Constitutional monarchies and semi-constitutional monarchies: a global historical study, 1800–2017

ABSTRACT

The issue of executive power sharing in democratic countries with a monarch as head of state has received little scholarly attention. In many ways this make perfect sense; since there is no place for a powerful hereditary monarch in a democratic system, one could argue that systems with powerful monarchs do not qualify as democracies. Nevertheless, there are many examples of political systems, classified as democracies by most reputable categorizations or indices, where the monarch has, or has had, more or less the same position as a president in semi-presidential systems. The aim of the present study is to study to what extent the occurrence of semi-constitutional monarchies, i.e. democratic regimes in which power is shared between a prime minister and a monarch, can be explained by reference to Huntington’s notion of the King’s dilemma and the size of countries. The study is global and encompasses the time period 1800–2017.

KEYWORDS:

• Monarchies
• King’s dilemma
• regime classification
• global historical comparison

Introduction

Half a century ago, Samuel P. Huntington (1968), when discussing regime transformations from autocracy to democracy, took the view that an absolute monarch who considered reaching a compromise by means of which he or she would retain some of his or her powers within the framework of a democratic system, was likely to get disappointed. In the long run, the monarch faced a zero-sum game; either try to retain his or her powers as an absolute monarch or be stripped of all powers and, at best, continue as a ceremonial head of state of a democracy. The basic logic behind this statement is simple: a leader who has not been elected by the people has little or no legitimacy to rule in a democratic polity.

At the same time, monarchies are not on the verge of extinction; currently there are approximately 30 democracies with a monarch as head of state and among authoritarian regimes, monarchies in particular have been shown to be very stable (e.g. Kailitz, 2013; Magaloni, 2008). Yet, the question how much powers monarchs possess has not aroused a great deal of interest among political scientists. Whereas, the relationship between presidents and prime ministers has been widely discussed in the literature on semi-presidentialism (e.g. Åberg & Sedelius, 2018; Brunlíc & Kubát, 2019; Cheibub et al., 2010; Duverger, 1980; Elgie, 1999; Sartori, 1997; Schleiter & Morgan-Jones, 2009), the issue of executive power sharing in democratic monarchies has been more or less completely neglected.

To some extent this makes perfect sense; since there is no place for a powerful hereditary monarch in a democratic system, one could easily argue that such systems do not qualify as democracies. On the other hand, there are many systems classified as democracies by most reputable categorizations or indices where the monarch has, or has had, more or less the same position as a president in a semi-presidential system. It is therefore essential to answer the questions why these systems – anomalies and anachronisms – emerge and persist.

To begin with, some conceptual clarifications are needed. Commonly, monarchies are classified into absolute monarchies and constitutional monarchies, but since this classification essentially is based on the powers the monarch possesses, the difference between the categories is one of degree rather than kind. In practice, the difference between the categories largely follows the dividing line between democracies and autocracies. Absolute monarchies are systems where the monarch, the hereditary ruler, possesses powers to such an extent that the countries in question do not qualify as democracies. These systems are of little interest for the purpose of the present study.

Monarchies that meet the criteria of democracy are generally considered constitutional monarchies. However, within this category of countries the powers of the monarch can still vary, which makes it necessary to apply a further categorisation. Following Corbett et al. (2017), I therefore introduce an additional category labelled ‘semi-constitutional’ monarchy. The authors define as semi-constitutional monarchies ‘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, p. 691).

In the present study, countries must be democratic in order to qualify as semi-constitutional monarchies. Countries that meet the criteria of democracy with a monarch as a head of state are consequently either constitutional monarchies or semi-constitutional monarchies. The crucial question is to draw a line between the two categories. Based on the definition by Corbett et al. the monarch in a semi-constitutional monarchy should ‘have the capacity to exert a large measure of political influence’, but it is of course very difficult to determine when the amount of influence is large enough to justify a classification as ‘semi-constitutional monarchy’.

Establishing cut-off points along the power scale is highly problematic and, to a certain extent, arbitrary. Here, I have chosen to apply a generous criterion for inclusion in the category semi-constitutional monarchies. Accordingly, all democracies where the monarch has held executive powers, legislative powers, powers over domestic policy, or powers to dissolve the legislature are defined as semi-constitutional monarchies and the rest of the cases, where the monarch is powerless on all dimensions, are conferred to the category constitutional monarchies.

This cut-off point makes sense for two reasons. First, it is uncontroversial, in the sense that it separates systems where the monarch has ceremonial powers only from systems where the monarch can exercise at least some influence in the political sphere. Second, since a monarch in a semi-constitutional monarchy lacks democratic legitimacy (in contrast to a president in a semi-presidential system) very powerful monarchs are, by definition, not possible within democratic contexts, and the threshold must consequently be situated at low levels of the power-scale.

Explaining powerful monarchs

The literature on the role of monarchs in democratic systems is scarce. Based on the few studies that have been conducted in the field, there are two plausible explanations for why powerful monarchs occur in democratic settings. First, since powerful hereditary heads of states do not sit well with democratic principles it is natural to consider systems with powerful monarchs as anomalies, which are likely to occur especially in countries which experience a transition from autocratic monarchical rule to democracy. Huntington (1968, pp. 177–191) launched the expression ‘the king’s dilemma’ in order to describe the challenges monarchs faced when trying to reconcile monarchic rule with strives for modernisation. According to Huntington, there were three options available for a monarch confronted with this dilemma: transformation, coexistence, and maintenance. The first alternative included a variety of ways in which power could be transferred from the monarch to another institution, such as the people, a bureaucratic elite, the military etc. The second option was to combine monarchical powers with democracy, or, in Huntington’s (1968, p. 180) words, ‘to institutionalize competitive coexistence in the polity of two independent sources of power’. The third strategy available for the monarch was to try to maintain his or her power, for instance by allowing persons with a middle-class background to receive high positions in the government, by fighting modernisation or by intensifying repression.

For the purpose of the present study, we are concerned with situations where absolute monarchies democratise, which means that the first and, particularly, the second strategies are relevant. To some extent, the two strategies overlap and merely reflect differences in degree rather than in kind. If the first option is chosen, power is transformed from the monarch to the people, whereby a democratic, constitutional monarchy emerges where ‘the king reigns but does not rule’ (Huntington, 1968, p. 177). In other words, it reflects a complete transformation of power, where the monarch is confined to the position as a ceremonial head of state. In the second alternative, the monarch preserves a substantial part of his or her powers, but coexists with democratic institutions, and, very explicitly, with a prime minister who emerges from and/or is responsible to parliament.

Since the monarch presumably can be expected to be reluctant to give up his or her powers, a handy compromise in a democratising 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. Such a development is very much in line with Huntington’s (1968, p. 178) observation that a radical shift from absolute monarchy to constitutional monarchy is highly unlikely, and would require ‘either time or revolution’.

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, pp. 180–181). Accordingly, we can expect that the time period during which a monarch is in possession of powers is limited, and that pressures for reducing the powers of the monarch will grow as democracy consolidates. Particularly the death of the monarch provides a good opportunity to strip the monarch of his or her powers (see also Huntington, 1968, 180).

The other plausible explanation for powerful monarchs in democracies is size. Corbett et al. (2017) point out that smallness is likely to counteract the gloomy prospects for the monarch outlined in ‘the king’s dilemma’. Smallness, Corbett et al. (2017) argue, is linked to two features that are likely to make monarchies persist in democratic settings, institutional fidelity and personalisation (see also Jugl, 2020, p. 287). By institutional fidelity they refer to the fact that the birth and existence of small states are linked to actions taken by monarchs at crucial moments in history, which means that there is a ‘tendency to consider the monarchy an essential component of the identity of the state’ (Corbett et al., 2017). Personalisation, again, stems from the assumption that relations between the people and the rulers become more intimate in small entities. Based on a study of Bhutan, Liechtenstein, and Tonga, Corbett et al., 2017 (pp. 701–702) conclude that

the significance of personalisation is magnified in small states due to the close proximity between the ruler and the ruled. The result is that monarchs are not remote or distant figures, thus undermining the potential for the regime to become a symbol of oppression. (also Dahl & Tufte, 1973, pp. 66–109)

In other words, both a previous monarchic regime and small size are expected to be conducive for semi-constitutional monarchism. However, based on the arguments laid out above, they differ with regard to how they are linked to the phenomenon in question. Huntington’s (1968) line of reasoning suggests that the previous regime form explains why semi-constitutional monarchies emerge whereas Corbett et al. (2017) link small size to the persistence of semi-constitutional monarchic rule. Indeed, it is difficult to see how small size could explain why powerful monarchs emerge in democratic entities. All semi-constitutional monarchies, since they combine a democratic form of government with a non-powerless hereditary head of state, struggle with a problem of legitimacy. It is therefore unlikely that a democratising country, previously not ruled by a monarch, would deliberately opt for this regime form in a situation where all other alternatives were available. This argument should be valid in small and large states alike.

Measuring monarchic powers

When monarchies form the object of research, focusing exclusively on constitutional provisions is likely to aggrandise the powers of the monarch. In many countries, the process of democratisation was slow, and the monarch was gradually divested of his or her powers. Often, these changes were not reflected in the constitutions of the countries. During the period 1809–1974, the Swedish constitution stipulated that executive powers were conferred 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 Norway, again, 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, p. 172).

When measuring monarchical powers it is therefore advisable to rely on political practice rather than constitutions. In recent years, it has become much easier to make cross-country comparisons on the basis of political practice. The V-dem-dataset (Coppedge et al., 2018) is particularly important for the purpose of the present study as it contains a number of variables which compares the powers of the head of state with those of the head of government based explicitly on political practice.

The powers of the monarchs are measured with reference to nine questions in the V-dem dataset. Five of these refer explicitly to powers in the executive sphere, whereas two refer to legislative powers. I also include a question measuring the influence of the monarch over domestic policy. Finally, I account for the ability of the monarch to dissolve the legislature, which is a power that affects the legislative sphere directly and the executive sphere indirectly (e.g. Neto & Lobo, 2009; Shugart, 2005, pp. 334–335).

Empirical patterns of power sharing

Countries and cases with powerful monarchs

To some extent, all semi-constitutional monarchies operate in a gray area between autocracy and democracy. In essence, the more powers the monarch possesses, the less democratic the country. When identifying semi-constitutional monarchic systems the task of separating democracies from autocracies is therefore crucial. The present study takes as its point of departure the dataset by Boix et al. (2013; 2018), who make use of a dichotomous qualitative scale, classifying more than 200 countries as either democracies or autocracies on a yearly basis for the time period 1800–2015. Since the dataset does not cover the period 2016–2017, I have for these two years classified the countries of the world into the categories democracy or autocracy.¹ Countries classified as democracies by Boix, Miller and Rosato are included in the study. The dataset does not contain extensive information on regime characteristics for the European miniature states Liechtenstein and Monaco. For these countries, classifications have been made by the author for the time periods for which data is lacking (Liechtenstein 1866–1990, Monaco 1862–1993). For the sake of validity, I have complemented the dataset by Boix, Miller and Rosato by making use of V-dem’s Liberal democracy index (D) (v2x_libdem). The index varies between 0.0 and 1.0. I have chosen to also include countries with scores equalling or exceeding 0.4 on the scale.² Regarding the few countries not included in the V-dem dataset, they have been included in the study if they are classified as ‘free’ by Freedom House. The only country for which this assessment has been of relevance is Tonga, which is included in the population for the years 2012–2017.

I then proceed by excluding republics and independent countries ruled by the monarch of another country. The categorisation is based on (Anckar & Fredriksson, 2019 and the V-dem dataset). Such cases include former British colonies in which the British monarch formally acts as head of state but where she is represented by a Governor-General. In these cases, the V-dem database considers the Governor-General as the head of state. Although Governor-Generals are formally appointed by the British monarch, their connection to the British Crown is often very vague. In practice, Governor-Generals are chosen by the parliaments and/or prime ministers of the countries and often the persons chosen to this position are local politicians or dignitaries. Therefore, their statuses resemble more a president in parliamentary systems than a monarch in a hereditary monarchy.

Finally, three monarchies classified as democracies are not included in the V-dem dataset, namely Liechtenstein, Monaco and Tonga. The countries are included in the present study but their classifications along the nine power dimensions are based on other sources.³ Table 1 gives an overview of the population of cases. Altogether, 20 monarchies are classified as democracies during the time period 1800–2017 and the number of yearly observations amounts to 1,243.⁴

Table 1. Democracies with a monarch as head of state 1800–2017.

Belgium 1894–1913, 18–39, 44–2017	Monaco 1962–2017
Bhutan 2009–2017	Nepal 1991–2001
Denmark 1901–2017	Netherlands 1888–1939, 45–2017
Greece 1864–1914, 1935, 1946–1966	Norway 1905–39, 45–2017
Italy 1919–1921	Spain 1977–2017
Japan 1952–2017	Sweden 1911–2017
Laos 1954–1958	Thailand 1975, 1983–90, 1992–2005, 2011–13
Lesotho 2002–2017	Tonga 2012–2017
Liechtenstein 1921–2017	United Kingdom 1885–2017
Luxembourg 1890–1913, 18–39, 44–2017	Yugoslavia 1921–1928

I then proceed by identifying the cases where the monarch has been powerful on the nine power dimensions of the V-dem dataset. Results are presented in Table 2 and they tell us that the number of cases and countries where a monarch possesses powers is surprisingly high. This is notably the case regarding the power to appoint the prime minister. However, it is evident that the V-dem experts have interpreted this question loosely, and included many countries where the monarch is not involved in choosing the prime minister, but only makes the final, formal, appointment. Still, with regard to the other prerogatives, the cells are far from empty. The number of cases varies between 72 and 386 and the number of countries between 4 and 13.

Table 2. Monarchical powers in democracies. V-dem database.

Monarchic powers	Cases	Sum
1.Monarch appoints cabinet^a	Monaco 1962–2017, Nepal 2000–01, Sweden 1911–16, Yugoslavia 1921–28	Cases: 72 Countries: 4
2.Relative power of monarch equals or is stronger than power of PM^b	Belgium 1918, 1959, Liechtenstein 1921–2017, Luxembourg 1944, Monaco 1962–2017, Netherlands 1945, Sweden 1911–16, Yugoslavia 1921–28	Cases:171 Countries: 7
3.Monarch dismisses ministers^c	Bhutan 2016, Greece 1864–1914, Laos 1954–58, Liechtenstein 1921–2017, Luxembourg 1944, Monaco 1962–2017, Nepal 1991–2001, Netherlands 1945, Sweden 1911–17, Thailand 1975, 1983–90, 92–2005, 2011–13, Yugoslavia 1921–28	Cases: 264 Countries: 11
4. Monarch can remove Prime Minister^d	Bhutan 2014–17, Greece 1864–1907, 1911–14, 1955–66, Laos 1954–58, Lesotho 2013–16, Liechtenstein 1921–2017, Luxembourg 1900–39, 44–2017, Monaco 1962–2017, Nepal 1991–2001, Spain 1977, Thailand 1975, 1983–90, 1992–2005, 2011–13, Yugoslavia 1921–28	Cases: 386 Countries: 11
5. Prime Minister appointed by monarch^e	Belgium 1894–1913, 1918–39, 1946–2017, Denmark 1901–42, 1945–2017, Greece 1864–99, 1935, 1946–66, Italy 1919–21, Japan 1952–2017, Luxembourg 1900–39, 1944–2017, Monaco 1962–2017, Nepal 1991–2001, Netherlands 1888–1939, 1945, Norway 1905–39, 1945–2017, Spain 1977–2017, Sweden 1911–75, United Kingdom 1885–2017, Yugoslavia 1921–28	Cases: 945 Countries: 14
6. Monarch in control over domestic policy^f	Bhutan 2009–17, Greece 1864–1914, 1935, 1950–66, Lesotho 2002–16, Liechtenstein 1921–2017, Luxembourg 1900–39, Monaco 1962–2017, Spain 1977, Thailand 1975, 1983–90, 92–2005, 2011–13	Cases: 313 Countries: 8
7. Monarch can dissolve legislature^g	Bhutan 2013–16, Greece 1874, 1946–66, Italy 1919–21, Liechtenstein 1921–2017, Luxembourg 1944, Monaco 1962–2017, Sweden 1911–16, Thailand 1975, 1983–90, 92–2005, 2011–13, Tonga 2012–2017, Yugoslavia 1921–28	Cases: 229 Countries: 10
8. Monarch has veto powers^h	Bhutan 2009–16, Greece 1864–1910, Italy 1919–21, Laos 1954–58, Liechtenstein 1921–2017, Luxembourg 1900–39, 44–2008, Monaco 1962–2017, Nepal 1991–2001, Netherlands 1945, Spain 1977, Thailand 1975, 1983–1990, 92–2005, 2011–13, Tonga 2012–2017, Yugoslavia 1921–1928	Cases: 374 Countries: 13
9. Monarch can introduce legislation^i	Bhutan 2009–17, Greece 1864–1914, 1935, 1946–66, Italy 1919–21, Laos 1954–58, Liechtenstein 1921–2017, Monaco 1962–2017, Nepal 2000–01, Norway 1905–08, Sweden 1911–16, Yugoslavia 1921–28	Cases: 263 Countries: 10

^a HOS 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?’ Yes= Responses 3 or 4.

^b Relative power of the HOS (D) (v2ex_hosw). ‘Does the head of state (HOS) have more relative power than the head of government (HOG) over the appointment and dismissal of cabinet ministers?’ Strong monarch = 0.5 or 1.

^c HOS 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?’ Yes = 2 or 3.

^d HOG appointment in practice (v2expathhg). ‘Is the head of government (HOG) appointed by the head of state (HOS)? If several bodies were involved in the appointment process, select the one that exerted the most critical impact on the decision’ (Yes = 6, head of state).

^e HOG 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?’ (Yes = 5, head of state).

^f 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?’ (Yes = 5, head of state).

^g HOS 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?’ (Yes = responses 2 or 3).

^h HOS ‘veto power in practice’ (C) (v2exdfvths, *_osp, *_ord) (yes = responses 2, 3, 4).

ⁱ HOS ‘proposes legislation in practice’ (C) (v2exdfpphs, *_osp, *_ord) (yes = responses 0, 1).

Combinations of monarchical powers

The next step is to assess to what extent power dimensions are cumulated in the units of analysis. To begin with, I exclude the power to appoint the prime minister from further analysis for the reason outlined above. Also, combining all the rest of the dimensions makes little sense, since many of the questions from the V-dem dataset overlap, especially with reference to the powers in the executive sphere. For the sake of parsimony, I restrict the analysis to four power dimensions. Every case listed under any of questions 1–4 is categorised as meeting the criterion of monarchic executive powers (EP), and every case listed under questions 8 or 9 is considered to fulfil the criterion of monarchic legislative powers (LP). In addition, I include powers over domestic policy (DPP), referring to question 6 and dissolution powers (DP), referring to question 7 in the analysis. This leaves us with 16 possible power combinations, presented in Table 3.

Table 3. Combinations of monarchical powers in democracies 1800–2017. V-dem dataset.

EP*LP*DPP*DP	Bhutan 2014–16, Greece 1874, 1955–1966, Liechtenstein 1921-2017, Monaco 1962–2017, Thailand 1975, 1983–90, 1992–2005, 2011–13.	Cases: 195 Countries: 5
EP*LP* DPP	Bhutan 2014–17, Greece 1864–1914, 55–66, Liechtenstein 1921–2017, Luxembourg 1900–39, Monaco 1962–2017, Spain 1977, Thailand 1975, 83–90, 92–2005, 2011–13.	Cases: 288 Countries: 7
EP* LP* DP	Bhutan 2014–16, Greece 1874, 1955–1966, Liechtenstein 1921–2017, Luxembourg 1944, Monaco 1962–2017, Sweden 1911–16, Thailand 1975, 83–90, 92–2005, 2011–13, Yugoslavia 1921–28.	Cases: 210 Countries: 8
EP*DPP* DP	Bhutan 2014–16, Greece, 1874, 1955–66, Liechtenstein 1921–2017, Monaco 1962–2017, Thailand 1975, 1983–90, 1992–2005, 11–13.	Cases:195 Countries: 5
LP*DPP* DP	Bhutan 2013–16, Greece 1874, 1955–1966, Liechtenstein 1921–2017, Monaco 1962–2017, Thailand 1975, 83–90, 92–2005, 2011–13.	Cases:196 Countries: 5
EP*LP	Bhutan 2014–17, Greece 1864–1914, 55–66, Laos 1954–58, Liechtenstein 1921–2017, Luxembourg 1900–39, 44–2008, Monaco 1962–2017, Nepal 1991–2001, Netherlands 1945, Spain 1977, Sweden 1911–16, Thailand 1975, 83–90, 92–2005, 2011–13, Yugoslavia 1921–1928.	Cases:387 Countries: 12
EP*DPP	Bhutan 2014–17, Greece 1864–1914, 1955–1966, Lesotho 2013–2016, Liechtenstein 1921–2017, Luxembourg 1900–1939, Monaco 1962–2017, Spain 1977, Thailand 1975, 83–90, 92–2005, 2011–13.	Cases:291 Countries: 8
EP*DP	Bhutan 2014–16, Greece 1874, 1955–66, Liechtenstein 1921–2017, Luxembourg 1944, Monaco 1962–2017, Sweden 1911–16, Thailand 1975, 83–90, 92–2005, 2011–13, Yugoslavia 1921–28.	Cases:210 Countries: 8
LP*DPP	Bhutan 2009–17, Greece 1864–1914, 1935, 50–66, Liechtenstein 1921–2017, Luxembourg 1900–39, Monaco 1962–2017, Spain 1977, Thailand 1975, 83–90, 92–2005, 2011–13.	Cases:306 Countries: 7
LP*DP	Bhutan 2013–16, Greece 1874, 1946–66, Italy 1919–21, Liechtenstein 1921–2017, Luxembourg 1944, Monaco 1962–2017, Sweden 1911–16, Thailand 1975, 83–90, 92–2005, 2011–13, Tonga 2012–17, Yugoslavia 1921–28.	Cases: 229 Countries: 10
DPP*DP	Bhutan 2013–16, Greece 1874, 1950–66, Liechtenstein 1921–2017, Monaco 1962–2017, Thailand 1975, 83–90, 92–2005, 2011–13.	Cases: 201 Countries: 5
EP	Belgium 1918, 1959, Bhutan 2014–17, Greece 1864–1914, 55–66, Laos 1954–58, Lesotho 2013–16, Liechtenstein 1921–2017, Luxembourg 1900–39, 44–2017, Monaco 1962–2017, Nepal 1991–2001, Netherlands 1945, Spain 1977, Sweden 1911–17, Thailand 1975, 83–90, 92–2005, 2011–13, Yugoslavia 1921–28.	Cases: 399 Countries: 14
LP	Bhutan 2009–17, Greece 1864–1914, 1935, 46–66, Italy 1919–21, Laos 1954–58, Liechtenstein 1921–2017, Luxembourg 1900–39, 44–2008, Monaco 1962–2017, Nepal 1991–2001, Netherlands 1945, Norway 1905–08, Spain 1977, Sweden 1911–16, Tonga 2012–17, Thailand 1975, 83–90, 92–2005, 2011–13, Yugoslavia 1921–28.	Cases: 411 Countries: 15
DPP	Bhutan 2009–17, Greece 1864–1914, 1935, 50–66, Liechtenstein 1921–2017, Lesotho 2002–16, Luxembourg 1900–39, Monaco 1962–2017, Spain 1977, Thailand 1975, 83–90, 92–2005, 2011–13	Cases: 313 Countries: 8
DP	Bhutan 2013–16, Greece 1874, 1946–66, Italy 1919–21, Liechtenstein 1921–2017, Luxembourg 1944, Monaco 1962–2017, Sweden 1911–16, Thailand 1975, 83–90, 92–2005, 2011–13, Tonga 2012–17, Yugoslavia 1921–28.	Cases: 229 Countries: 10
None	Belgium 1900–13, 19–1939, 44–58, 60–2017, Denmark 1901–2017, Japan 1952–2017, Lesotho 2017, Netherlands 1888–1939, 1946–2017, Norway 1909–39, 1945–2017, Spain 1978–2017, Sweden 1918–2017, United Kingdom 1885–2017	Cases: 793 Countries: 9
Missing data	Belgium 1894–99, Luxembourg 1890–99	Cases: 16 Countries: 2

Note: EP = Executive powers; LP = Legislative powers; DPP = Domestic policy powers; DP = Dissolution powers.

The results show that there are five countries where the monarch has been powerful on all four dimensions: Bhutan, Greece, Liechtenstein, Monaco and Thailand. Bhutan had been a hereditary monarchy since 1907 when the country enacted a new constitution in 2008. The constitution transformed Bhutan into a constitutional monarchy as the monarch gave up a significant part of his powers. It has generally been pointed out that the case of Bhutan is unique in the sense that democratic reforms were voluntarily initiated by the king and not reluctantly, as a result of popular protests or demands. Indeed, the majority of the Bhutanese population was, in fact, against the reforms (e.g. Quintino, 2018; Sinpeng, 2007). This depiction is somewhat qualified by authors who have pointed out that there was indeed some pressure for democratic reforms particularly from external actors (e.g. Iyer, 2019; Muni, 2014). However, as Corbett et al. (2017) have noted, the heart of the matter is that ‘the king simultaneously devolved authority and cemented the monarchy’s place in Bhutanese socio-political life’ [whereby] he ensured a peaceful transition to democracy and sidestepped Huntington’s ‘King’s Dilemma’, at least for a time’ (Corbett et al., 2017) see also Sinpeng (2007, p. 39).

In Greece, regime developments during the last 150 years have been a real roller-coaster ride. The country surpassed the threshold of democracy in 1864, and is the first example of executive power sharing. Greece’s period as a semi-constitutional monarchic system ended when disagreements between Prime Minister Venizelos and King Constantine regarding whether Greece should join forces with the allies or remain neutral during World War I led to the breakdown of the democratic system in 1915. Greece returned to democracy in 1926 under a republican constitution. In 1935, military strongman Georgios Kondylis forced Prime Minister Tsaldaris to resign and reinstalled the semi-constitutional monarchic system. However, already in 1936, Greece returned to authoritarian rule under Ioannix Metaxas and the country remained autocratically ruled until 1946, when parliamentary elections were held and the semi-constitutional monarchic system was effectively restored. In 1967, the military coup ended the monarchy and when Greece returned to democracy in 1974 it adopted a republican parliamentary system in which the president effectively had very few powers.

The monarchs of Liechtenstein and Monaco still possess extensive powers. Veenendaal (2013, p. 58) notes that ‘[i]n both countries, executive and judicial power is traditionally located in the hands of the Prince, who delegates this power to selfappointed government ministers and judges’. One important difference between Liechtenstein and Monaco is that the principle of parliamentarism is not recognised in the Monegasque constitution (Grinda, 2007, p. 76, 88).

In Liechtenstein the Prince Regnant appoints the government, which must enjoy the confidence of both the legislature and the Prince Regnant. The hereditary monarch also has the power to dissolve parliament and veto powers. In contrast to many other monarchies, where the monarchs are powerful constitutionally but not in reality, the monarchs of Liechtenstein have in fact used, or threatened to use, their powers, and this is especially the case with the contemporary monarch, Hans Adam II (Beattie, 2004, pp. 174–225; Corbett et al., 2017, pp. 699–700).

In Monaco, the powers of the Prince are even greater. According to the constitution from 1962, the Prince is the dominant (in fact, the only) actor in the exercise of executive powers and has very far reaching powers in the legislative sphere; he has exclusive right to initiate laws, absolute veto powers and can dissolve the legislature at will. Based on purely constitutional provisions it can, indeed, be questioned whether Monaco actually qualifies as democracy (e.g. Veenendaal, 2013, pp. 57–58). Yet, Boix et al. classifies the country as a democracy during the whole period it is included in the dataset (i.e. from 1994) and Freedom House has classified the country as ‘free’ since 1993.

In Monaco, it is rather difficult to compare constitutional provisions with political practice, as there have been few conflicts between the Prince and the parliament. The parliament has always been dominated by one party (although not always the same), which has been loyal to the monarch. Nevertheless, it is evident that in comparison with other monarchs operating within a democratic framework, the Prince of Monaco is an extremely powerful actor in terms of both constitution and practice (see Chagnollaud de Sabouret, 2015; D’Onario, 2014).

In Thailand, the position of the monarch was very strong until the year 1932, when a coup was launched by a relatively small group consisting of both military personnel and civilians. In the new constitution, adopted in the same year, the powers of the monarch were reduced substantially. The coup did not transform Thailand into a democracy, however. For the next four decades, Thailand was ruled by the military, a period that ended when Thailand surpassed the threshold for democracy in 1974. Ever since, the country has hovered between democracy and autocracy.

It is somewhat difficult to assess the influence of the monarch in Thai politics. In a widely cited article, McCargo (2005) uses the term ‘network monarchy’ in order to describe Thailand’s mode of governance. In essence, this system makes it possible for the monarch to exercise his powers behind the scenes, for instance ‘by working through proxies such as privy councillors and trusted military figures’ (McCargo, 2005, p. 501). Although this ‘network monarchy’ has been challenged, in particular by prime minister Thaksin Shinawatra, who was ousted of power in a military coup, in 2006, it is evident that the concept is still highly relevant for describing the Thai form of governance. Kanchoochat and Hewison (2016, p. 377), for instance, maintain that ‘the groups … [making up] the network monarchy were energised by the rise of Thaksin and became united in opposition to pro-Thaksin governments’. The death of King Bhumibol in 2016 has brought the question of monarchic powers to the fore in Thailand. At present, there are no signs that the new monarch, King Varjiralongkorn, would accept a mere ceremonial role in Thai politics. Quite to the contrary, there are many indications that the new king is eager to increase his powers (Mérieau, 2017).

Moving on, the results showed that In Luxembourg, Sweden, Spain and Yugoslavia the monarch has been in possession of powers in three out of four power dimensions. In the data set by Boix et al. (2013), Luxembourg is classified as a democratic country since 1890. Although the current constitution still formally grants the Grand Duke a leading role in the executive sphere and the power to appoint and dismiss members of the government as well as the power to dissolve the legislature at will, the monarch of Luxembourg possesses significantly less powers in practise.

Sweden, again, passed the threshold of democracy in 1911, when universal male suffrage was introduced. For a few years, the king and the prime minister shared executive powers, but it was highly unclear which of the actors that was the most powerful one. The power struggle culminated in 1914, when King Gustav V publicly challenged Prime Minister Karl Staaff. However, with the election of 1917, power shifted from the king to the prime minister when the King accepted to appoint a government which enjoyed the support of a parliamentary majority.

Like Sweden, Spain fits into the category where a monarch holds powers for a short transitional phase as a country democratises. Before Franco died in 1975, he had appointed the then Prince Juan Carlos as his successor. King Juan Carlos, however, was determined to liberalise Spain. After a short power struggle with holdovers from the Franco regime, democratic elections were held in 1977, and during this year, 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.

Yugoslavia, or as it was called at the time, The Kingdom of Serbs, Croats and Slovenes, was created in 1918, and surpassed the threshold of democracy in 1921. The constitution adopted the same year gave King Alexander I extensive powers. Unlike in Sweden and Spain, where the monarch retained some powers during a transitional phase as democracy consolidated, the Yugoslavian monarch gradually increased his powers, and in 1929, he abolished the constitution and concentrated powers into his own hands, thus returning Yugoslavia to the category of autocratic systems.

Among the remaining cases, the monarch has possessed powers in only one or two spheres. The Italian experience resembles the Yugoslavian one in the sense that the monarch retained some powers as the country transited from autocracy to democracy, after which autocracy was restored. Italy turned to democratic rule in 1919. The old constitution, the Statuto Albertino, was still in force and gave the monarch far reaching powers. It is notable that the constitution did not even contain a provision that the government must enjoy the confidence of parliament, although the principle was accepted implicitly (Caciagli, 2010, pp. 1027–1028). In 1922, Benito Mussolini became prime minister and the short democratic period ended.

Muck like Italy and Yugoslavia, neither Laos nor Nepal conforms to a model where executive power is gradually transferred from the monarch to a government responsible to parliament, after which democracy becomes consolidated. In Nepal, the democratic constitution of 1990 was adopted reluctantly by King Birendra, after it had become apparent that he would otherwise have faced a rebellion (Nepal et al., 2011, p. 887). Although the constitution was parliamentary and the prime minister was the dominant political actor, the king continued to exercise significant influence in the executive field. In 2001, King Birendra was killed by a family member and King Gyanendra assumed the throne. The following year the new king postponed the scheduled elections indefinitely and concentrated executive powers into his own hands, whereby Nepal returned to authoritarian rule.

Laos gained its independence in 1953 but the semi-constitutional monarchic constitution had been adopted several years earlier, namely in 1947, in close cooperation with French officials. The 1947 constitution stipulated that Laos was not to become fully independent but to remain within the French Union. It provided for a democratic form of government albeit with a hereditary monarch with powers to chair cabinet meetings and considerable veto powers in the legislative sphere (Digithèque dematériaux juridiques et politiques). The strong position of the hereditary monarch was attributed to the fact that the royal family of Laos had been extremely loyal to the French. Thus, by vesting powers into the hands of a loyal monarch the French most likely wanted to ensure that its influence in the country continued. The democratic era of Laos ended in 1959, after the military forced Prime Minister Sananikone to resign. However, the monarch continued to be an influential (although not dominant) actor in Laotian politics until the end of the monarchy in 1975, when the communists came to power.

As shown by Corbett et al. (2017), constitutional developments in Tonga constitute a very good example of the King’s dilemma. As pressures for democratic reforms started in the absolute monarchy, ‘King Tupou IV and his government responded by using all the strategies that Huntington outlines, including attempts to co-opt progressive elites, repression, prosecution, and intimidation’ (Corbett et al., 2017, p. 695). Consequently, democratic reforms could not take place until Tupou IV died in 2006 and was succeeded by Tupou V. A new constitution, which restricted the powers of the monarch, was adopted in 2010. Nevertheless, the monarch continues to have extensive powers, especially in the legislative sphere. In 2012, Tupou V died, and was succeeded by his brother Tupou VI. During his reign, it appears as the monarch has strengthened his powers, a fact which became apparent at the latest in August 2017, when the King dissolved the legislature, thereby indirectly dismissing the prime minister.

Lesotho, again, meets the criteria of democracy since 2002. The powers of the monarch have been constitutionally weak ever since the country became independent in 1966. However, the relationship between the king and the prime ministers has been conflict-ridden in the past. The king is deemed to have had considerable powers in domestic policy until 2016. Moreover, the V-dem country experts consider him to have had powers to remove the prime minister during 2013–2016, a period during which the main parties had difficulties in securing a majority in parliament.

Finally, the monarchs of Belgium and Netherlands have possessed powers for very short periods of time. These cases, however, refer to exceptional periods in the history of the countries. In the V-dem dataset, the Belgian monarch is considered to have had a strong position with regard to government formation during the year 1918, the same year Belgium was liberated from German occupation. In contrast to the King, the Belgian government had been in exile during the war, and new elections did not take place until 1919. The V-dem country experts also regard the Belgian king to have been influential in the executive sphere in the year 1959, most probably a reflection of King Baudouin’s active role in the independence process of Belgian Congo. In the Netherlands, the monarch is considered to have been powerful in 1945. After the end of the German occupation, the country was ruled by a caretaker government, appointed by the monarch and functional for a short period, when the parliament had not yet become functional after the occupation.

Explanations of executive power-sharing

We can then give a final assessment of how our two plausible explanations of monarchic power fare. In Table 4, the size dimension is combined with a dimension accounting for whether the powers of the monarch are related to a transitory stage from autocracy to democracy. All countries where the monarch has been in position of powers in any of the four dimensions listed in Table 3 are included. Thresholds within the size category are, to some extent, always arbitrary. However, based on our theoretical proposition we would expect that the crucial differences are the ones that exist between very small entities and larger ones. One natural dividing line emerges impromptu, as there is a cluster consisting of the five miniature states Bhutan, Liechtenstein, Luxembourg, Monaco, and Tonga. The smallest category, then, is made up of states with a population of less than 1 million. A second category is composed of somewhat larger countries, with a population ranging from 1 to 3 millions, whereas all other countries are considered large. Regarding the other dimension of interest, I apply six categories. The first one is made up of countries where the monarch has ‘inherited’ at least some of his or her powers when the country transited from autocracy to democracy, after which democracy has become consolidated. The second category consists of similar cases in which democracy did not consolidate, and the country returned to autocracy. The third and fourth category refer to situations where the semi-constitutional monarchic form of government has emerged from an autocratic regime without a monarch as head of state, whereas the fifth and sixth categories describe situations where the semi-constitutional monarchic system has developed from another democratic form of government, and subsequently either democratised (the second last column) or not (the last column).

Table 4. Patterns of emergence and consolidation of semi-constitutional monarchies 1800–2017.

Population	Previous regime
	Autocracy with monarch as head of state		Other autocracy		Democracy
	Outcome		Outcome		Outcome
	Democratic consolidation	Democratic breakdown	Democratic consolidation	Democratic breakdown	Democratic consolidation	Democratic breakdown
< 1 million	Bhutan 2009–17, Liechtenstein 1921–2017, Luxembourg 1900–2017^a, Monaco 1962–2017, Tonga 2012–17
1–3 millions	Lesotho 2002–16, Norway 1905–08	Laos 1954–58
> 3 millions	Spain 1977, Sweden 1911–17	Greece 1864–1914, Italy 1919–21, Nepal 1991–2001, Thailand 1975, 83–90, 92–2005, 11–13, Yugoslavia 1921–28		Greece 1944–66	Belgium 1918^a, 1959, Netherlands 1945^a	Greece 1935

^aDisregarding periods of occupation during WW1 and WW2.

On a general level, there is very strong support for the assumptions laid out in the theoretical part of the study. In almost all cases where a monarch has held powers in a democracy, the powers of the monarch are directly ‘inherited' from or related to the pre-democratic era. However, semi-constitutional monarchies do not always exist as a transitional phase during which democracy is introduced and subsequently consolidated. In the population Bhutan, Liechtenstein, Luxembourg, Monaco, Norway, Lesotho, Spain, Sweden, and Tonga fully conform to such a pattern. However, In Greece (1864–1914), Italy (1919–1921), Laos (1954–1958), Nepal (1991–2001), Thailand and Yugoslavia (1921–1928) the semi-constitutional monarchic system coincided with a transition to democracy, but in all these countries the democratic form of government subsequently broke down.

In general, the evidence supports Huntington’s argument in the sense that powerful monarchs have not been long-lived in democracies. However, three countries, all of which have powerful monarchs at the time of writing, contradict this rule. Liechtenstein, Luxembourg, and Monaco have had powerful monarchs operating within a democratic context for many decades. Among the other cases, only Greece between 1864 and 1914 qualify as a long-term semi-constitutional monarchy. Thus, much in line with Corbett et al. (2017), we reach the conclusion that small size appears to be very important for explaining why powerful monarchs persist in democracies. Such a conclusion is corroborated by the cases of Bhutan and Tonga, both of which are small and where the monarch is in possession of significant powers. However, since none of the two countries has been a democracy for more than decade, it is difficult to predict future developments. It is far from self-evident that the countries will have a democratic form of government with a powerful monarch as head of state after two or three decades.

If we disregard the exceptional cases of Belgium and the Netherlands dealt with above, Greece is the only country where the powers of the monarch have not been directly inherited from the pre-democratic period. In Greece, the monarchy was restored in 1935, after a referendum installed (and probably heavily rigged) by Prime minister and General Georgios Kondylis. This restoration of the monarchy is best explained from a political actor perspective. The main reason for Kondylis’ support of the return of the monarchy was apparently strategic; his ambition was to follow the example of Benito Mussolini, and merely retain the monarchy as a means of legitimising his actions.

During the Second World War, the king was in exile. After the liberation of Greece, there was strong opposition towards the monarchy. The question whether Greece should be a republic or a monarchy was settled in a referendum in 1946, where the monarchists won by more than two thirds of the vote. Greece was then ruled as a semi-constitutional monarchy until the military coup in 1967. The military junta abolished the monarchy in 1973, in an attempt to consolidate its position. After the fall of the military regime, voters approved the introduction of a republican form of government by a clear majority.

The introduction of a republic form of government in 1974 constitutes an example of a case where the monarch gradually loses his legitimacy among the population by stretching his constitutional prerogatives to its limit. King Constantine II was highly controversial. He came to power in 1964, and almost immediately clashed with prime minister Papandreou over the control of the military. In the subsequent years, King Constantine appointed a number of governments, all of which lasted for short periods of time. This turbulent period paved the way for the military takeover in 1967. In this respect, the Greek case provides an excellent illustration of how difficult it is for the monarch to coexist with party government, very much in line with Huntington’s (1968, pp. 180–181) prediction.

Whereas Greece has switched between republic and monarchic forms of governments, Thailand has remained a monarchy, but oscillated between democracy and military rule. In Thailand, the semi-constitutional 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. Finally, there is a religious connection. Theravada Buddhism is the dominant religion in Thailand. In this tradition, the king is believed to be ‘meritorious’, meaning that he is in possession of ‘vast reservoirs of merit accumulated in past lives, which can be translated into the improvement of this-worldly conditions of those who are linked with them’ (Keyes, 1977, p. 288).

Conclusion

The first task of the study was to identify all democratic regimes in which the monarch has been in possession of executive or legislative powers during the time period 1800–2017. Based on the V-dem dataset I identified seventeen countries that qualified as semi-constitutional monarchies during the last two centuries. I then proceeded by testing the assumption that semi-constitutional monarchies would emerge primarily in countries which transit from autocratic monarchies to democracies and that small size was conducive for the survival of the regime type in question.

In general, empirical findings strongly supported these propositions. Semi-constitutional monarchic regimes emerge in former autocratic monarchies as they democratise and rarely persist for long periods. The results also showed, that while semi-constitutional monarchic forms of government tend to emerge in rather similar settings and under similar circumstances (i.e. when countries move from autocracy to democracy for the first time), the regimes can eventually be substituted with authoritarian or democratic forms of government. In those rare cases, where semi-constitutional monarchic regimes become long-lived, the size of the political units plays an important role, suggesting that small size appears to be crucial for the legitimacy of strong monarchs in democratic settings. It is particularly noteworthy that all three long-lasting semi-constitutional monarchies are characterised by their extreme smallness.

The present work has pinpointed the need to put more focus on two largely overlooked research areas in political science. First, whereas there is already a quite extensive literature on executive power sharing between presidents and prime ministers, similar studies regarding the power-sharing arrangements between monarchs and prime ministers are, with very few exceptions, conspicuous by their absence. Yet, the results of the present study show, that monarchs have possessed significant powers in a substantial part of the democratic countries with a monarch as head of state. This finding alone, lays good ground for further research in the field.

It can be readily assumed that the issue of how the powers of the monarch in relation to the prime minister should be measured will be crucial in future research efforts. In the present contribution, I have made use of V-dem’s dataset. Although the dataset is useful for cross-country studies, it certainly has its limitations. Since the coding is based on expert surveys among a large number of country experts it is evident that the coding criteria can vary substantially between the experts (e.g. Skaaning, 2018, pp. 110–111). In addition, the dataset does not account for all possible power prerogatives. For the purpose of the present study, it would, for instance, have been particularly useful to also account for to what extent the monarch is in control of foreign policy.

Second, physical factors have generally played a surprisingly subordinate role in comparative politics. The most obvious example of this shortcoming concerns the link between size and democracy. Theoretically, it can be traced back to Ancient Greece, but so far, very few empirical studies have been conducted where size has been given the primary focus among the explanatory variables. The results of the present study has demonstrated that there is a strong relationship between population size and the stability of semi-constitutional monarchies. However, it is quite plausible that physical determinants in general and size in particular can play an important role in explaining regime choice and regime survival on a more general level and future studies are accordingly advised to fully explore such patterns.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Note on contributor

Carsten Anckar is professor of political science (comparative politics) at Abo Akademi University, Finland. He has published widely in the field of political science. His current research focuses on regime classifications and regime transitions.

Notes

1 The basic strategy has been to compare Freedom House’s scores with Boix, Miller and Rosato’s classifications during the period 2012–2015 and thereafter check if Freedom House’s scores have changed during the years 2016–2017. Values have also been compared with the scores countries have received on the Polity 2 scale (an effective measure of the degree of democracy, ranging from –10 to +10) in the Polity IV-dataset (Marshall et al., 2018).

2 The motivation for choosing 0.4 as a cut-off point is to make sure that no relevant case is left out from the study. Applying a higher threshold, say 0.5. would be too strict in comparison with Boix, Miller and Rosato's threshold for inclusion in the category of democracies, and would, for instance, mean that the following countries, all classified as democracies by Boix et al. in 2015, would fall below the threshold of democracy in 2015: Albania, East Timor, El Salvador, Guyana, India, Mexico, Moldova, Nepal, and Solomon Islands.

3 E.g. D’Onario 2014; Grinda, 2007; Veenendaal, 2013; Beattie, 2004; Chagnollaud de Sabouret, 2015; Quintino, 2014; Matangi Tonga (https://matangitonga.to/).

4 Sometimes it is not easy to draw the line between inheritance and elections. In Lesotho, the college of chiefs determine who will be the next person in line for succession as monarch. However, since this designation shall be ‘in accordance with the customary law of Lesotho (art. 45 of the constitution)’ and the kings have come from the same family during the country’s independence, Lesotho is classified as a monarchy.