2 Separation of powers ≠ presidentialism

The separation of powers between executive and assembly is often explicitly conflated with the presidential form of government, as if one could not be had without the other (Calabresi 2001: 54). This is to some extent the legacy of the so-called old institutionalists, who tended to focus on the cases of Great Britain and the United States and paid little attention to systems that were neither parliamentary nor presidential (Bagehot 1867; Laski 1940; Wilson 1844). Their hunch was “that the basic forms of democratic government follow an essentially dichotomous pattern” (Lijphart 1997b: 128). In parts of the literature, this dichotomous thinking has persisted until today. As a result, presidential government is defended in terms of the separation of powers (Calabresi 2001) and the direct election of the chief executive is proposed as the most appropriate way to democratize separation-of-powers systems like that of the European Union (Sonnicksen 2017).

We ought to stop conflating powers separation with presidentialism and instead evaluate forms of government along two separable analytical dimensions. One is the separation of powers: roughly, the degree to which the origin and survival of the executive is separated from the assembly. The other is executive personalism: roughly, the degree to which the power of the executive is constitutionally focused on a single human being. The main goal in this chapter is to show how the separation of powers can be decoupled from executive personalism—and that semi-parliamentary government is a way to do so.

Table 2.1 summarizes the argument in a simplified manner. In presidential and parliamentary government, the separation of powers and executive personalism are perfectly correlated; semi-parliamentarism decouples them. Executive personalism without the separation of powers could hardly be considered democratic; it would be some form of elective dictatorship.

Table 2.1 is too simple a typology, of course, as there are other hybrid forms of government. Most studies and textbooks in political science work with a trichotomy that distinguishes the two pure types from the most prevalent hybrid: semi-presidentialism (e.g. Cheibub et al. 2014; Samuels 2007). Yet, since the latter ties the separation of powers very closely to executive personalism, too, the common trichotomy also reinforces the conflation of the separation of powers and executive personalism. I argue that we ought to distinguish six basic forms of government and highlight how three of them can be understood as efforts to reap some or all of the benefits of the separation of powers, while limiting or avoiding executive personalism. This chapter pays special attention to these oft-neglected hybrids: elected prime-ministerial government in Israel (from 1992 to 2001), assembly-independent government in Switzerland, and semi-parliamentary government in Australia (and Japan).

I begin by elaborating on the concept of executive personalism. Then I discuss the six basic types of government, which come in three logical pairs: presidentialism and parliamentarism, the Israeli and Swiss hybrids, and semi-presidentialism and semi-parliamentarism. The discussion of the two pure types remains short, as a fuller exploration is reserved for Chapters 5 and 9. The chapter ends by synthesizing the argument in a simple typological framework.

As noted in Chapter 1, executive personalism refers to the extent to which constitutional rules (a) vest executive power in a single human being; (b) who is democratically authorized (more or less) directly by the voters; and (c) who cannot be dismissed for political reasons by any collective and representative entity such as an assembly or a political party.¹ The concept says nothing about how much power the chief executive has, but executive personalism seems to have a causal tendency, under a broad range of background conditions, to strengthen presidential power and undermine and erode formal and informal constraints on executive. While this causal hypothesis is one reason for focusing on executive personalism, it must be distinguished from the concept as such.²

As to formal constraints on the executive, presidentialism in the United States is a good example. The history of the US presidency has been one of the gradual expansion of its power and importance, and this expansion has been fueled by executive personalism (Ginsberg 2016: 38–52; Prakash 2020). Madison’s idea of checks and balances underestimated the unifying power of presidential leadership, as well as the pressure to work around the deadlock created by numerous institutional veto players (Howell and Moe 2016, 2020). As a result, presidents “dismantled the Madisonian system piece by piece, paving the way for our current president-centered system of national administration” (Posner 2016: 42). Much of the power of the US president is thus not kept in check anymore by constitutional constraints, but “by public scrutiny, the media, and the challenge of leading different institutions and groups in an enormous and diverse country” (Posner 2016: 43). Or so one might hope.

As to informal checks and balances, the picture is similar. It is certainly plausible that more strongly organized parties with an independent leadership and an institutionalized bureaucracy are able to “try to curb the excesses of the president in order to protect their own political prospects” (Rhodes-Purdy and Madrid 2020: 321; see also Martínez 2021). Yet, the problem is that weak organizational and programmatic capacities of political parties are, to a large extent, endogenous to executive personalism.

Samuels and Shugart (2010) have shown that parties in presidential and semi-presidential systems often become “presidentialized.” This means that they “delegate considerable discretion to their leaders-as-executives to shape their electoral and governing strategies, and that parties lose their ability to hold their agents to account” (Samuels and Shugart 2010: 37). As a result, “political parties in pure and semi-presidential systems are unlikely, under most conditions, to act as voters’ representational agents as they do in parliamentary systems” (Samuels and Shugart 2010: 247). Chief executives may even be able to reverse the principal–agent relationship and dominate their parties, rather than being responsible to them. This also means that they can often rely on “their” parties to undermine the separation of powers (Levinson and Pildes 2006). Parties may not be strong enough principals to control the president, but they can be cohesive and polarized agents enough to shield him or her from effective checks and balances.

While my arguments draw heavily on the important theory by Samuels and Shugart (2010), I also argue that it needs clarification and extension. Its main claim is that “to the extent that the constitutional structure separates executive and legislative origin and/or survival, parties will tend to be presidentialized” (2010: 37). Yet, the authors only analyze versions of the separation of powers that imply the direct election of either a president or a prime minister. Assembly-independent government in Switzerland and semi-parliamentary government in Australia (and Japan) are excluded from their analysis (Samuels and Shugart 2010: 28), but these are precisely the versions of the separation of powers that limit or prevent executive personalism by avoiding any direct executive elections. It is not the separation of powers that presidentializes parties, but executive personalism. Or so I argue in this chapter.

The two pure forms of government have been distinguished in myriad ways (Lijphart 1992b), but virtually all leading scholars today agree on two definitional criteria (Cheibub et al. 2014; Elgie 2018; Lijphart 1984; Samuels and Shugart 2010; Shugart and Carey 1992). One concerns the origin of the executive: Is the (chief) executive selected by the legislature or elected independently from the legislative assembly, in popular elections? The other dimension concerns the executive’s survival in office: Can the chief executive and cabinet be removed by the assembly in a political no-confidence vote with a simple or absolute majority—or is there only an impeachment procedure of a more judicial and typically supermajoritarian nature (Pérez-Liñán 2020)? These two institutional dimensions affect the degree of branch-based powers separation as well as the degree of executive personalism.

In the two pure types of parliamentary and presidential government, the separation of powers and executive personalism are perfectly aligned (Figure 2.1). Pure presidentialism separates the origin and survival of the chief executive in a way that maximizes executive personalism. Presidents, as the chief executives, are popularly (usually directly) elected by the voters for a fixed term; they cannot be removed by the assembly in a political no-confidence vote. Hence, the president’s authority does not depend on the assembly or any other collective entity at the representative level. Presidentialism achieves the separation of powers by vesting massive executive authority in a single human being. As a result, parties in presidential systems tend to become presidentialized (Samuels and Shugart 2010).

While presidentialism’s executive personalism is, to some extent, a historical overhang from monarchy (Colomer 2013; Nelson 2014; Prakash 2020: Chapter 1; Scheuerman 2005), the power concentration in a single human being has also been justified in democratic terms. One idea is that it increases clarity of responsibility (DiClerico 1987: 304); another is that the direct election of chief executives is inherently “more democratic” than their indirect selection by the assembly (Arato 2000: 321; Calabresi 2001: 67; Lijphart 1992a: 13). However, if these normative ideas of democratic responsibility and popular control are taken seriously, one might expect proponents of presidentialism to be equally strong champions of the possibility of direct recall; that is, the dismissal of presidents in a recall referendum.³ Presidents would then be responsible to their voters on an ongoing basis. The democratic principal would have more control over its directly elected agent between elections. Executive personalism would become more fully democratized. In fact, though, direct recall is often ignored in comparisons of presidentialism and parliamentarism, it is only possible in a few presidential systems, and, at the time of writing, no recall election has ever succeeded in removing a national executive from office (Pérez-Liñán 2020).

Presidential constitutions also try to limit the perils of executive personalism through various means. Standard examples include impeachment procedures and the prohibition of presidential re-election (Baturo and Elgie 2019). As I discuss in Chapter 9, however, these mechanisms are problematic counterweights or tend to weaken the potential benefits of the separation of powers. A much rarer approach—used in Uruguay from 1952 to 1967—is to directly elect a multi-person, collegial executive whose members share power equally and make decisions collectively (Altman 2008, 2020; see also Orentlicher 2013). Its downsides include a potential lack of efficiency and resoluteness (Altman 2020: 322–323).

Pure parliamentarism avoids executive personalism by avoiding any separation of powers between the executive and the legislature. Chief executives are selected by an assembly majority and remain accountable to it on an ongoing basis.⁴ They, together with the entire cabinet, can—for purely political reasons—be dismissed by a simple or absolute assembly majority in a no-confidence vote.⁵ This also implies that chief executives remain agents of their party. If a party wants to remove its prime minister, it is generally able to do so through a formal no-confidence vote in parliament or as a matter of intra-party politics.⁶ Hence, even if politics becomes “personalized” in the sense that candidates’ personal characteristics matter greatly, parties remain in control of their prime-ministerial candidates and, if successful, their prime minister (Samuels and Shugart 2010). Under pure parliamentarism, however, this control comes at the costs of losing the potential benefits of the separation of powers (Chapter 5).

If pure parliamentarism and presidentialism exhausted the available design options, the separation of powers and executive personalism would be two sides of the same coin. In fact, though, a number of hybrids blend elements of the two pure types. I begin with two rare hybrids that combine separated power along one dimension—executive origin or survival—with fused power along the other. I argue that these hybrids are rare because they mix presidential and parliamentary features in ways that create severe tensions. Understanding these tensions will help us to appreciate the more robust hybrids discussed afterwards.

Under this hybrid form of government, chief executives are directly elected, but their survival in office depends on the parliamentary majority—separate origin but fused survival. This system was used in Israel between 1992 and 2001 for two general elections in 1996 and 1999 and a prime-ministerial election in 2001. It was also discussed at various points in Italy, Japan, and the Netherlands. Samuels and Shugart (2010: 28) refer to it as elected prime-ministerial.⁷

From the perspective of democratic legitimacy, this hybrid may be viewed as combining the best of presidentialism and parliamentarism: Voters can choose chief executives directly and get rid of them at least indirectly, via a no-confidence vote in the assembly. In practice, however, this system proved unsuccessful and was quickly abandoned. The lessons of this failed experiment are important for the argument in this book. They show how (a) the separation of powers (here of the executive’s origin) was sought as a solution to the problems that a highly fragmented parliament created for pure parliamentarism; and how (b) the efforts to limit the perils of executive personalism along the survival-dimension ultimately led to a self-defeating constitutional design.

Prior to 1992, the fragmentation of the party system in Israel had led to severe problems of building and maintaining cabinets. Since attempts to make the electoral system less proportional had failed, reformers hoped that the direct election of the prime minister would reduce the leverage and blackmail potential of smaller parties. It “would replace the uncertainties of narrow, fluid, and fickle parliamentary majorities with a decisive and unambiguous choice of the head of the executive power” (Medding 1999: 205). The underlying normative idea—typically associated with presidentialism and the Westminster model of parliamentarism—was that “a government must be always as much as possible a direct expression of popular will” (Ottolenghi 2001: 115).

These hopes were shattered, however, because it proved impossible to get the benefits of separate origin without also accepting the risks of executive personalism. One desired effect of electing the prime minister directly was to give the winner a clear mandate from the people and a far greater zone of independence in the day-to-day running of the government. Yet to realize these benefits, it would have been necessary to also weaken the power that the parliamentary majority would have over the prime minister. With this power intact, the small parties would retain much of their bargaining leverage, and the popular mandate of the prime minister would partly be cancelled. To work as intended, the system would have needed some degree of powers separation along the survival-dimension, too—as in a presidential system.

The protagonists of the reform understood this point. In their original plan, prime ministers would not have needed a parliamentary confidence or investiture vote to install their government, and they could only have been dismissed in a no-confidence vote of a supermajority in parliament: at least 70 of the 120 members of the Knesset. This supermajority requirement would have rendered prime ministers somewhat more independent of the parliamentary majority. They would have been somewhat freer to act “like a president, making decisions without needing to keep a weather eye on the shifting moods and alliances in parliament” (Ottolenghi 2001: 112).

But this is where the risks of executive personalism came into play. While the intention was to make the government more powerful vis-à-vis the fragmented parliament, direct election of the prime minister would have implied a simultaneous power shift to the single person occupying the office of the prime minister. And this led to fears that prime ministers “might turn populist or even authoritarian” (Ottolenghi 2001: 112). As a result, the reform could only gain approval without any strengthening of the separation of powers along the survival-dimension: Elected prime ministers could be dismissed with an absolute majority of 61 votes and they would also need a parliamentary vote to install the government. In the case of a lost no-confidence vote, new prime-ministerial and assembly elections were required.⁸ These changes meant that the independent electoral legitimacy of the executive made no difference to its survival in office.

The reform, thus, could not achieve its goals. It ended up being counterproductive because the separation of the executive’s origin increased the partisan fragmentation of parliament even further. Reformers had hoped for a “coattail effect,” so that the candidacies for prime minister of the two largest parties, Labor and Likud, would also increase these parties’ vote share in the Knesset. Yet the opposite was true: More voters engaged in “ticket splitting” and gave their party vote to one of the smaller parties. This pushed party system fragmentation to new highs and exacerbated all the problems of coalition-building and maintenance that had motivated the reform (Medding 1999; Ottolenghi 2001).

What is more, while the direct election of the prime minister did not achieve its goals, the increased executive personalism nevertheless caused a “presidentialization” of the major political parties (Samuels and Shugart 2010: 181–188). In the electoral arena, Labor and Likud diverted resources away from the Knesset election because they needed other parties’ support to win the prime-ministerial race. Without this win, gaining a large share of Knesset seats would have been of little use. The major parties began to choose candidates on the basis of their appeal to a broad constituency, rather than long intra-party service. These candidates sought to appear to be “above” parties, as they focused their campaigns on undecided centrist voters. They also toned down their campaign rhetoric to appease smaller parties and their supporters. The major parties became more vote-seeking and less focused on seeking ideologically rooted policies.

An analogous transformation happened in the governing arena, as direct elections weakened the influence of the prime minister’s party over the office holder and, thus, deprived the principle of collective responsibility “of its party core” (Medding 1999: 205). Since the legitimacy conferred by direct elections was personal, prime ministers felt they should govern independently of partisan constraints, and their parties could do nothing about this. In a pure parliamentary system, the prime minister’s party can fire the person occupying the office of the prime minister without losing its hold on the office itself. Not so in Israel’s hybrid system, as the party had no guarantee that it would retain the premiership in a new election (Samuels and Shugart 2010: 187).

In sum, elected prime-ministerial government in Israel failed to strike a balance between the potential benefits of the separation of powers and the perils of executive personalism. Fused survival cancelled some of the potential benefits of separated origin, while many of the downsides of executive personalism still materialized.

The federal Swiss form of government can be seen as the opposite of the Israeli hybrid. The cabinet is elected by the assembly, but its survival in office does not depend on this assembly: fused origin but separate survival (Figure 2.2). It is often called directorial government or, following Shugart and Carey (1992), “assembly-independent” government.

From the perspective of democratic legitimacy, this hybrid seems to combine the worst of presidentialism and parliamentarism: Voters can neither elect the executive directly, nor can they remove it from office indirectly, via the assembly. In practice, however, it has proved to be highly resilient, which is partly explained by the strength of direct democracy and the conventions of “concordance” (Konkordanz). Here, I focus on how the Swiss version of assembly-independent government manages to capitalize on the benefits of powers separation while successfully containing the dangers of executive personalism.

Let us start with the latter. The members of the Swiss cabinet—called Federal Council—serve a fixed term and, thus, cannot be sanctioned by their parties or the assembly majority during their time in office. They can only be denied re-election when their term is completed. This implies a greater degree of executive personalism than in a parliamentary system. Yet this personalism is simultaneously contained in various ways, which was an explicit goal of the framers. While they acknowledged certain advantages offered by an office such as that of the President of the United States, the Constitutional Reform Committee (1992[1848]: 173) “could not think of proposing the creation of an office so contrary to the ideas and habits of the Swiss people who might see therein evidence of a monarchical or dictatorial tendency … Our democratic feeling revolts against personal pre-eminence.”

The perils of executive personalism are reduced in three main ways. The first is the fused origin of the executive. No person can become a member of the Federal Council against the wishes of the assembly majority, by directly appealing to voters. The second is the collegial nature of the executive. The assembly does not elect a single president, who then selects, and can fire, the other members of the cabinet. Instead, a joint sitting of the assembly’s two chambers elects each of the seven members of the Federal Council, who serve in the cabinet as equals.⁹ Third, the electoral system for the Federal Council, which provides for the individual and sequential election of the seven members, has a systematic tendency to elect more centrist candidates from all parties, as it allows all members of parliament to influence the relative chances of the candidates of their rival parties (Stojanović 2016). For example, in 1999 the centrist Joseph Deiss, one of several Christian Democratic candidates, was elected with 50.2% of the votes in the sixth voting round, after receiving only 8.2% in the first round (Stojanović 2016: 52–53).¹⁰

While Swiss institutions thus limit the power of any individual cabinet member and counteract monarchical or dictatorial tendencies, the separation of survival is crucial for stabilizing Swiss concordance. It liberates the assembly majority from the task of keeping the cabinet in office, so that different legislative coalitions can be formed on different issues. It is important to understand how this stabilizes Swiss conventions.

The term “concordance” is often understood as a synonym for the so-called Magic Formula in Switzerland: the convention that the four largest parties ought to be represented in the Federal Council, with three parties having two members and one party one member.¹¹ This convention is typically seen as an integral component of so-called “consensus” or “consociational” democracy in Switzerland (Lijphart 1984: 24; Freiburghaus and Vatter 2019). However, its emergence was “not the outcome of consensus but of political struggle between Christian Democrats and Radicals” (Stojanović 2016: 55, emphasis in the original), and this struggle reflected the underlying majoritarian institutions (see, e.g. Marti 2019: 38–41).¹² The Christian Democrats (CVP, then KVP) sought to reduce the power of the Liberals (also called Radicals) in the Federal Council, who had achieved a majority of four seats in 1953. They wanted to become the pivotal (median) force in the Council, being able to form majorities either with the Liberals on their right or the Social Democrats on their left. They therefore made a political deal with the Social Democrats, who demanded two Council seats. This deal needed some fortunate circumstances and was thus fully executed only in 1959. The Magic Formula was resisted by the Liberals, so that the Social Democrats had to win their two seats in head-to-head contests against Liberal candidates (and they also had to drop their preferred candidate for the second seat in the third voting round to get a more moderate candidate elected). As a result of this struggle, Christian Democrats, Liberals, and Social Democrats ended up with two seats and the People’s Party (SVP, then BGB) with one. Only later did this seat allocation rule become known as the Magic Formula.¹³

While the broad Magic Formula coalition further reduces the dangers of executive personalism, it also presents a serious challenge for policymaking. The Swiss government has become highly polarized, as the two largest parties—Social Democrats and the People’s Party—occupy rather extreme positions (Bochsler et al. 2015; Vatter 2016). Such a government coalition would be extremely difficult to form and stabilize under parliamentarism because coalition parties’ support for the cabinet typically requires their status as legislative veto players on all or most issues, usually codified in a coalition contract. Ideological heterogeneity thus tends to lead to deadlock and cabinet instability under parliamentarism (Tsebelis 2002). But in Switzerland there is no coalition contract and parties are not veto players; they can be outvoted on individual pieces of legislation. In particular, the parties on the left and right wing may be excluded from the minimal-winning coalition (Schwarz et al. 2011; Traber 2015). Linder and Mueller (2021: 192) note that “the political centre—Christian-Democrats and the Radicals—is the most important policy shaping actor in the parliamentary arena.” Of course, this is what we would expect under majoritarian decision rules. The separation of survival creates a form of legislative flexibility that contributes to the resilience of Swiss concordance.

The Swiss combination of fused origin and separate survival has created a behavioral–institutional equilibrium that is not only rather unique but also has important downsides. Due to the Magic Formula, elections do not have much of an (immediate) impact on the composition of the cabinet and voters have no clear choice between alternative political directions. This was obvious, for example, after the parliamentary elections in October 2019. In the wake of increased public concerns about climate change, the Greens and Green Liberals were the biggest gainers of the election. Both parties more than doubled their previous vote shares, to 13.2% and 7.8%, respectively. The Greens surpassed the Christian Democrats (11.4%) and were only slightly behind the Liberals (15.1%). In the subsequent re-election of the Federal Council in December, they attacked one of the seats of the Liberals and tried to get their head of the party, Regula Rytz, elected instead. Yet the assembly’s center-right majority rejected the attack and re-elected Liberal foreign minister, Ignazio Cassis. The “Green wave” was thus stopped rather abruptly at the gates of the government. The reduced importance of elections for cabinet composition is also considered one of the main reasons why turnout in Switzerland is very low (Blais 2014; Franklin 2004). In 2019, it was only 45.1%.¹⁴

In sum, while assembly-independent government in Switzerland has been resilient and fairly successful, it is part of a very complex and demanding behavioral–institutional equilibrium. It is not surprising, therefore, that the Swiss hybrid has not been an export success.

How can a mixture of presidential and parliamentary features become more robust and less contradictory? The answer, I submit, is that one branch must be constitutionally separated along both institutional dimensions: origin and survival. The way to achieve this without going back to the pure types is to divide either the executive or the assembly into two democratically authorized parts. In this way, one part of the executive (the president) can be separated along both dimensions or the prime minister and cabinet can be separated along both dimensions from one part of the assembly (the second chamber).¹⁵ The final two hybrids have this structure (Figure 2.3).¹⁶

Under semi-presidentialism voters directly elect a president, who survives in office independently from the legislature, but there is also a prime minister, who—together with the cabinet—is dependent on the political confidence of parliament (Duverger 1980; Elgie 1999, 2011).¹⁷ The executive is thus divided into two parts. One part of this “dual executive,” the president, has both origin and survival separated from the assembly, while the other part, the prime minister, has its survival fused with the assembly.

In sharp contrast to the Israeli and Swiss hybrids, semi-presidentialism “has become the most emulated democratic regime type in the world” (Samuels and Shugart 2010: 40). Currently, the constitutions of more than 50 countries—not all of which are democracies—qualify as semi-presidential (Åberg and Sedelius 2020).

Semi-presidentialism’s prevalence is not rooted in its appreciation by scholars. While many constitutional experts warn against this form of government, it suits the self-interest of politicians involved in constitution-making. It provides “a neat compromise between political forces that want presidentialism, usually because they calculate that their party will win the presidency, and those that want parliamentarism, usually because they believe that they are not strong enough to win the presidency, but stand a chance of entering a coalition government, thereby sharing in executive power” (Elgie 2016: 60).¹⁸

Like presidentialism, semi-presidentialism embraces executive personalism as integral to the separation of powers (Lacerda 2020). The difference to pure presidentialism is that the framers of the US Constitution did not anticipate the rise of mass parties. By contrast, semi-presidentialism was designed as a response to these parties. In Weimar Germany, and later in France, leading proponents of semi-presidentialism mistrusted political parties’ capacity to govern and wanted “plebiscitary” presidential elections that would place the president “above” the parties (Samuels and Shugart 2010: 39–40; Weber 1986).¹⁹

When we analytically distinguish the separation of powers from executive personalism, a specific rationale of the former is difficult to formulate for semi-presidentialism. Robert Elgie’s (2011: 14–15) authoritative literature review mentions only two, although he grants the first merely some “intuitive logic.” The idea is that semi-presidentialism allows for some degree of power-sharing within the executive, especially in the context of a polarized society (Elgie 2011: 14). This logic is hardly convincing, though. If power-sharing is the goal, collegial government under parliamentarism or Swiss-style assembly-independent government appears as the much better option (Lijphart 2012). Semi-presidentialism’s executive personalism is likely to undermine the adequate representation of a societal group by a president. A presidential candidate might have to distance himself from the interests of the group to get elected, and the group has little control over an elected president (Samuels and Shugart 2010).

Elgie’s second rationale for semi-presidentialism compares it to presidentialism. Sartori (1997: 124) argues that “[w]hile pure presidentialism is a stalemate-prone structure, semi-presidentialism proposes a gridlock-avoiding machinery.” What he means is that the latter allows for the oscillation of power between the president and the prime minister, “reinforcing the authority of whoever obtains a majority” (Sartori 1997: 125). This argument is based on the French experience with “cohabitation” in the 1980s and 1990s. This term describes a situation where the president and prime minister are from opposing parties and where the president’s party is not represented in the cabinet. In France, the president—due to his informal partisan influence—held executive power when he enjoyed majority support in the assembly but had to accept the formal authority of the prime minister under cohabitation.

The problem with this argument is that cohabitation might also lead to conflict between president and prime minister. Åberg and Sedelius’s (2020: 1125) review of the literature finds that “intra-executive conflict is more common during instances of cohabitation” and that “[a]s expected, cohabitation can lead to severe tension and undermine general performance, especially when a democracy is young, or when there is no clear-cut constitutional provision setting out the distribution of power among the key actors.” In France, the constitution was changed to reduce the likelihood of cohabition. In 2000, voters approved a referendum reducing the president’s term from seven to five years, and the National Assembly then passed a bill to the effect that presidential elections would precede parliamentary elections. Since this reform, cohabitation has been avoided.

While the rationale of powers separation thus remains rather unclear, semi-presidentialism is designed to increase executive personalism. The extent of this increase depends on the strength of the president. Here, the distinction between the premier-presidential and president-parliamentary subtypes of semi-presidentialism becomes important (Samuels and Shugart 2010: 30; Shugart and Carey 1992). In the former subtype, the prime minister and cabinet are formally accountable exclusively to the assembly majority (as in Figure 2.3). In the latter, the prime minister and cabinet are dually accountable to the president and the assembly majority. In terms of Figure 2.3, these systems also imply a dashed line from the president to the prime minister and cabinet.

Executive personalism is greatest in the president-parliamentary subtype of semi-presidentialism. Presidents are typically so dominant that these systems are often treated as “effectively ‘presidential’” (Chaisty et al. 2018: 26) for many purposes. Presidents’ power can even be greater than that of their counterparts in a pure presidential system because they may have the right to dissolve the assembly under certain conditions. President-parliamentarism is also the subtype that is more clearly associated with authoritarian government and/or poses the greatest dangers for the consolidation of democracy (Åberg and Sedelius 2020; Elgie 2011; Stykow 2019).

In the premier-presidential subtype, too, much depends on the specific constitutional powers of the president (Shugart 2005: 338–340). If the president lacks most—or all—of the relevant powers, premier-presidentialism can be barely distinct from parliamentarism with a directly elected figurehead. Ireland is a case in point. Political practice might not always reflect the letter of the constitution, though. On the one hand, we have already seen in the French example that presidents can be dominant, even within a premier-presidential system (especially if they are the head of their parties and enjoy majority support in the assembly). On the other hand, Austria is a well-known example of a president-parliamentary system that, due to constitutional conventions and a particular party-system environment, has effectively functioned like a pure parliamentary system (Müller 1999).

In sum, semi-presidentialism lacks a convincing powers-separation rationale and is essentially about executive personalism. This personalism may not matter much when semi-presidentialism functions like parliamentarism in a particular country or when cohabitation shifts power from the president to the prime minister for a limited period of time. But then a directly elected president does not give the political system much of an advantage either, at least not in terms of the separation of powers.

As Figure 2.3 suggests, semi-parliamentarism is, in some sense, the mirror image of semi-presidentialism (Ganghof 2018a). While semi-presidentialism divides the executive into two democratically authorized parts, only one of which—the prime minister—depends on assembly confidence for its survival in office, semi-parliamentarism divides the assembly into two equally legitimate parts, in the simplest case two chambers, only one of which possesses the power to dismiss the prime minister in a no-confidence vote.²⁰ While semi-presidentialism shifts the locus of powers separation into the executive, semi-parliamentarism shifts it into the assembly. It separates the fused powers of the executive and the first-chamber majority from the second chamber. Only the latter is a legislature in a narrow sense of the term, while the first chamber majority becomes fused with the executive. The cabinet-supporting first chamber majority relates to the second chamber as the president relates to the entire assembly in a presidential system (compare Figures 2.1 and 2.3). As a result, the branch-based separation of powers is decoupled from executive personalism.

Chapter 3 will elaborate on the definition and origins of semi-parliamentarism; here, it is best to consider an example: the Australian state of Victoria (Stone 2008; Taylor 2006). The Parliament of Victoria has two chambers: the first chamber (Legislative Assembly) and the second chamber (Legislative Council). Crucially, both chambers are directly elected for concurrent terms and do not differ in their democratic legitimacy. The logic of representation also does not differ between the two chambers: both represent voters’ ideological preferences. The idea of territorial representation plays no special role in the second chamber. Yet, while both chambers also have a veto over ordinary legislation, only the first chamber can dismiss the cabinet in a no-confidence vote.²¹ Even though individual cabinet members are drawn from both chambers, the cabinet originates and survives separately from the second chamber. This is why Victoria can be described as a semi-parliamentary system.²²

The rationale of semi-parliamentarism is entirely based on the separation of powers. It can combine central advantages of the Swiss and Israeli hybrids, while completely avoiding executive personalism. As in Israel, it is possible for voters to more or less directly select a candidate for the office of the prime minister. The first chamber is elected in single-member districts under majoritarian (“ranked choice”) rules, which have so far succeeded in creating an almost pure two-party system. The winning side usually gains an absolute majority of seats and can form a government on its own. First-chamber elections can thus become highly “personalized,” in the sense that much attention focuses on the prime-ministerial candidates, but they cannot become “presidentialized” in the sense of Samuels and Shugart (2010). In contrast to Israel, though, the prime minister is not confronted with the need to cobble together a fixed-majority coalition in a fragmented parliament. While the Victorian second chamber is elected under proportional rules and represents various minor parties, it need not vote the cabinet into office and cannot dismiss it in a no-confidence vote. The cabinet is thus free to build second-chamber majorities on an issue-by-issue basis—just as in Switzerland. For example, after the concurrent elections of both chambers in 2018, the Labor Party controlled a large majority in the first chamber (62.5% of all seats) and formed a one-party government but had to govern as a “minority cabinet” in the second chamber (45%). The balance of power in this chamber was held by eight minor parties, only one of which (the Greens) also gained seats in the first chamber. Victorian governments made ample use of the resulting flexibility in coalition-building (Ganghof et al. 2018).

In sum, while semi-presidentialism and semi-parliamentarism are, in some ways, mirror images of one another, their rationales could hardly be more different. Semi-presidentialism is essentially about executive personalism, while semi-parliamentarism is essentially about the separation of powers.

Let us summarize the argument by integrating all six basic forms of government into a simple typological framework (Table 2.2). This framework modifies and extends the typological approach of Lijphart (1984: 70) and was first presented in Ganghof (2014).²³ It remains focused on the two crucial questions of how the executive comes into office (origin) and whether it can be removed from office in a political no-confidence vote (survival), but differs from other approaches in two main ways. First, the typology includes democratic criteria in a symmetrical manner. Many definitions require that, in a presidential or semi-presidential system, the president must be authorized in direct or quasi-direct elections, but they say nothing about the democratic legitimacy of the assembly under any executive format. In Table 2.2, the requirement of direct elections is specified for the executive and the assembly alike.²⁴ This also implies that the typology applies only to democratic systems.²⁵ Second, the typology allows for the possibility that the political executive’s survival in office is only partly dependent on the assembly. One way in which this partial dependency can exist is that only one part of a dual political executive is dependent on the assembly—the prime minister, but not the president under semi-presidentialism.²⁶ The other is that the political executive is dependent on the confidence of only one part of a dual assembly: the first, but not the second chamber under semi-parliamentarism (Table 2.2).²⁷

The two institutional dimensions of this typology are directly connected to the two sources of executive personalism. One source is the direct (or quasi-direct) authorization of a single human being: the president in a presidential or semi-presidential system and the prime minister in an elected prime-ministerial system. All three systems in the upper row of Table 2.2 are thus institutionally personalized, to various degrees, along the origin-dimension. The other source of executive personalism is that the members of the executive are not politically responsible to some collective and representative entity on an ongoing basis. Hence both executive formats in the right column are personalized along the survival-dimension. Presidentialism is personalized along both dimensions. This fact is crucial to much of the critical debate around it (see Chapter 9).²⁸

Two systems avoid executive personalism: parliamentary and semi-parliamentary government. The crucial difference between them is that parliamentary government has to give up on the potential benefits of the branch-based separation of powers, while semi-parliamentarism is still able to reap some or all of them. This is why semi-parliamentary government deserves our attention. One of the main conclusions of Samuels and Shugart (2010: 261) is that if reformers “truly want parliamentarized parties, they should keep or adopt parliamentarism.” My conclusion is that if reformers want parliamentarized parties and a branch-based separation of powers, they should keep or adopt semi-parliamentarism.