. However, we still have not empirically assessed how the powers of monarchs develop over time in these dual executives. Recall, that the basic argument laid out in Chapter 4, suggested that, due to their lack of democratic legitimacy, monarchs with more than ceremonial powers were not expected to be long-lived in democratic settings. There are, however, some theoretical propositions for the expectation that pressures for stripping monarchs of powers should be mitigated in small countries. According to Corbett et al. (2017), the prospects for powerful monarchs to survive in small settings are explained by two components, institutional fidelity and personalization. Institutional fidelity refers to the important and decisive roles played by monarchs at crucial moments in the history of small nations. Indeed, many miniature nations managed to survive mainly or solely because of actions taken by

140

C. ANCKAR

the monarchs, and therefore, there is a ‘tendency to consider the monarch an essential component of the identity of the state’ (Corbett et al., 2017, 692). Personalization, in turn, refers to an assumption that relations between the monarch and the people are more intimate in small units than in large ones. This argument bears strong resemblance to Dahls and Tufte’s (1973) discussion of how relationships between the people and power holders are affected by the size of units. In small units, leaders should be informed about the preferences of the citizens ‘by direct observation and communication’ (Dahl & Tufte, 1973, 87). When relations are more intimate, communication between leaders and citizens are expected to be reciprocal. With reference to a qualitative case study of Bhutan, Liechtenstein, and Tonga, Corbett et al. (2017, 701–702) noted that personalization was an important feature of the states. Monarchs, they asserted, ‘are not remote or distant figures’, a fact that ‘undermin[es] the potential for the regime to become a symbol of oppression’. Although the discussion above has centred on monarchic survival in democratic contexts, population size has also been linked to the survival of authoritarian monarchic regimes. In an extensive theoretical and empirical assessment, Jugl (2020, 284) finds a strong link between smallness and monarchic survival, and makes the argument that ‘three fundamental characteristics of small states (a feeling of vulnerability, social proximity and institutional centralization) stabilize the three pillars of autocratic stability (legitimation, repression and co-optation)’. Furthermore, in a recent article, Gerring et al. (2021, 605–606) note that the relation between size and the monarchic regime form is of a different nature in different eras. Whereas they confirm the link between smallness and monarchies in modern times (from World War I and onwards), in earlier centuries, the association was reversed. Finally, it is worth emphasizing that population size is explicitly linked to the persistence and not to the emergence of monarchic rule. We cannot disregard the fact that all democracies with a monarchic head of state to varying extents have to struggle with a problem of democratic legitimacy in the sense that the head of state inherits his or her position. It is therefore difficult to find any theoretical arguments for why a democratizing country, previously not ruled by a monarch, would choose a monarchic regime form if all other alternatives were available (also Anckar, 2021).

5

PATTERNS OF TRANSFORMATION IN DUAL EXECUTIVES

5.3.2

141

Empirical Patterns

I shall continue by conducting a number of statistical analyses where the ambition is twofold. A first ambition is to give a more comprehensive view of how the powers of the head of state are affected by previous levels of powers held by the head of state. A second ambition is to assess to what extent the powers of the head of state are affected by the control variables introduced above. The dependent variable takes into account whether or not presidential powers have increased, decreased or remained the same a given year. Data-availability for number of parties in government , government support in parliament , and party system fragmentation is limited. In addition, not all of the independent variables are expected to affect the dependent variable in all three types of dual executives. I, therefore, run several regression models with different constellations of control variables. In cases where all the three above-mentioned variables with missing cases fail to reach the level of statistical significance in the extensive model, I only include in subsequent models, the variable which was closest to reaching the level of statistical significance in the extensive model. I start by giving an overview of the patterns of association in the whole population. In Table 5.6, the dependent variable accounts for whether or not the powers of the head of state have increased a given year. The results show that powers of the head of state are particularly likely to increase in systems where the president is popularly elected. A non-popular election of the head of state also increases the probability for increasing powers when compared to systems with a monarch as head of state, but this relationship is weaker. In all but one of the regressions, we detect no association between the lagged variable measuring the powers of the head of state and the dependent variable, suggesting that the likelihood of an increase in the powers of the head of state is not affected by the previous level of power. In the regression models with all control variables except cohabitation included, there is a strange weak positive relationship between the variables, but this association is not detected in the regressions with a more limited number of independent variables and, consequently, with a higher number of cases included in the analyses. It is noteworthy that none of the control variables is statistically linked to the dependent variable in any of the regression models. We can thereafter shift focus on explaining decreasing levels of powers of the head of state. This is done for the whole population in Table

0.471 (0.363) 1.601 −0.058 (0.084) 0.944 −0.008 (0.010) 0.992 0.029 (0.076) 1.029 −0.155 (0.386) 0.857 −3.260*** (0.747) 0.038 0.026 1,236 −0.462 (0.374) 0.630 −3.393*** (0.554) 0.034 0.024* 1,482

−0.189 (0.271) 0.828 −3.433*** (0.198) 0.032 0.022** 2,119

0.676** (0.239) 1.965

0.606 (0.313) 1.833

−0.003 (0.008) 0.997

– Default

−4.517*** (0.801) 0.011 0.073*** 1,965

Default 1.302* (0.533) 3.676 1.692** (0.512) 5.428 −0.077 (0.084) 0.926 −0.007 (0.010) 0.993 −0.010 (0.096) 0.990

−4.882*** (0.431) 0.008 0.074*** 2,363

Default 1.395** (0.476) 4.035 2.016*** (0.453) 7.508 −0.042 (0.059) 0.958

−4.950*** (0.274) 0.007 0.082*** 3,969

Default 1.441*** (0.325) 4.225 2.030*** (0.307) 7.611

0.073 (0.052) 1.075

0.090 (0.060) 1.094

0.200* (0.084) 1.221

0.032 (0.055) 1.033

0.071 (0.064) 1.074 – Default

Reg. 6

Reg. 5

Reg. 4

Reg. 3

Reg. 2

In each cell, regression coefficients are listed first, with standard errors in parentheses, followed by odds ratios ***p < 0.001, **p < 0.01, *p < 0.05

Nagelkerke R 2 N

Constant

Cohabitation

Party system fragmentation in lower chamber

Government support in parliament

Number of parties in government

Popular election

– Default

0.153 (0.087) 1.165

Powers of head of state (lag 1)

Selection mode Inheritance Non-popular election

Reg. 1

Explaining increasing powers of the head of state 1850–2019. Logistic regression analyses

Independent variables

Table 5.6

142 C. ANCKAR

5

PATTERNS OF TRANSFORMATION IN DUAL EXECUTIVES

143

5.7. Interestingly enough, we now find a consistent positive association between the lagged power index and decreasing levels of powers of the head of state. This finding shows that powers of the head of state are more likely to decrease when existing powers are extensive than when they are limited. We also note that there is a positive association between non-popularly elected presidents (compared to monarchs) and decreasing levels of powers. This finding is interesting, as it suggests that non-elected presidents are more likely to lose powers than monarchs. Recall, that in Table 5.6, findings showed that non-popularly elected presidents were more likely to receive increasing powers compared to monarchs. These results indicate that in comparison with monarchs, the powers of nonpopularly elected presidents tend to shift not only upwards along the scale, but also downwards. Furthermore, it is of interest to note that in all regressions where monarchy constitutes the reference category, coefficients for popularly elected presidents are not significant, indicating that popularly elected presidents are not more unlikely to lose powers than monarchs. Finally, we note that in line with the findings when explaining increasing powers of the head of state, the control variables are unimportant for explaining decreasing powers of the head of state. We then turn to assessing patterns of association within the three respective dual executive categories. Tables 5.8 and 5.9 present results for systems where the president is not elected by popular vote. The results are easily interpreted: none of the variables are important for explaining increasing powers of the president but when we turn to explaining decreasing powers of the president, a different picture emerges. As shown in Table 5.9, the previous level of presidential powers is consistently positively related to decreasing powers of the president. In other words, powerful presidents are more likely to lose powers than weak presidents. The control variables, on the other hand, are irrelevant as explanations of why the powers of presidents decrease. Moving on to the category of popularly elected presidents, results are presented in Tables 5.10 and 5.11. Here, the results are, in one sense, a complete reflection of the findings for non-popularly elected presidents. Based on Table 5.10, we reach the conclusion that the previous level of presidential powers does not explain why presidential powers increase. However, as shown in Table 5.11, the previous level of presidential powers is positively related to decreasing presidential powers. It is

0.024 (0.041) 1.024 0.002 (0.322) 1.002 −3.750*** (0.278) 0.024 0.082*** 1,650

−0.032 (0.064) 0.968 −3.280*** (0.159) 0.038 0.072*** 2,119

−0.798** (0.250) 0.450

−0.563 (0.309) 0.570

−0.923* (0.368) 0.397 −0.117 (0.073) 0.890 −0.007 (0.009) 0.993 0.075 (0.045) 1.078 0.034 (0.371) 1.035 −3.177*** (0.624) 0.042 0.102*** 1,236 −4.684*** (0.641) 0.009 0.132*** 1,965

Default 1.491** (0.441) 4.441 0.528 (0.483) 1.696 −0.127 (0.071) 0.881 −0.006 (0.008) 0.994 0.070 (0.045) 1.072

In each cell, regression coefficients are listed first, with standard errors in parentheses, followed by odds ratios ***p < 0.001, **p < 0.01, *p < 0.05.

Nagelkerke R 2 N

Constant

Cohabitation

Party system fragmentation in lower chamber

Government support in parliament

Number of parties in government

Popular election

– Default

−4.593*** (0.334) 0.010 0.099*** 2,363

Default 1.166** (0.361) 3.208 0.197 (0.407) 1.217 −0.012 (0.053) 0.988

−4.334*** (0.184) 0.013 0.082*** 3,969

Default 1.064*** (0.218) 2.898 0.298 (0.234) 1.347

0.369*** (0.046) 1.446

0.411*** (0.066) 1.508

0.521*** (0.084) 1.684

0.383*** (0.056) 1.466

0.433*** (0.071) 1.542 – Default

Reg. 6

Reg. 5

Reg. 4

Reg. 3

Reg. 2

– Default

0.507*** (0.088) 1.661

Powers of head of state (lag 1)

Election mode Inheritance Non-popular election

Reg. 1

Explaining decreasing powers of the head of state 1850–2019. Logistic regression analyses

Independent variables

Table 5.7

144 C. ANCKAR

0.364 (0.247) 1.439 −0.523 (0.780) 0.592 −0.094 (0.172) 0.910 −0.006 (0.022) 0.994 0.032 (0.075) 1.032 −3.429* (1.420) 0.032 0.024 536

Powers of head of state (lag 1)

−3.537*** (0.499) 0.029 0.012 632

0.166 (0.208) 1.181 −0.750 (0.758) 0.472 −0.009 (0.135) 0.991

Reg. 2

In each cell, regression coefficients are listed first, with standard errors in parentheses, followed by odds ratios ***p < 0.001, **p < 0.01, *p < 0.05

Nagelkerke R 2 N

Constant

Party system fragmentation in lower chamber

Government support in parliament

Number of parties in government

Cohabitation

Reg. 1

Independent variables

−3.387*** (0.251) 0.034 0.0004 1,000

−0.065 (0.208) 0.937

−0.060 (0.205) 0.941 −0.599 (0.542) 0.549

−3.289*** (0.259) 0.037 0.007 1,000

Reg. 4

Reg. 3

Table 5.8 Explaining increasing powers of the head of state 1850–2019 in democratic republics where the president is not popularly elected. Logistic regression analyses 5 PATTERNS OF TRANSFORMATION IN DUAL EXECUTIVES

145

1.297*** (0.209) 3.658 −0.552 (0.663) 0.576 −0.242 (0.159) 0.785 0.002 (0.016) 1.002 0.045 (0.048) 1.046 −4.309*** (1.135) 0.013 0.303*** 536

Powers of head of state (lag 1)

−4.206*** (0.477) 0.015 0.238*** 632

0.926*** (0.142) 2.524 −0.622 (0.639) 0.537 0.010 (0.113) 1.010

Reg. 2

In each cell, regression coefficients are listed first, with standard errors in parentheses, followed by odds ratios ***p < 0.001, **p < 0.01, *p < 0.05

Nagelkerke R 2 N

Constant

Party system fragmentation in lower chamber

Government support in parliament

Number of parties in government

Cohabitation

Reg. 1

Independent variables

−3.945*** (0.239) 0.019 0.163*** 1,000

0.827*** (0.109) 2.286 −0.014 (0.113) 0.986

Reg. 3

−3.949*** (0.237) 0.019 0.163*** 1,000

0.827*** (0.109) 2.287

Reg. 4

Table 5.9 Explaining decreasing powers of the head of state 1850–2019 in democratic republics where the president is not popularly elected. Logistic regression analyses

146 C. ANCKAR

0.129 (0.095) 1.137 −0.029 (0.450) 0.971 −0.040 (0.098) 0.961 −0.009 (0.012) 0.991 −0.005 (0.153) 0.995 −2.623** (0.949) 0.073 0.012 700

Powers of head of state (lag 1)

−2.733*** (0.590) 0.065 0.007 851

−0.004 (0.009) 0.996

0.069 (0.068) 1.072 −0.310 (0.432) 0.733

Reg. 2

In each cell, regression coefficients are listed first, with standard errors in parentheses, followed by odds ratios ***p < 0.001, **p < 0.01, *p < 0.05

Nagelkerke R 2 N

Constant

Party system fragmentation in lower chamber

Government support in parliament

Number of parties in government

Cohabitation

Reg. 1

Independent variables

−2.817*** (0.198) 0.060 0.009 1,119

0.045 (0.057) 1.046 −0.057 (0.131) 0.944

Reg. 3

−2.875*** (0.193) 0.056 0.002 1,119

0.056 (0.057) 1.057

Reg. 4

Table 5.10 Explaining increasing powers of the head of state 1850–2019 in democratic republics where the president is popularly elected. Logistic regression analyses 5 PATTERNS OF TRANSFORMATION IN DUAL EXECUTIVES

147

0.329** (0.096) 1.390 0.292 (0.463) 1.339 −0.093 (0.096) 0.912 −0.010 (0.011) 0.990 0.191 (0.144) 1.210 −3.791*** (0.959) 0.023 0.064* 700

Powers of head of state (lag 1)

0.097 (0.120) 1.102 −4.029*** (0.518) 0.018 0.047** 907

0.270*** (0.075) 1.310 0.068 (0.396) 1.070

Reg. 2

In each cell, regression coefficients are listed first, with standard errors in parentheses, followed by odds ratios ***p < 0.001, **p < 0.01, *p < 0.05

Nagelkerke R 2 N

Constant

Party system fragmentation in lower chamber

Government support in parliament

Number of parties in government

Cohabitation

Reg. 1

Independent variables

−3.531*** (0.248) 0.029 0.050*** 1,119

0.238*** (0.060) 1.269 −0.052 (0.137) 0.949

Reg. 3

−3.591*** (0.243) 0.028 0.046*** 1,119

0.247*** (0.059) 1.281

Reg. 4

Table 5.11 Explaining decreasing powers of the head of state 1850–2019 in democratic republics where the president is popularly elected. Logistic regression analyses

148 C. ANCKAR

5

PATTERNS OF TRANSFORMATION IN DUAL EXECUTIVES

149

worth noting, though, that within the category of popularly elected presidents the odds ratios for the lagged powers of the head of state are quite close to 1, and lower than within the category of non-popularly elected presidents, indicating that the importance of the variable is fairly limited. Once again, the control variables turn out to be insignificant. Finally, I present results for the category of monarchies. However, since there are rather few cases where the powers of the monarchs have changed, running regressions even with a limited number of independent variables makes the models unstable, especially since some of the control variables have quite a few missing cases. To partly remedy this shortcoming, I conducted initial analyses where each of the control variables was included in regressions along with the lagged power scale. As it turned out, with respect to increasing powers of the head of state, the only control variable that was significantly associated with the dependent variables was party system fragmentation in parliament. With respect to decreasing powers of the head of state, only population size reached the level of statistical analyses. These variables were therefore the only control variables to be included in the regression analyses. Results are shown in Table 5.12, and they are in line with previous findings in the sense that there is a positive relationship between the lagged power index and decreasing powers of the head of state. However, now a strange weak positive association between the lagged power index and increasing powers of the monarch is detected, suggesting that the powers of the monarch are predicted to increase when the powers of the monarch are strong. However, this association disappears when party system fragmentation is included in the regression. The direction of association between party system fragmentation and increasing powers of the head of state also runs contrary to our theoretical assumption. In other words, an increase in party system fragmentation is associated with a lower likelihood that the powers of the monarch increase. With regard to the relationship between population size and decreasing powers of the head of state, findings are in line with our expectations. The likelihood that the powers of the monarchs decrease grows with increasing population size.

5.4

Discussion

The empirical analyses have returned a number of results that we need to analyse further. The most palpable results from the multivariate regressions concerned the relationship between the lagged power dimension

−5.279*** (0.369) 0.005 0.024 1,850

0.313* (0.148) 1.367

Reg. 2

In each cell, regression coefficients are listed first, with standard errors in parentheses, followed by odds ratios ***p < 0.001, **p < 0.01, *p < 0.05

Nagelkerke R 2 N

Constant

Population size (log) −2.242 (−1.419) 0.106 0.118* 1,115

0.283 (0.274) 1.327 −1.241* (0.591) 0.289

Powers of head of state (lag 1)

Party system fragmentation in lower chamber

Reg. 1

Independent variables

Dependent variable: Power increase

0.211*** (0.082) 1.235 −7.711 (1.373) 0.0004 0.049 1,855

0.524*** (0.116) 1.689

Reg. 1

−4.243*** (0.222) 0.014 0.031** 1,850

0.317*** (0.090) 1.374

Reg. 2

Dependent variable: Power decrease

Table 5.12 Explaining increasing and decreasing powers of the head of state 1850–2019 in democratic republics with a monarch as head of state. Logistic regression analyses

150 C. ANCKAR

5

PATTERNS OF TRANSFORMATION IN DUAL EXECUTIVES

151

and changes in the powers of the head of state. The findings quite consistently showed that previous levels of power were not related to increasing levels of powers of the head of state. However, in all three categories of dual executives, the likelihood of reducing the powers of the head of state grew as we moved upwards along the power scale. In other words, whereas systems with weak heads of state tend to be stable, systems with powerful heads of state tend to transform into systems with weaker heads of state. These patterns partly contradict the expected transformation patterns laid out in Table 5.1. With regard to presidents not elected by popular vote and monarchs, results are well in line with our assumptions. Especially in monarchies, but also to a lesser extent in republics with nonpopularly elected presidents, powerful heads of state conflict with the principle of democratic legitimacy. Therefore, as we expected, powerful heads of state in these systems are likely to be stripped of their powers. What is more surprising is the fact that the same pattern was observed also in countries with popularly elected presidents. Although the odds ratios were smaller than in the category of non-popularly elected presidents, findings were consistent in all regression models. This means that the more powers the president possesses, the higher the likelihood that the president loses powers. Furthermore, although the regression analyses confirmed the pattern of an association between the lagged power scale and a decrease of the powers of the head of state in all three categories of countries, they did not answer the question if, and to what extent, the relationships were consistent through all levels along the power scale. We could very well expect there to be thresholds effects operating. For instance, when non-popularly elected presidents’ powers reach above a certain value, the likelihood that they result in a subsequent reduction of powers in the following years increases substantially. A closer look at the predicted probabilities makes it possible to see if this is, indeed, the case. Figures 5.1–5.3 show three graphs in which the lagged power dimension is plotted against the predicted probabilities from bivariate regressions for each of the three dual executive variants (Fig. 5.2). For countries with popularly elected presidents and countries with a monarch as head of state, the relationship is quite linear, indicating that no thresholds effects whatsoever are discernible. The situation in countries with non-popularly elected presidents is different in two respects.

152

C. ANCKAR

Fig. 5.1 Predicted probability of loss of powers at t according to amount of powers at t − 1 in dual executives with non-popularly elected presidents

Fig. 5.2 Predicted probability of loss of powers at t according to amount of powers at t − 1 in dual executives with popularly elected presidents

5

PATTERNS OF TRANSFORMATION IN DUAL EXECUTIVES

153

Fig. 5.3 Predicted probability of loss of powers at t according to amount of powers at t − 1 in dual executives with a hereditary monarch

For one thing, we note that the relationship displays a more exponential pattern, indicating that the likelihood of the president losing power increases more and more when we move upwards on the power scale. This exponential pattern, however, is far from conspicuous. What is more interesting is the fact that the risk levels of losing powers are much higher in the category of non-popularly elected presidents than in the other two categories. The risk that a head of state scoring 4.00 on the power scale loses powers is 5 per cent in monarchies and 7 per cent in countries with popularly elected presidents. The corresponding value for a non-popularly elected president is much higher, 34 per cent. These results indicate that non-popularly elected presidents in particular are at the risk of losing powers. It is somewhat surprising to note that a corresponding result was conspicuous by its absence in monarchies. Since the democratic legitimacy of the power holder is non-existent in monarchies, we would expect monarchs to lose powers particularly in cases where the monarch is powerful. Yet this is clearly not the case. One plausible explanation is that changes along the power dimension are exceptionally rare in dual executives, and that results are highly influenced by a few cases. This,

154

C. ANCKAR

however, is not necessarily the case. The statement is true for Liechtenstein, which has scored the same value, 5.0, throughout its existence as a democracy (from 1921), but in Luxembourg the values have fluctuated between 3.67 and 0.58 since the year 1900. Bhutan is another illustrative case. It has met the criteria of democracy since 2009 and from that point in time it has scored the values 1.83 (2009–2012), 2.33 (2013–2015), 4.08 (2016), 3.75 (2017–2018), and 2.25 (2019). Another plausible explanation could be that monarchies tend to be characterized by some feature that counteract pressures for the reduction of monarchical powers in situations where the monarch is strong. Based on the literature in the field, we expected population size to fulfil this function. The results from the regression analyses confirmed that population size interacts with decreasing (but not increasing) powers of the monarch in regressions. This is an indication that although population size does not explain why monarchs receive more powers, it performs well in explaining why monarchs retain powers. Based on this finding, we can expect that powerful monarchs are to be found particularly in small entities. Indeed, a Pearson correlation analyses between population size in its logarithmized version and the power index among the 1,894 cases of dual executives with a monarch as head of state yields a correlation coefficient of −0.65 (significant at p0.00–0.99 1.00–2.99 3.00–4.99 5.00–7.00 Total

0 0 0 1 0 1

2 4 9 2 0 17

0 1 0 0 0 1

2 8 14 3 0 27

0 0 1 0 0 1

0 2 0 0 0 2

0 1 4 0 0 5

Each column returns, for each power category, the total number of cases

number of cases is rather low, I choose, in each cell, to return absolute numbers rather than percentages.2,3 Based on Table 6.1, we reach the conclusion that in dual executives with non-popularly elected presidents, most of the transformations from democracy to autocracy occur in situations where the powers of the president are moderate. What is also noteworthy is the fact that dual executives with non-popularly elected presidents to a very high extent transform into military regimes. Altogether, there are seventeen examples of this pattern. Four of these cases refer to transitions to authoritarian rule in African countries shortly after they received their independence. This was the case in Sudan 1958, Nigeria 1966, Somalia 1969, and Ghana 1972. Poland 1926, Portugal 1926, and Spain 1937, in turn, are examples of democracies that transformed into right-wing nationalist or fascist military regimes during the inter-war era in Europe. The rest of the cases are made up of Myanmar 1958, 1962, Sudan 1989, Pakistan 1977, 1999, 2018, Suriname 1980, Turkey 1980, Fiji 2006, and Bangladesh 2007. Five countries have switched from dual executives to personalist systems. Lithuania 1926 and Latvia 1934 resemble the cases of Poland, Spain, and Portugal in the sense that their dual executives were substituted with authoritarian right-wing governments under the leaderships of Anatanas Smetona in Lithuania and K¯arlis Ulmanis in Latvia. The same conclusion is valid for Austria, where democracy paved way to authoritarianism in 1933, first under Chancellor Engelbert Dollfuss and from 1934

6

HOW DUAL EXECUTIVES END

169

under Chancellor Kurt Schuschnigg. After Indonesia’s first general election, held in 1955, the country met the criteria of democracy for a short period of time during which it had a dual executive. However, already in 1957, President Sukarno concentrated power into his own hands, and the country moved to the category of authoritarian states where it remained until 1999. Albania, again, passed the threshold of democracy in 1992 and has remained in the camp ever since, except for the year 1996, when highly flawed parliamentary elections were held. Moving on, there are two cases of transitions to multi-party authoritarian rule, Sri Lanka in 1978 and Bangladesh in 2014. In Sri Lanka, the United National Party won a landslide victory in the 1977 elections. Its leader, J. R. Jayawardene was elected president and swiftly changed the constitution, enhancing the powers of the president to such an extent that the country no longer met the criteria of democracy. In Bangladesh, the Bangladesh Awami League, has been in power since the year 2009 under the leadership of Sheikh Hasina. The elections that took place in 2014 were marred with irregularities, boycotts and protests and effectively ended the democratic era. There is also one case where a single-party regime has emerged from a dual executive. Czechoslovakia experienced a short democratic period after the World War II before the communists, with Soviet support, eventually took power. There is one transition to an oligarchic system, namely in Fiji in 2000. For the first time in history, the election in 1999 brought an Indo-Fijian politician, Mahendra Chaudry, to the position of prime minister. In the ethnically divided country, tensions were high and one year later, Fijian nationalists launched a coup whereby Chaudry was ousted of office. The political system of Fiji at that time is considered to be oligarchic, due to the exclusion of the Indo-Fijian population from the political sphere along with the strong position of the Great Council of Chiefs, which among other things, still elects the president of Fiji. Finally, dual executives with non-popularly elected presidents almost never transform into single-head executives. The only exception to this rule is Belarus, where a dual executive was substituted with a presidential form of government in 1994. However, the case of Belarus is exceptional since there seemed to be quite a lot of confusion regarding the relationships between the legislature and the executive. The constitution did provide for a position of a prime minister, but it did not contain any provisions indicating that the prime minister could be ousted of office by a parliamentary vote of no confidence. Instead, article 107 explicitly stated

170

C. ANCKAR

that the prime minister was appointed and dismissed by the president. The constitution was amended shortly thereafter, in 1996, with provisions making the government dependent on the legislature. However, since other amendments gave the president very wide powers Belarus no longer qualified as a democracy from that year (see also Arkadyev, 2008). The picture is different in countries with popularly elected presidents. Based on findings presented in Table 6.2, we discover that more than three quarters of the regime transformations have occurred in countries situated in the highest power category. Contrary to the patterns observed for countries with non-popularly elected presidents, nearly half of the conversions are towards the personalist regime type. It is noteworthy that many of these cases refer to situations where a powerful president in a democratic system has taken measures to further increase his powers in order to strengthen his position or by attempting to sidestep constitutionally defined term limits. Thus, the system has moved further and further away from the democratic camp towards the authoritarian one under the leadership of the same president. This has most typically been the case in Russia under Putin and in Turkey under Erdogan but also in Peru under Fujimori, Sri Lanka under Rajapaksa, Guinea-Bissau under Vieira, and Niger under Tandja. The only country that do not adhere to this pattern is Madagascar, where Andry Rajoelina’s transitional government (which lasted for five years) rose to power subject to a military coup in 2009 and Germany in 1933, after Hitler’s seizure of power. Table 6.2 Powers of the head of state and transformation to other regime types from dual executives with popularly elected presidents Regime type at t Power category of HOS at t−1

Presidential

Multi-party authoritarian

Military

Personalist

Total

0.00 >0.00–0.99 1.00–2.99 3.00–4.99 5.00–7.00 Total

0 0 0 0 2 2

0 0 0 0 3 3

0 0 0 3 4 7

0 0 2 1 5 8

0 0 2 4 14 20

Each column returns, for each power category, the total number of cases

6

HOW DUAL EXECUTIVES END

171

Table 6.3 Powers of the head of state and transformation to other regime types from dual executives with a monarch as head of state Regime type at t Power category Military of HOS at t−1

Personalism

Absolute monarchy

Monarchic Oligarchy

Other oligarchy

Total

0.00 >0.00–0.99 1.00–2.99 3.00–4.99 5.00–7.00 Total

0 1 1 2 0 4

0 0 0 2 0 2

0 0 1 0 0 1

1 1 0 0 0 2

1 2 5 6 0 14

0 0 3 2 0 5

Each column returns, for each power category, the total number of cases

Transitions towards military regimes also occurred in more than a third of the cases. The cases in question are South Korea 1961, Cuba 1952, Peru 1962, 1968, Niger 1996, Central African Republic 2003, and Mali 2012. Finally, there are three cases where dual executives have transformed into multi-party authoritarian states. The cases in question, however, do not refer to radical changes in the regimes of the countries, but rather reflect small declines in the level of democracy in countries situated at the threshold between democracy and autocracy. Accordingly, small drops at V-dem’s electoral and liberal democracy scales turned Tanzania into the camp of authoritarian states in 2001 and 2016, whereas Boix et al., (2013) transmitted Mozambique to the category autocracies in 2004, presumably as a consequence of irregularities in the general elections held the same year. Interestingly enough, transitions towards a presidential form of government are extremely rare. The only cases are made up of Brazil 1963 and Kenya 2013. Both of the cases refer to a return to the old, ‘normal’ regime form after short interlude of dual executive rule. In Brazil, the sudden resignation of president Jânio Quadros in 1961 created fears among the military (and in the United States) that his left-leading VicePresident Joao Gulart would prove unreliable (Loureiro, 2014, 333). As a precautionary measure, the parliament adopted an amendment to the constitution whereby the office of prime ministers was established and the powers of the president reduced. The prime minister and the government where subject to a parliamentary vote of no confidence. Tancredo

172

C. ANCKAR

Neves assumed the position as prime minister, but dissatisfaction with the government was widespread and in 1963, after a plebiscite, Brazil returned to presidentialism. In the presidential election in Kenya in 2007, President Mwai Kibaki and the leader of the opposition, Raila Odinga, fought a close race. Eventually, Kibaki was declared winner, sparking heavy protests and acts of violence. After interventions from the international community and under increasing pressure, Kibaki and Odinga signed an agreement, which, among other things, included a power sharing arrangement. Kibaki held on to the presidency whereas Odinga assumed the newly established position as prime minister. However, this was, indeed, a temporary arrangement. A new constitution was adopted in 2010, which re-established the single executive and the position as prime minister was abolished after the presidential elections in 2013. Turning, then, to the category of monarchies, Table 6.3 reveals a more dispersed pattern. Democracies with a monarch as head of state transform into various types of autocracies in fairly equal proportions. Transitions occur at different levels along the power scale, although most of the democratic breakdowns occur in situations where the monarch is in possession of fairly strong or strong powers. We also note that it is extremely rare that constitutional or semi-constitutional monarchies transform into absolute monarchies. It has happened on two occasions, in Yugoslavia in 1929 and Nepal in 2002. Yugoslavia, (or, at that time, The Kingdom of Serbs, Croats, and Slovenes) adopted a democratic constitution in 1921, which gave King Alexander I extensive powers. The unity of the newly established Kingdom was threatened as there were strong pressures to introduce a federal state. On 6 January 1929, in an attempt to unite the country, the King abolished the constitution and concentrated powers into his own hands. In Nepal, democracy was introduced in 1991 after the country had been ruled as an absolute monarchy throughout its existence. A decade later, King Gyanendra postponed the scheduled elections and concentrated executive powers into his own hands, throwing Nepal back to authoritarian rule. However, dissatisfaction with the King was widespread and the political parties united in an effort to restore democracy. In parallel, a grass-root revolution with a leftish agenda, the so-called Maoist insurgency had been going on for many years (Thapa & Sharma, 2009). In April 2006, the King was forced to give up his powers. The constituent

6

HOW DUAL EXECUTIVES END

173

assembly, elected in 2008, finally abolished the monarchy and Nepal became a republic and has remained in the democratic camp ever since. There are two cases where countries have adopted an oligarchic regime, after the democratic breakdown. Pakistan 1956 and Fiji 1987 were formally ruled by the British monarch when democracy broke down. Both countries subsequently returned to democratic rule, but this time under republican constitutions. The case of Laos in 1959 is special. The country had become independent in 1953, under a constitution that granted the King extensive powers. However, in 1959, the military forced the prime minister of Laos to resign, and executive power was transferred to ‘the three princes of Laos’, i.e. three persons of royal descent, representing either royal, neutral, or communist views. However, the King remained influential, and the system introduced in Laos could best be described as a monarchic oligarchy. A well-known transfer from a dual executive with a monarch as head of state to personalist rule took place in Italy in 1922. After his ‘March on Rome’, Benito Mussolini was appointed prime minister and the fascist era began. A much less known transition in the same direction occurred in the Solomon Islands in 2000, when Prime Minister Ulufa’alu was captured by rebel forces and the opposition was able to replace him by Manasseh Sogavare. Monarchic regimes have transferred into personalist systems on two occasions in Thailand (1976 and 1991). However, as we have seen, in Thailand the position of the King has largely remained the same in times of democracy and autocracy, In addition, drawing the line between military and civilian rule is not always easy. The military committed a coup in October 1976, after which the King appointed a civilian prime minister, Thanin Kraivichien. Disappointed with the civilian government, the military launched another coup in 1977, successfully removing Thanin from power. The regime transformation in 1991 followed a similar pattern. The military launched a coup, but swiftly turned power over to a civilian government over which the military exercised heavy influence. The two remaining democratic breakdowns in Thailand, in 2006 and 2014, are classified as transformations to military regimes. On both occasions, the military launched coups against democratically elected regimes. Between 2006 and 2008, the country had an interim government led by the former Supreme commander of the Thai army, Surayud Chulanont. The coup in 2014, again, was led by General Prayut Chan-o-cha, who has remained prime minister of Thailand ever since. Two of the cases of transitions to military regimes occurred in Greece, more specifically in 1936

174

C. ANCKAR

and 1967. As we recall, Greece is unique in the sense that it has switched from a democratic republic to a democratic monarchy. This occurred in 1935, when monarchy was re-introduce after a referendum. However, the referendum was called (and most likely heavily rigged) by Greece’s strongman, general Georgios Kondylis, whose ambition was to introduce a Greek variant of Mussolini’s regime form, where the monarch’s main function was to legitimize the rule of Kondylis himself. Greece did indeed return to authoritarian rule already in 1936, not under Kondylis, who had by then died of a heart attack, but under Ioannis Metaxas. The other regime transformation in Greece occurred in 1967, when a military junta took power and King Constantine II had to flee the country. Democracy was restored in 1974, but the same year Greek voters in a referendum opted for a republican form of government. The final case in the category is Sierra Leone, which short experience as a democracy ended with a military coup in 1967.

6.4

Multivariate Patterns

The discussion above has focused on providing a detailed description of the patterns of transformations from dual executives to other regime types. The basic argument was that the powers of the president in combination with the mode of selection of the head of state were crucial for explaining why dual executives die. However, it is evident that there are many other plausible reasons for why dual executives end. In the next part of this chapter, I shall introduce a number of alternative explanations for why dual executives transform into other regime forms. These potential explanatory variables are largely derived from the literature on democratic breakdown. Some of the variables discussed are specifically linked to dual executives whereas others have been showed to affect the risk of democratic breakdown in general. In Chapter 5, I made the assumption that a number of features linked to what Tavits (2009: 35–40) has referred to as the ‘political opportunity framework’ were expected to affect the possibilities of the head of state to exert influence on the policy process. It is evident that the same characteristics also can explain why dual executives do not survive. To begin with, I discussed the possibility that cohabitation, i.e. situations where the president and the prime minister represented different political parties, would affect the powers of the head of state. Although the empirical analyses failed to detect any such patterns of interactions, it is

6

HOW DUAL EXECUTIVES END

175

still possible that cohabitation can help to explain why dual executives die. In fact, there is much more theoretical and empirical evidence regarding the latter research question than the former. The basic argument is that cohabitation threatens system stability. If the president’s party is not represented in government, the potential for conflicts between the president and the prime minister is high. Presidents can attempt to impede the work of the government in several ways, ultimately, if he or she has the power to do so, by dismissing the government or by dissolving the parliament (Elgie & Schleiter, 2011, 46–47). Periods of cohabitation could be particularly problematic for democratic survival in neopatrimonial states where ‘the struggle by opposing parties to compete for government patronage simultaneously, through a politically divided executive, inevitably leads … to severe political breakdown…’ (Kirschke, 2007, 1390). Several authors have warned against the perils of cohabitation for system stability, often referring to examples from individual countries (Ghergina & Miscoiu, 2013; Linz & Stepan, 1996, 286; Stepan & Suleiman, 1995; Suleiman, 1980; Villalón & Idrissa, 2005). At the same time, systematic comparisons assessing the effects of cohabitation are relatively scarce. One finding shows that in Eastern Europe, cohabitation was related to higher levels of intra-executive conflicts, but that its impact was overshadowed by whether or presidential and parliamentary elections were held concurrently or not (Sedelius, 2006, 193–194). Some evidence also suggests that periods of cohabitation can lead to democratic breakdowns, especially in newly democratized entities (Kirschke, 2007). Other studies, again, have failed to detect any association between cohabitation on the one hand and government instability or regime collapse on the other (Elgie, 2008; Elgie & Schleiter, 2011; Fernandes & Magalhães, 2016; Kim, 2015; Schleiter & Morgan-Jones 2009; Sedelius & Ekman, 2010, 519). The view that cohabitation is a vice for the functioning of semipresidentialism does not stand unchallenged, however. Some authors have pointed out that cohabitation makes it possible for power sharing arrangements between representatives of different political camps, and particularly that it prevents the president from becoming too dominating at the expense of the parliament (Moestrup 2007; Samuels & Shugart 2010, 260). This line of reasoning is very much in line with Sartori’s (1997, 125) argument that in semi-presidential systems ‘the problem of

176

C. ANCKAR

divided majorities finds a solution by “head-shifting”, by reinforcing the authority of whoever obtains the majority’. Although cohabitation is likely to increase the likelihood of regime collapse in dual executives, it is far from obvious that it alone makes democratic systems collapse. In her extensive comparison of France and Weimar Germany, Cindy Skach (2005) reached the conclusion that semipresidentialism is likely to suffer breakdowns particularly in cases where countries are ruled by divided minority governments, i.e. situations where ‘neither the president nor the prime minister, nor any party or coalition, enjoys a substantive majority in the legislature’ (Skach, 2005, 15). This conclusion is supported by Russia’s brief, and failed, period as a democracy during the Yeltsin era (Colton & Skach, 2005). At the same time, the relationship between divided minority governments and regime collapse has either not been detected in large-N studies (Elgie & Schleiter, 2011; Kim, 2015) or, in cases where such tendencies have been observed, the results have been inconclusive (Elgie, 2008). Anyhow, it is evident that the government’s support base in parliament is likely to affect the stability of the regime. In cases where the government cannot rely on a parliamentary majority, there is a possibility that the president takes a more active role in politics, thus undermining the position of the prime minister and increasing the risk of system collapse, particularly in settings where democracy has not yet been consolidated. I shall therefore make use of two measures, one which only captures the extent to which the government is in control of the parliament, measured by the seat share of the parties in government in parliament, and one which accounts for the extent to which the divided minority government criterion is obtained. This happens in periods of cohabitation where the seat share of the parties in government in parliament is less than 50%. Another variable connected to the political opportunity framework discussed in Chapter 5 was party system fragmentation. Party system fragmentation is also relevant when discussing the possibilities that dual executives transform into autocracies. A recurring argument is that the presidential form of government is especially problematic for democratic consolidation in newly democratized countries, as there is an inherent risk that presidents challenge the democratic institutions and aim to transform themselves to ‘presidential strongmen’ (Linz, 1994; Linz & Valenzuela, 1994; Stepan & Skach, 1993). Scott Mainwaring (1993) in particular, has challenged, this notion, arguing instead that it is the combination of presidentialism and multi-partism that is detrimental for democratic stability.

6

HOW DUAL EXECUTIVES END

177

According to Mainwaring (1993, 223), ‘[s]ome problems typical of presidential systems—conflict between the executive and legislature resulting in immobilism, presidential efforts to bypass congress, congressional efforts to limit presidential action—are often exacerbated by multiparty systems’. In contrast, two-party systems, since they are less polarized ideologically and tend to bring about one-party majority governments, are likely to counteract the pitfalls of presidentialism. It is less clear, if party system fragmentation also contributes to the fall of dual executives. Based on Skach’s (2005) study referred to above, it makes sense to assume that party system fragmentation can be detrimental also for semi-presidential systems. After all, party system fragmentation is a key factor for understanding the failure of semi-presidentialism in Weimar Germany, where ‘the party system … was fragmented, zero-sum, polarized, and lacked the capability of forming a majority coalition’ (Shen, 2011a, b, 241). Correspondingly, the success of semi-presidentialism in Taiwan is partly explained by its less fragmented party system (Shen, 2011a, b). It is worth pointing out that, empirically, the association is not exclusively based on case studies; in an extensive study of semi-presidential systems, Elgie and Schleiter (2011) found evidence of a link between party system fragmentation and breakdown of semi-presidential regimes. However, as party system fragmentation was only included as a control variable, they did not elaborate on the finding. In addition to variables related to the opportunity framework, there are other plausible determinants of regime breakdown that we need to consider. First, it is evident that economic factors are likely to have an impact on whether democracy survives or not. Lipset’s (1959) proposition that socioeconomic development fosters democracy is probably one of the most researched areas in comparative politics. This is not the place for a detailed exposition of findings arrived at in all studies. Suffice it to say that Lipset’s proposition has received support in many studies at the same time as his general theory has been qualified and improved in a number of respects (Barro, 1999; Bollen, 1979, 1983; Boix, 2011; Burkhart & Lewis- Beck 1994; Cutright, 1963; Diamond, 1992; Epstein et al., 2006; Hadenius 1992; Knutsen et al., 2019; ; Teorell 2010; Vanhanen, 1990). However, there are also studies where any relationship between the variables has not been detected (Acemoglu et al., 2008, 2009). It is evident that many indicators of socioeconomic development could affect the propensity for democratic breakdown. The general expectation

178

C. ANCKAR

is that in periods of economic hardship, dissatisfaction among the public grows, resulting in increased likelihood of a regime breakdown (Przeworski & Limongi, 1997; Svolik, 2008). However, since the empirical material is extensive both in time and space, sophisticated measures of socioeconomic development cannot be used. I therefore operate with two indicators that are commonly used in studies in the field namely wealth, measured as GDP per capita and growth, measured as change in GDP per capita (Brownlee 2009; Elgie & Schleiter, 2011). For dual executives with a monarch as head of state I also include population size among the control variables. As discussed in Chapter 5, Corbett, et al. (2017) have made the argument that the prospects for powerful monarchs to survive in small settings are particularly favourable due to the institutional fidelity and personalization that are expected to characterize small societies. Recall, that the patterns of transformations in Chapter 5, showed that powerful monarchs tended to retain their positions particularly in the smallest countries. It therefore makes sense to assume that the risk that dual executives headed by monarchs transform into other regimes form is lower in smaller countries than in larger ones. Diffusion patterns are also likely to have an impact on regime breakdown. In Chapter 4, the density of similar regimes in the geographic proximity was introduced as a plausible explanation of why countries choose dual executives. It also makes sense to consider the variable when the ambition is to account for the end of dual executives. As we recall, diffusion is commonly included in studies on democratization. The prospects for democratization are expected to increase, with increasing shares of democracies in neighbouring countries. However, diffusion can also explain opposite trends. Several studies have shown that diffusion patterns are relevant for explaining the decline of democracy (Huntington, 1991; Kneuer & Demmelhuber, 2016, 2021; Weyland, 2016). Thereby it makes sense to assume that the share of dual executives in the geographical proximity is linked to the prospects for the survival of the regime type. Since the ambition is to explain the end of dual executives per se, and not transformation patterns among dual executives, I have opted for focusing on the overall share of dual executives in the region rather than using the three types of dual executives (nonpopularly elected presidents, popularly elected presidents, and monarchs). Like in Chapter 4, I use the variable Region (geographic) (E) (e_regiongeo) from the V-dem dataset for categorizing the regions of the world.

6

HOW DUAL EXECUTIVES END

179

Finally, there is reason to believe that the likelihood of democratic regime survival is dependent on how firmly the democratic regime form is rooted in society. Linz and Stepan’s famous argument stipulates that democratic consolidation occurs along three dimensions: behaviourally, attitudinally, and constitutionally, until democracy has become ‘the only game in town’ (Linz & Stepan, 1996, 5). In this constellation, time plays a crucial role. With the passing of time, democratic routines become established, a growing number of elections are held, and political office holders are substituted by other leaders, a fact which is of crucial importance for the consolidation of democracy (Lindberg, 2006, 99–141; Moehler & Lindberg, 2009). As existing political elites grow accustomed to working within the frames of the democratic institutions and new generations enter the political scene ‘a … process of Darwinian selectivity in favor of convinced democrats [occurs]’ (Rustow, 1970, 358). The aforesaid boils down to the assumption that the legitimacy of the democratic form of government is likely to increase with increasing time under democratic rule (Gerring et al., 2005; Huntington 1991, 260–265; Mishler & Rose, 2007). Once again, however, empirical findings regarding the relationship between democratic age and democratic stability is equivocal. A positive association between the variables has been verified by Svolik (2008) as well as Kapstein and Converse (2008); whereas findings from other studies have failed to detect such a relationship (Epstein et al., 2006; Przeworski et al., 2000). Clearly, the operationalization of what Gerring et al. (2005) refer to as ‘the democracy stock’ can be done in a number of ways. One alternative is to account for the general level of democracy over an extensive period of years. For example, Gerring et al. (2005) measures the ‘democracy stock’ by including democracy scores based on Polity 2 over a period of 100 years. Another possibility is to calculate the number of years during which a country meets the criteria of democracy over a certain period (Denk & Anckar, 2014). A third possibility is to only account for the period during which a country has been democratic prior to a regime breakdown or, in cases where no democratic failures have occurred, to the last year the study encompasses, which is the strategy that will be applied in the framework of the present research effort. The main reason for the choice of this strategy is that the theoretical propositions discussed above suggest that democracy consolidates through learning processes, where the political elites and the masses eventually grow accustomed to how the democratic institutions operate and

180

C. ANCKAR

the ‘rules of the game’. Although it is evident that previous democratic experiences far back in time can have an impact on how democracy takes root in newly democratized settings, it is more probable that each new period under democratic rule initiates a new process of democratic tutelage, where the risk of democratic setback is expected to decline gradually. In addition, one can expect that the risk of autocratic reversal should be imminent in situations where a previous autocratic reversal has occurred in a not so distant past. In such cases, a great deal of the population has experiences with autocratic rule and many persons, particularly those who were influential in or benefitted from, the old authoritarian regime can be quite sympathetic towards the idea of a return to autocracy. We can then turn to the empirical part, where the ambition is to assess to what extent the introduction of control variables affects the patterns of relationships that have emerged thus far. In the previous section, I have discussed a number of plausible determinants of regime breakdown. However, as the number of instances where dual executives have transformed into other regime forms is rather limited, running multiple regressions including all control variables is not a meaningful exercise. Altogether, there are 63 cases where dual regimes have broken down. However, with regard to the variables that refer to the opportunity framework there are a number of missing cases, which heavily reduces the number of instances with regime breakdowns in those regression models where the variables government support in parliament , divided minority government , and party system fragmentation in parliament are included. In order to remedy this shortcoming, I initially run regressions where each control variable is separately included in regressions along with the independent variables. Only the control variables that are statistically related to the dependent variable (p < 0.05) will be included in further analyses. In all regressions, the independent variables are measured one year prior to the regime breakdown. I begin with a presentation of the findings from the total population of cases. Results are given in Table 6.4. Initial analyses show that the following control variables are statistically linked to end of dual executives: GDP per capita, regional density of dual executives, and duration of democracy. These variables will be included in regressions along with selection mode and powers of the head of state. In line with our expectations, we find that the propensity for regime breakdown increases with increasing levels of powers of the head of state. This finding is consistent in all regression models. With regard to the

−5.680*** (0.319) 0.003 0.109*** 4,110

Reference 1.450*** (0.332) 4.265 −0.274 (0.386) 0.760 0.524*** (0.074) 1.690

0.806 (1.397) 2.238 0.196*** 3,611

Reference 0.801* (0.376) 2.227 −1.200* (0.528) 0.301 0.545*** (0.108) 1.724 −0.705*** (0.152) 0.494

Reg. 2

−0.024*** (0.006) 0.976 −0.045*** (0.012) 0.956 −3.102*** (0.433) 0.045 0.200*** 4,110

Reference 0.869** (0.335) 2.384 −0.568 (0.415) 0.567 0.337*** (0.086) 1.401

Reg. 3

Reference 0.717 (0.380) 2.048 −0.918 (0.492) 0.399 0.392*** (0.106) 1.480 −0.339* (0.169) 0.712 −0.020** (0.007) 0.981 −0.041** (0.015) 0.960 −0.504 (1.434) 0.604 0.237*** 3,611

Reg. 4

HOW DUAL EXECUTIVES END

In each cell, regression coefficients are listed first, with standard errors in parentheses, followed by odds ratios ***p < 0.001, **p < 0.01, *p < 0.05

Nagelkerke R 2 N

Constant

Democratic duration

Regional density of dual executives

GDP/cap (log)

Powers of head of state

popular election

Selection of HOS Inheritance non-popular election

Reg. 1

Explaining end of dual executives 1850–2019

Independent variables

Table 6.4

6

181

182

C. ANCKAR

mode of selection of the head of state, we find that a non-popular election mode (compared to inheritance) increases the likelihood of regime breakdown. We also note a weak tendency suggesting that systems with popularly elected presidents are more resistant to regime breakdown than systems with a monarch as head of state. This association, however, only reaches the level of significance in one of the regressions. With regard to the control variables, we note that GDP per capita is negatively associated with the dependent variable. This pattern is in line with our expectations; regime breakdowns are more likely to occur when the economy is in bad shape than in economically prosperous times. Regional density of dual regimes and duration of democracy are also connected to regime breakdowns in the expected directions. The likelihood of regime breakdown decreases with increasing shares of dual executives in neighbouring countries and with increasing years of democratic rule. Both of the variables retain their explanatory power in all models where they are included. Nevertheless, for both of these variables, the odds ratios are close to 1, meaning that their impact is weak. Let us thereafter turn to assessing relationships separately for the three categories of dual executives. I begin with the category of non-popularly elected presidents. Here, initial analyses show that the following control variables are statistically linked to end of dual executives: government support in parliament , GDP per capita, GDP growth, regional density of dual executives, and duration of democracy. Accordingly, these variables are eligible for inclusion in regressions along with the powers of the head of state. Results from the multiple regression analyses are presented in Table 6.5. A first observation is that the powers of the head of state do not affect the likelihood of regime breakdown. Although such an association is detected in the bivariate regression model, it disappears when control variables are added to the model. We further note that government support in parliament loses its explanatory value when included in regressions along with other control variables. In the most comprehensive model (regression 5), only democratic duration reaches the level of statistical significance. However, as the inclusion of government support in parliament reduces the number of cases quite profoundly, I ran an additional analysis without this variable (regression 6). The duration of democracy retained its explanatory value. In addition, economic wealth and

−8.161*** (1.585) 0.0000003 0.076* 652

0.182 (0.290) 1.199 0.061** (0.023) 1.063

Reg. 2

4.212* (1.954) 67.513 0.165** 859

−0.951*** 0.235 (0.386) −0.104** (0.033) 0.901

0.271 (0.195) 1.312

Reg. 3

−0.022** (0.008) 0.978 −0.080** (0.023) 0.923 −1.707*** (0.477) 0.181 0.153*** 1,056

0.148 (0.149) 1.159

Reg. 4

In each cell, regression coefficients are listed first, with standard errors in parentheses, followed by odds ratios ***p