Hostname: page-component-7b7567c8c-wxvwp Total loading time: 0 Render date: 2026-11-14T03:17:43.816Z Has data issue: false hasContentIssue false
  • English
  • Français

Review Article: Citizens, Presidents and Assemblies: The Study of Semi-Presidentialism beyond Duverger and Linz

Published online by Cambridge University Press:  26 August 2009

Petra Schleiter and
Edward Morgan-Jones
Rights & Permissions [Opens in a new window]

Abstract

Semi-presidential regimes have attracted increasing attention from scholars and constitutional reformers over the last quarter century. Yet, despite this popularity, there is no consensus on how to understand this constitutional format. Since Duverger defined semi-presidentialism as a ‘new political system model’, and Linz argued that the constitutional format shares many of the ‘perils of presidentialism’, subsequent research has questioned the conceptual status of semi-presidentialism as a distinct regime type, and whether it has any distinct effects on politics. In this article we review the progress of recent work on semi-presidentialism and suggest that the conceptual tools to clarify some of the major debates in the field are now available in the form of principal–agent theoretical work on democratic constitutions.

Information

Type
Review Article
Information
British Journal of Political Science , Volume 39 , Issue 4 , October 2009, pp. 871 - 892
Copyright
Copyright © Cambridge University Press 2009

Semi-presidential regimes have attracted increasing attention from scholars and constitutional reformers over the last quarter century. Like the French Fifth Republic, these regimes combine a directly elected president with a prime minister and cabinet accountable to the assembly.Footnote 1 The number of semi-presidential regimes increased dramatically with the democratic revolutions of the early 1990s: in 1946 only three West European countries were semi-presidential (Iceland, Austria and Finland); by 2002 semi-presidential regimes accounted for 22 per cent of all democracies worldwide.Footnote 2 Geographically this constitutional format originated in Western Europe, but has now spread to Eastern Europe, Africa, Asia and Latin America, incorporating states as diverse as France, Mali, Portugal, Peru, Russia and Taiwan.Footnote 3 Thus, in terms of its increasingly frequent adoption and the interest it arouses, semi-presidentialism is, in Shugart’s words, ‘a regime type whose time has come’.Footnote 4

The seminal contributions of two scholars have in large part framed the research agenda on semi-presidentialism: Duverger’s proposition that semi-presidentialism constitutes a ‘new political system model’,Footnote 5 and Linz’s argument that the perils of presidentialism affect semi-presidentialism too, and make it a type of democracy prone to crisis and breakdown.Footnote 6 However, both contributions have sparked major and unresolved debates. In the first of these debates scholars such as Lijphart and Siaroff have questioned Duverger’s contribution on the grounds that the behavioural and institutional variation displayed by semi-presidential regimes undermines the analytical value of classifying them as a distinct regime type.Footnote 7 The second controversy centres on the effects of popularly elected presidents on politics. Here Shugart and Carey, Cheibub and others have challenged Linz’s asymmetrical focus on the adverse effects of popularly elected presidents and argued that there are a number of ways in which presidents potentially enhance (rather than simply detract from) democratic performance.Footnote 8 The third and most general dispute revolves around the question whether the different democratic regime types such as presidentialism, parliamentarism and semi-presidentialism have any distinct effects at all on politics.Footnote 9 Tsebelis and Eaton make two points contra Linz and others in this respect. On a theoretical level Tsebelis views the distinction between presidential, semi-presidential and parliamentary systems as epiphenomenal to the preferences and number of veto players, which he argues determine the performance of a regime by affecting its ability to achieve policy change.Footnote 10 On an empirical level, Eaton argues that analysts who study regime type effects on policy have yet to demonstrate in controlled empirical work that such effects exist.

In this article we review the impressive progress that political scientists have made with respect to semi-presidentialism over the last decade,Footnote 11 and suggest that the conceptual tools to clarify these three major debates in the field are becoming available through recent principal–agent theoretical work on democratic constitutions.Footnote 12 Principal–agent theory focuses on the delegation of authority that lies at the heart of representative democracies: voters – the ultimate democratic principal – delegate to elected politicians, such as presidents and assembly members (voters’ agents), who act in turn as principals in their own right vis-à-vis the government (their agent).Footnote 13 We argue in his article that the analytical strengths of the approach in relation to the debates outlined above are threefold. First, a principal–agent framework provides a theory of differences in the nature of representative democracies against which the contrasting views in the controversy about semi-presidentialism as a regime type can be assessed. Secondly, it offers the conceptual tools to analyse the advantages and drawbacks of different delegation regimes for citizens as the ultimate democratic principals in ensuring that their agents act on their behalf, which help to clarify the debate about the impact of popularly elected presidents on the politics of semi-presidential regimes. Thirdly, a principal–agent approach speaks to the general debate about the effects of semi-presidentialism, parliamentarism and presidentialism on politics, by highlighting aspects of regime performance that go beyond Tsebelis’s elegant but narrow focus on the capacity for policy change, and which are powerfully conditioned by the democratic regime type.

The sections that follow review the literature on semi-presidentialism. We begin by turning to the debate about the conceptual status of semi-presidentialism, before discussing the effects of the regime type on the stability and performance of these democracies.

Conceptualization

The Debate

While Duverger’s seminal article defined semi-presidentialism as a new regime type, it did not ground this claim in a theory of how representative democracies differ in nature. As a result, the debate about the conceptual status of semi-presidentialism has become the longest running in the field. It revolves around two questions: first, should semi-presidential political systems be defined with reference to their institutions or behaviour? And secondly, does their heterogeneity (behavioural and institutional) make semi-presidential regimes an empty set or residual category?

The first question is raised by Duverger’s original definition of semi-presidentialism as a political system in which: ‘(1) the president … is elected by universal suffrage; (2) … possesses quite considerable power; (3) [and] … has opposite him … a prime minister and ministers who possess executive and governmental power and can stay in office only if the parliament does not show its opposition to them’.Footnote 14 This definition combines institutional and behavioural (quite considerable presidential power) criteria. As a result, it gives rise to tautological arguments when used, for example, in studies of the effects of institutions on the behaviour of these regimes. For this reason the majority of subsequent scholars have avoided conflating behaviour and institutions in the definition of the regime type and most have followed Elgie’s strictly institutional definition.Footnote 15 However, even establishing an institutional definition of semi-presidential regimes has proved difficult because of their institutional heterogeneity. Indeed, Siaroff contends that ‘there is really no such thing as a semi-presidential system when viewed through the prism of presidential powers.’Footnote 16

This debate raises two questions: apart from the methodological concern of avoiding tautological arguments, what theoretical reasons are there to think of differences in the nature of representative democracies as defined by institutions? And how can regime-defining institutional variation be distinguished from institutional variation within the regime type? Both questions require a theory of what defines the nature of representative democracies, which we argue, is provided by recent developments in principal–agent theory.Footnote 17

A Principal–Agent View of Semi-Presidential Constitutions

Principal–agent theory identifies the fact that democratic politicians are agents of the electorate as the defining feature of representative democracies in contrast to other types of political regimes. It views democratic politicians as representatives who exercise authority delegated to them by their voters to whom they remain accountable.Footnote 18 From this perspective, the nature of democracies differs with the way in which constitutions structure who wields political authority on behalf of voters, and what they are empowered to do.Footnote 19 In their ideal-typical form semi-presidentialism, parliamentarism and presidentialism, the three most common types of democratic constitutions vary along two fundamental, regime-defining dimensions. First, they vary in the number of agents whom voters empower to act on their behalf – that is voters may empower just one agent (parliament in parliamentary regimes) or two (president and assembly, as in presidential and semi-presidential regimes). Secondly, these democratic constitutions vary in the authority they grant to elected politicians to form and direct the government, which they structure in one of two ways: either they unequivocally subordinate the government to one agent of the electorate (the president in presidential regimes and the assembly in parliamentary regimes), or they do not, as in semi-presidential regimes, and thereby give rise to opportunities for both agents of the electorate, president and assembly to exercise some degree of joint authority vis-à-vis the government.

Semi-presidential constitutions do not unequivocally subordinate the government to either president or assembly because the government’s survival depends, on the one hand, on the confidence of the assembly, and, on the other hand, on the direct election of the president creating a second channel for voters to influence how the country is governed. It is this popular election, which gives presidents in semi-presidential regimes the status of head of state with a popular mandate separate from the assembly that may be used to affect the government. Semi-presidential constitutions typically reinforce the opportunities that the president’s popular election creates by granting presidents a range of powers in at least one of three areas (government formation, government termination and legislation). But it is the popular election of the president rather than the level of presidential executive power that differentiates semi-presidentialism from parliamentarism. Presidents in parliamentary regimes may also wield some constitutional powers to influence governments or policy but are elected by, and therefore remain agents of, parliament. As a result, the role of parliamentary heads of state is to facilitate parliamentary government, not to promote a separate and popularly legitimated set of presidential goals in influencing government and politics.Footnote 20 The definition which captures the nature of the semi-presidential delegation chain in a minimal and unambiguous way is Elgie’s institutional definition of semi-presidentialism as characterized by ‘a popularly elected, fixed term president [who] exists alongside a prime minister and cabinet who are responsible to parliament’.Footnote 21 This is the definition of semi-presidentialism that we adopt in this article, and which has been used by nearly all of the recent work in the field.Footnote 22

Yet, given the basic structure of their delegation chain, semi-presidential constitutions vary considerably in the level of powers they attribute to president and assembly, and in how they structure the interaction between these two sets of actors.Footnote 23 It is this variation that has most often led analysts to question the utility of semi-presidentialism as a distinct regime type.Footnote 24 We therefore investigate this variation in some detail and discuss its theoretical status in relation to the definition of semi-presidentialism as a regime type, turning first to powers over the government and then to the legislative sphere.

With respect to government formation, semi-presidential regimes range from granting the president no more than the option of vetoing a government by refusing to invest it (examples include Bulgaria, Niger), via presidential initiative in naming a premier, subject to assembly investiture (Portugal 1982, Lithuania, Macedonia, Mongolia, Poland, Slovenia, Romania, Russia), to presidential power to name a premier without investiture (France, Austria, Ukraine 1996, Taiwan, Mali). With respect to government dismissal, semi-presidential regimes vary similarly from reserving the power to dismiss the government for an assembly that the president cannot dissolve (Lithuania, Romania, Slovenia, Mongolia), via reserving power to dismiss the government for an assembly that can be dissolved by the president (France, Portugal, Mali), to allowing president and assembly equal powers to dismiss the government (Austria), or giving the president powers to dismiss the government while restricting the assembly’s ability to pass a vote of no confidence (Russia).Footnote 25 Most semi-presidential regimes make the assembly either the dominant, or a co-equal player in controlling the government, only a minority of these constitutions tip the balance in favour of the president.

From a principal–agent perspective, this institutional variation is clearly distinguishable from regime-defining institutional variation because it does not alter the fundamental, regime-defining feature of semi-presidentialism: that two agents of the electorate, assembly and president have – according to the formal constitutional rules – a status and mandate, which gives both the opportunity to exercise authority over the government. Nevertheless, within-regime type variation in the powers to control government composition is consequential because it affects the probability with which assembly parties or presidents can achieve their preferred outcomes in this situation of shared de jure authority.

Turning to legislative agenda control, we see similarly pronounced institutional variation. Semi-presidential regimes typically vest agenda powers primarily in the executive, but they can divide agenda control in different ways between president and government. Like presidential regimes, semi-presidential constitutions may grant presidents legislative powers ranging from a block veto (Moldova, Niger, Poland, Portugal, Romania, Russia), or a partial veto (Bulgaria, Mongolia) via amendatory observations to legislation (Ukraine, Lithuania), to legislative initiative, the power to declare legislation urgent, initiative to call a referendum, and the ability to issue executive orders or even legislative decrees (Russia).Footnote 26 Semi-presidential constitutions can also vest agenda powers in the government, and the French Constitution goes perhaps furthest among semi-presidential constitutions in giving dominant agenda powers to the government. The French government controls the legislative agenda, can use ordinances, and invoke an array of restrictive procedures to manage parliamentary decision making including closed rule, package votes and confidence votes.Footnote 27

Again, from a principal–agent perspective, this institutional variation is distinct from regime-defining institutional variation because it does not alter the fundamental nature of the delegation chain. Nevertheless, this type of within-regime variation in legislative agenda powers matters because it affects the probability with which assembly parties or the president are likely to achieve their preferred legislative and policy aims. Moreover, constitutional variation shapes the probability of legislative conflict between these actors. Semi-presidential regimes that grant agenda powers predominantly to the government in effect fuse control of the government and legislative agenda control, in a manner that parallels some aspects of parliamentarism. They afford president and assembly only indirect legislative agenda control – which is contingent on their ability to control the composition of the government. As a result, the main area of negotiation between president and assembly is government composition. A case in point is France: once government composition has been negotiated, legislative agenda control essentially follows, with some minor exceptions (such as the presidential power to refuse countersignature of government ordinances), and requires little further bargaining. Legislative gridlock is unlikely. By contrast, semi-presidential regimes with significant presidential agenda powers institutionally separate control of the government from control of the legislative agenda in a manner that parallels some aspects of presidentialism. Russia is a case in point. Here, in situations of cohabitation, an assembly that is able to control the government will have to continue to negotiate with the president over legislation.Footnote 28

Turning from institutions to behaviour, semi-presidential regimes can, as Lijphart and Duverger observed, display diverse behavioural patterns that range from more ‘presidential style’ governance, say, in France under President de Gaulle, or Russia under President Putin, to more ‘parliamentary style’ governance, say, in Austria under President Klestil.Footnote 29 This variation led Lijphart to doubt whether semi-presidentialism is in fact a distinct regime type, and prompted Duverger and others to question whether the constitutional format is a good predictor of political practice in these regimes.Footnote 30 But recent work, which approaches semi-presidential regimes from a principal–agent perspective, makes clear that it is precisely the formal constitutional structure of the semi-presidential chain of delegation that shapes the pattern and scope of potential variation in the de facto control of semi-presidential governments.Footnote 31 In other words, the behaviour of semi-presidential regimes can vary to this degree only because they make both the assembly and the president agents through which voters can potentially influence government and policy. As a result, the patterns of behavioural variation that semi-presidential regimes can display differ fundamentally from presidentialism and parliamentarism.

First, semi-presidentialism entails the possibility that the assembly or the president can act as the primary principal of the government at different points in time within the same regime.Footnote 32 Thus, a range of semi-presidential regimes have at various times seen governments formed and directed by the assembly, but have at other times had governments which were primarily formed and directed by the president without active party participation. Ukrainian presidents, for instance, have been major players in the appointment and dismissal of entirely non-partisan governments throughout a large part of the post-Soviet period.Footnote 33 In Finland, President Kekkonen appointed a series of non-partisan presidential cabinets between 1956 and 1982.Footnote 34 And in Bulgaria a close associate of President Zhelev, Ljuben Berov, was appointed to lead a non-partisan government in 1992, when ‘no obvious party-based cabinet was in sight’.Footnote 35 But at other times, all of these countries have had governments that were primarily composed and controlled by assembly parties. This potential for shifts in primary control of the government between president and assembly is a consequence of the semi-presidential chain of delegation, and can occur in semi-presidential regimes with very different levels of presidential constitutional power. The president of Bulgaria is constitutionally weak, the Finnish president is moderately powerful, and the Ukrainian president’s constitutional powers are very extensive.

Secondly, and relatedly, semi-presidentialism can give rise to situations of cohabitation, in which a president of one political persuasion faces a prime minister and government controlled by the parliamentary opposition. Cohabitation can arise because president and assembly are separately elected and parties opposed to the president may be able to dominate negotiations over government composition. The most recent French period of cohabitation between the conservative President Chirac and socialist Prime Minister Jospin (1997–2002) is a case in point. These possibilities illustrate that the differences between the semi-presidential chain of delegation and that of other types of representative democracies have important political consequences, which cannot be reconciled with the view that semi-presidentialism is merely a subspecies of presidentialism, or of parliamentarism, as proposed by Reestman.Footnote 36

None the less, the relative weight of the regime type in accounting for the behaviour of semi-presidential systems remains the subject of a major debate, which is complicated by the institutional variation among these regimes.Footnote 37 In the sections that follow we examine the effects of semi-presidential institutions on the behaviour of these systems, turning first to democratic stability and then to other aspects of democratic performance including representation and accountability, government formation and durability and the policy and legislative process.

Semi-presidentialism and democratic stability

Linz’s concern with the adverse effects of popularly elected presidents on democratic stability has powerfully shaped research on how semi-presidentialism affects democratic survival. Extending his critique of presidentialism to semi-presidential regimes, Linz focused on the potential for conflict, gridlock and breakdown, which semi-presidential regimes may entail.Footnote 38 Thus, Linz and Stepan see ‘grounds to be wary about a “dual executive” … if the president is directly elected and the prime minister is responsible to a directly elected parliament, there is a possibility for deadlock and constitutional conflict’.Footnote 39 Skach develops this reasoning further and argues that the ‘shared power, but unequal legitimacy and accountability [of president and prime minister], structure theoretically predictable and empirically verifiable tensions into the semi-presidential type’.Footnote 40 These tensions, if compounded by divided minority government (that is situations in which the government is supported by a legislative minority which opposes the president), Skach argues, increase the risk that powerful presidents will marginalize the assembly, and precipitate a slide towards authoritarianism.Footnote 41

The Linzian perspective, though, has not gone unchallenged. Pasquino and Sartori, for instance suggest that semi-presidentialism supports democratic survival by virtue of its flexibility.Footnote 42 In Pasquino’s words, semi-presidential regimes:

do not degenerate into plebiscitary democracy as is possible for presidential systems nor into assembly government as occurs in parliamentary systems. On the whole, under most circumstances, semi-presidential systems appear endowed with both more governmental capabilities and more institutional flexibility than parliamentary and presidential systems respectively.Footnote 43

And Frisson-Roche goes so far as to conclude that semi-presidentialism seems to be ‘the model which in both relative and absolute terms best enables a rapid transition between dictatorship and democracy, under the condition that the political actors accept to share power’.Footnote 44

A principal–agent perspective suggests that there is likely to be some truth in both sets of arguments: the attraction of the semi-presidential reliance on two agents of the electorate – assembly and president – is the potentially flexible control of the government combined with checks and balances. Its drawback lies in the potential for co-ordination problems and conflicts between these two agents. The perspective thereby highlights that two-agent designs in and of themselves are unlikely to be uniformly problematic or beneficial, and that their effects on democratic stability are likely to vary with the manner in which they structure the interaction between president and assembly.

Conceptual and empirical work suggests that semi-presidential constitutions, which enable stable solutions to conflicts between president and assembly that reflect the preferences of the electorate (their principal), reduce the risk of conflict and regime collapse. Historically, such designs have been chosen in the majority of semi-presidential regimes. Shugart and Carey first advanced the debate in this direction when they argued that a central quality of semi-presidential constitutions is their ability to produce stable outcomes (rather than gridlock or escalating conflict) in the cabinet appointment and dismissal game. The authors note that semi-presidential constitutions differ in how they structure the transaction between president and assembly over control of the cabinet and argue that semi-presidential regimes in which the cabinet is accountable only to the assembly induce stable bargaining outcomes. Shugart and Carey label such regimes premier-presidential, and contrast them with the president-parliamentary type, in which the president appoints a cabinet that is then dually accountable, and dismissible by both its democratic principals.Footnote 45 They argue that such ‘symmetry in [government] removal powers contributed to interbranch conflict in Portugal (1976–82), and to chronic cabinet instability in Germany’s Weimar Republic as well … [and encouraged] conflict and instability of the sort that led to the coup in Peru in 1968’.Footnote 46 Moestrup’s comparative study of semi-presidentialism in eighty-three young democracies lends considerable support to Shugart and Carey’s analysis. Her descriptive statistics suggest that ‘on average premier-presidential regimes have better FH [Freedom House] scores (3.0 versus 3.9) and better chances of democratic survival’.Footnote 47

In addition, case-oriented work draws attention to the importance of incentives for assembly and president not just to reach stable resolutions to their conflicts, but also to resolve their conflicts in accordance with the preferences of their ultimate principal – the electorate. Russia between 1991 and 1993, for example, had a constitution that resembled the (supposedly compromise-inducing) premier-presidential type and did not give the president the power to dissolve the assembly.Footnote 48 However, this constitution allowed situations to emerge in which the president faced a hostile assembly that was deeply unpopular, which would give the president, who had considerable executive powers, conflicting constitutional and electoral incentives. In 1993 President Yeltsin found himself in precisely this situation, so that he stood to gain electoral credit by responding to voter concerns about the unpopular assembly’s decisions ‘and defying the constitutional rules and checks of a hostile parliament [which he could not dissolve]’.Footnote 49 This problem helps to account for the destabilizing conflicts that marked Russian politics in 1993 and culminated in collapse of the First Russian Republic.

However, only a small minority of semi-presidential regimes feature such seriously problematic constitutional designs. As Shugart notes in recent work, many of the newly adopted (and supposedly conflict-prone) president-parliamentary constitutions moderate the potential for interbranch conflict in a number of ways and none ‘has Weimar’s … complete list of unilateral powers’.Footnote 50 Shugart’s conclusion – that regime-destabilizing constititutional features are not a general characteristic of semi-presidential regimes – receives powerful support from the empirical work of Cheibub, Moestrup and Elgie.Footnote 51 Cheibub traces three collapses of democracy among semi-presidential regimes in the thirty years up to 2002 (Congo, Comoros Islands and Niger), and his controlled comparative analysis of transitions to dictatorships in all parliamentary and semi-presidential democracies from 1946 to 2002 shows that semi-presidential regimes are no more likely to become dictatorships than parliamentary democracies.Footnote 52 Likewise Moestrup’s descriptive analysis of the frequency of democratic breakdowns suggests that ‘semi-presidential regimes are no more or less likely than either presidential or parliamentary regimes to suffer democratic breakdown through coups or otherwise’.Footnote 53 These results receive further support from Elgie’s analysis which examines descriptively whether there is in fact a link between democratic breakdowns under semi-presidentialism and the partisan conflict of divided minority government and cohabitation as assumed by the Linzian perspective. Elgie concludes ‘there is insufficient evidence to make the claim that cohabitation is perilous under semi-presidentialism … Niger is the only country where cohabitation has ever been directly responsible for the collapse of democracy’.Footnote 54 And while Elgie concurs that divided minority situations are prone to conflict, he stresses that ‘divided minority government is not necessarily fatal for partial democracies, and that full democracies with the possible exception of the Austrian First Republic, have always managed to survive divided minority government even if it has occurred in the earliest years of the transition process’.Footnote 55

In sum, the literature indicates that some, empirically rare, semi-presidential designs should be avoided. Thus, as a principal–agent perspective suggests, the manner in which semi-presidential regimes structure the interaction between president and assembly has important consequences for their chances of democratic survival. But the available comparative evidence also suggests, contra Linz, that semi-presidentialism as a regime type is as supportive of democratic survival as parliamentarism. The mechanisms of conflict management available in semi-presidential constitutions appear to allow the great majority of these regimes to manage cohabitation and divided minority situations successfully. These findings have shifted the focus of the research agenda from the Linzian concern with democratic stability towards the performance of semi-presidential regimes as democracies in areas as diverse as representation, government stability and the legislative effectiveness, and this is what we turn to next.

Semi-Presidentialism and Democratic Performance

Representation and Accountability

Representation can be defined as the ‘relationship between citizens’ interests and political outcomes, in which rulers act to meet the interests of the public’.Footnote 56 If we think of the ‘difference between the public’s preferences and how they are translated into policy by governments … as the “representation gap” ’,Footnote 57 then that gap can have two potential sources: politicians (as the agents of voters) may either be unable to act on behalf of their principals, or unwilling to do so. In other words, representation can entail classical principal–agent problems and the extent to which semi-presidentialism increases or reduces these problems – and thus the representation gap – has been subject to debate.

From a Linzian perspective semi-presidentialism is expected to widen the representation gap by reducing the willingness of politicians to represent a broad range of views within the electorate faithfully. The winner-takes-all nature of the presidential office, combined with the national constituency of presidents, is thought to encourage a false sense of mandate and to undermine a president’s willingness to compromise and to give due representation to minority views.Footnote 58 A principal–agent perspective diverges from the Linzian view by pointing out that the reliance of semi-presidentialism on two separately elected agents of voters can potentially enhance representation by reflecting a greater diversity of citizen interests, and may allow for more differentiated accountability of the president and the assembly for different aspects of policy.Footnote 59

The literature on representation has suggested a number of measures that can be used to guage the representation gap, including the congruence of the policy positions of governments and voters, and voters’ subjective assessments of the quality of representation.Footnote 60 The evidence with respect to both measures suggests that popularly elected presidents with significant executive powers may widen the representation gap. Thus Strøm et al. provide descriptive evidence on policy congruence, which suggests that more powerful presidents may detract from close representation of the median voter.Footnote 61 The authors compare the policy congruence of governments and median voters in fifteen West European democracies in the 1980s and 1990s. For the semi-presidential regimes amongst these democracies, they report variation ranging from extreme convergence (Ireland), via moderate divergence (Finland and Austria) to pronounced divergence between the government and the median voter in Portugal and France. This variation in policy congruence is consistent with the level of constitutional power enjoyed by presidents in these countries as reported by Strøm et al. (Ireland 6.5, Austria 7.5, Finland (1991) 10.5, France 11, Portugal 14–15) – broadly speaking, countries with more powerful presidents appear to represent voters less faithfully.Footnote 62 It is important to note that this is not a controlled analysis of policy congruence. Hence, it does not take into account the effects of electoral systems, the structure of voter cleavages, and other factors that may influence the distance between the policy preferences of governments and median voters. None the less, these correlations point to a potential disadvantage of semi-presidential regimes (relative to parliamentary systems) in their capacity to ensure policy congruence between voters and politicians, which deserves fuller investigation.

The impression that the size of the representation gap in semi-presidential regimes may be associated with presidential power is strengthened by other (controlled) comparative work on voter perceptions of representation. Whitefield examines how citizen’s perceptions of representation are structured by institutional (as well as economic and cultural) factors in thirteen post-communist states between 1993 and 1997. He finds that the most powerful ‘macro effect on … the representation gap … is the level of presidential power … [G]reater levels of presidential power are associated with higher representation gaps’.Footnote 63 Thus subjective and objective measures of representation appear to correlate negatively with presidential power.

Recent work by Samuels and Shugart develops a theoretical account for this representation gap by exploring how the separation of powers and purpose affects the representational strategies of presidents. The more constitutional power presidents have, the two authors argue, the greater their influence on policy – and the more distinct the mandates of president and assembly, the clearer the president’s responsibility for national policy. This, Samuels and Shugart suggest, shapes the representational strategies chosen by presidents. Like other politicians, presidents must choose what emphasis to place on closely following their programmes as mandated (mandate representation) or delivering results by the next election that serve their voters (accountability representation).Footnote 64 In contrast to the position of a prime minister in a parliamentary system, the powers of presidents are separate from those of the assembly, as are their terms, which are fixed. Moreover, the president’s mandate is separate from that of his or her party in the assembly. This gives a president greater leeway than a prime minister ‘to be above his/her party and to switch directions independently of party dictates’.Footnote 65 In other words, as the separation of powers and purpose rises, presidents have weaker incentives than prime ministers to focus on mandate representation, but the voters are more able to identify the president as responsible for national policy and to exercise accountability by choosing to vote out (or to return to office) the president or his/her chosen successor. ‘Meanwhile, the same voters may continue to hold legislators accountable for quite different things, such as providing a continued flow of government services to poor or marginalized areas.’Footnote 66 Thus, the more semi-presidential regimes depart from parliamentary-style collective government accountability, the more they give voters the option of holding presidents and legislators accountable for different aspects of political performance. While semi-presidentialism might therefore be expected to detract from mandate representation, Samuels and Shugart’s work suggests that it has compensating strengths with respect to accountability.

Accountability can be understood as ‘the degree and means by which elected policymakers are responsible to citizens’.Footnote 67 The ability of voters to reward or sanction incumbents is ‘conditional on two dimensions: 1) their capacity to assign responsibility for outcomes to incumbents and 2) their ability to act upon their responsibility attributions’.Footnote 68 Samuels and Shugart’s work suggests that the separation of powers and purpose enhances accountability along both dimensions. This view contrasts with the Linzian perspective, which expects semi-presidentialism to detract from accountability in both ways.Footnote 69 The complex delegation regime and the requirement for transaction between president and assembly is thought to make it difficult to establish where responsibility for government and policy lies. Moreover, powerful presidents are thought to be able to deprive voters of the opportunity to use their vote to reward or punish. As Skach argues:

[i]n semi-presidentialism, a president can put his prime minister in charge of solving difficult economic crises or near-crises, as Valery Giscard d’Estaing did with Raymond Barre in 1976, and then wait for the results. The president then ‘takes the credit for any substantial improvement in the economy; and finally, he can sacrifice his Prime Minister if the government’s economic policies aggravate the crisis. The dual executive system, therefore, has the advantage of shielding a President from trouble’.Footnote 70

On this basis, Skach concludes that semi-presidentialism affords voters less accountability than either presidentialism or parliamentarism.Footnote 71

The empirical evidence offers little support to the Linzian perspective, but suggests that Shugart and Samuels correctly identify accountability as a strength of regimes with popularly elected presidents. Controlled, survey-based studies demonstrate that voters in semi-presidential systems are able to assign responsibility for policy with great precision. Lewis-Beck examines voters’ perceptions of responsibility for the economy and their voting intentions in France, and finds that voters correctly assign responsibility for economic performance to the prime minister, not the president, during periods of cohabitation.Footnote 72 Moreover, Lewis-Beck and Nadeau demonstrate that French voters do not just target their perceptions of responsibility correctly, they also weigh responsibility for economic performance:

A Prime Minister who is not backed by the powers of the President cannot be as effective in shaping the national economy. There are simply things he or she cannot do that the President can. When the executive is unified, with a President commanding a dutiful Assembly shepherded by his hand-picked lieutenant, economic intervention is more easily accomplished.Footnote 73

As is consistent with this sophisticated understanding of the weight of responsibility for economic policy, the economic vote is ‘strongest under unified government, in presidential elections where an incumbent is running for re-election … Under cohabitation the target of the economic vote shifts to the Prime Minister,’Footnote 74 and its impact on voter choice in presidential elections is reduced.

While both studies focus only on France as an example of a semi-presidential regime, their conclusions are supported by the work of Hellwig and Samuels, who examine whether semi-presidentialism also gives voters the opportunity to act on these assignations of responsibility by punishing or rewarding politicians. The authors analyse regime type effects on electoral outcomes in seventy-five countries and 562 elections between 1975 and 2002, and conclude that voters in semi-presidential regimes can and do sanction politicians. They find that under conditions of unified government the direct election of the president leads to enhanced accountability to the public – ‘voters reward or punish incumbent presidents to a far greater extent than they reward or punish parliamentary parties in semi-presidential systems, but only under unified government’.Footnote 75 Under cohabitation, voters punish or reward the prime minister’s party for economic performance. This work suggests that voters do not just assign responsibility correctly, but are also able to effectively hold their politicians to account. Interestingly, though, Hellwig and Samuels’s work suggests that the scope of a president’s executive powers has no effect on these patterns of accountability in semi-presidential regimes.

In sum, over recent years researchers have begun to focus on the performance of semi-presidential regimes as democracies, including the effectiveness of the representation and accountability links between voters and politicians. A number of open questions remain. The difference that institutional variation within semi-presidential regimes does or does not make to electoral accountability needs to be investigated in more detail. Moreover, Whitefield’s work raises interesting questions about the relationship between the representational strategies chosen by politicians and voters’ subjective perceptions of the quality of representation. However, the basic conclusion supported by the work available to date is clear and supports the expectation generated by a principal–agent perspective that the semi-presidential delegation regime is likely to have distinct advantages and drawbacks for voters. The semi-presidential chain of delegation appears to create a clear risk for citizen-principals with respect to representation, which is that presidents, empowered by their own mandate, and secure in their position by virtue of their fixed term, may switch policy direction independently of party dictates. Empirically, semi-presidentialism appears to widen the mandate representation gap between voters and politicians as their agents. However, with respect to accountability the semi-presidential chain of delegation appears to have distinct strengths by enabling voters to ensure the effective accountability of their politicians, and specifically, of the two parts of the dual executive, through separate presidential and assembly elections. The next section considers how elected politicians then translate voter choices into governments – the second link in the chain of delegation and accountability.

Government Composition and Durability

While the potential for some degree of joint control over the government by president and assembly is a defining feature of the regime type, the question of how it affects the nature of governments and their durability has long been a central subject in the debate about semi-presidentialism. Various studies suggest that the regime type features an enhanced flexibility to ensure the formation and survival of governments even in difficult political circumstances because the cabinet can potentially draw support from president or assembly.Footnote 76 However, the more prominent view is that semi-presidentialism fosters conflict and cabinet crises, complicates coalition formation and encourages the appointment of cabinets that lack parliamentary support, especially when the assembly is broadly hostile to the president but cannot form a decisive majority;Footnote 77 ‘In such a situation the president might turn to extra-parliamentary leaders to provide support in forming a government … [and] the temptation to govern without or against the legislature, … cannot be excluded’.Footnote 78 This is likely to result in ‘a narrowing of the decision-making arena to a small group of handpicked, non-party ministers[, which] … violates the democratic principles of inclusion and contestation’.Footnote 79 From this perspective the compatibility of president-controlled governments with democratic representation and accountability is doubtful precisely because it is thought to provoke destabilizing conflict over governments.

A principal–agent approach differs from the Linzian perspective in two fundamental respects. First, it identifies presidents in semi-presidential regimes as agents of the electorate, who, like assemblies, play a role in translating electoral choices into governments. From this perspective presidential cabinet control is a priori nothing more and nothing less than a democratic alternative to assembly party government. Secondly, a principal–agent approach suggests that government composition and durability is likely to vary with the manner in which semi-presidential constitutions structure the negotiations between the assembly and the president. Shugart and Carey first developed this argument by hypothesizing that the relative de facto control of president and assembly over government formation varies with a president’s constitutional power – ceteris paribus they expected more powerful presidents to have greater influence on cabinet formation.Footnote 80 Moreover, these authors expected durable governments under most semi-presidential constitutions except in the president-parliamentary sub-set, which undermines stable outcomes because ‘either president or assembly can dismiss ministers’.Footnote 81 Our discussion turns first to government formation, then to durability.

There is by now significant empirical support for Shugart and Carey’s argument that presidential influence on cabinet composition varies with a president’s constitutional powers (amongst other things). Protsyk’s study of sixty-one semi-presidential governments in Eastern Europe and the Former Soviet Union shows that the balance of constitutional powers over cabinet composition between president and assembly correctly predicts the prime minister’s political proximity to the ideal point of president or assembly in three quarters of his cases.Footnote 82 Moreover, recent work has demonstrated that presidential influence on government composition changes and loosens the party–government relationship (in comparison to parliamentary regimes, where governments are typically nearly entirely partisan). Here Amorim Neto and Strøm broke new ground by suggesting that presidents may have a weaker preference for partisan cabinet ministers than assembly parties, because presidents are required to build national electoral coalitions, often across or above parties.Footnote 83 Their controlled comparative study shows that in the 1990s presidential influence on cabinets in twenty-four East and West European countries increased the share of non-partisan ministers in government. Schleiter and Morgan-Jones address the question of how politically important such non-partisan presidential appointees in semi-presidential governments are, and show that the general differences in the party–government relationship between parliamentarism and semi-presidentialism carry over into control of the top government posts, including the premiership.Footnote 84 Their study of 438 governments in twenty-eight European parliamentary and semi-presidential democracies throughout the second half of the twentieth century demonstrates that in parliamentary regimes 96–98 per cent of the top government portfolios are controlled by parties, whilst in semi-presidential regimes, fully one in six prime ministers and a quarter of all foreign ministers escape party control.

Unlike Skach, both studies draw attention to potential advantages as well as drawbacks of presidential cabinet control on democratic performance. As Amorim Neto and Strøm put it, ‘[s]tronger presidents may come to represent a form of parliamentarism that allows more channels for more diverse and direct citizen input, and perhaps more direct lines of accountability. At the same time, semi-presidentialism may imply a weaker role for political parties in coordinating this chain of democratic delegation’.Footnote 85

Indeed, the suggestion that presidential influence on the cabinet creates more channels for more diverse citizen input into cabinets receives empirical support from Cheibub’s work, which shows that presidential influence on the cabinet in semi-presidential regimes does not appear to undermine compromise, engagement with the legislature or coalition building in the manner that Linz and others suggest. Cheibub’s comparison of the frequency and majority status of government coalitions in all semi-presidential and parliamentary democracies (1946–2002) finds no prima facie support for the view that presidents complicate coalition formation.Footnote 86 Quite the reverse – the descriptive analysis that Cheibub presents suggests that semi-presidential governments are more frequently based on coalitions than parliamentary governments, and these coalitions achieve majority status more often. But just how meaningful the flexibility of semi-presidential regimes in government formation is, depends of course in part on the stability of the governments that are negotiated. This is the issue to which we turn next.

Broadly speaking, governments lose office when they cannot retain the confidence of their principals or muster legislative support for their policy agenda. Semi-presidential governments potentially have two elected principals: president and assembly. By definition, they require at least passive toleration by the assembly to survive. But in addition, their survival can be contingent on the president who may for example have powers to dismiss the cabinet or to dissolve the assembly. This dependence on two principals appears to make semi-presidential governments more short-lived than those of other democracies: Cheibub examines government durations in all democracies from 1946 to 2002 and reports that semi-presidential regimes have the shortest government durations, with an average of 2.9 years, compared to 4 and 4.7 years respectively for parliamentary and presidential governments.Footnote 87 This raises the question of whether the relatively short government durations of semi-presidential regimes reflect a reduced cabinet durability – defined as the expected duration of a cabinet, given its attributes, including its relationship to its principals.Footnote 88 Unfortunately, standard-setting comparative work on government durability in European democracies has not modelled the president’s influence and has subsumed semi-presidential cases with parliamentary democracies.Footnote 89

Although an average government duration of nearly three years suggests that serious government instability does not characterize semi-presidential regimes as a class, the case-oriented literature indicates that there is significant variation across semi-presidential regimes. There is some evidence that this variation is in part constitutionally structured, as Shugart and Carey propose, but these effects are not yet well understood.Footnote 90 Thus, Roper’s work suggests that more powerful presidents engender greater government instability,Footnote 91 but Strøm and Swindle’s study of strategic parliamentary dissolution in eighteen European democracies highlights presidential powers that may enhance government stability, such as the power to veto the dissolution of parliament by the prime minister. Moreover, these authors find no evidence that the unilateral presidential power to dissolve parliament correlates with a greater probability of assembly dissolution.Footnote 92

In sum, recent research on government composition and stability in semi-presidential regimes has significantly broadened the research agenda beyond the Linzian concerns. Scholars have examined the effects of presidents on the political orientation of the prime minister, the party–government relationship, the majority and coalition status of governments, and are beginning to explore the effects of presidential powers on variation in government stability. Their findings suggest that it is misleading to characterize the effects that presidential involvement with the government has on democratic performance as unequivocally negative. The empirical evidence lends support to the view that presidential influence on the government is compatible with democratic delegation and accountability – as would be expected from a principal–agent perspective. Moreover, as principal–agent theory suggests, the potential for joint control of the government, which is inherent in the semi-presidential chain of delegation, appears to have particular strengths and weaknesses. Semi-presidential regimes have a degree of flexibility in basing governments on presidential or assembly support and, while presidential influence loosens the party government relationship, it appears to facilitate coalition formation. But the potential for joint control of the government also correlates with shorter government durations, and although serious government instability is not a general feature of semi-presidentism as a regime type, its effects on government durability merit further investigation. We now turn to the effectiveness of semi-presidential governments in the legislative and policy process.

Policy and Legislative Process

Principal–agent theory suggests that the delegation of policy-making and legislative power from voters to elected politicians and governments can be more or less effective for the citizen-principals, and sees this effectiveness as influenced by the way in which constitutions allocate responsibility for policy and powers to propose, amend, veto and enact legislation. Constitutions thereby shape the motivations and ability of politicians to advance particular policy agendas on behalf of their constituents, and influence politicians’ incentives to co-operate in the policy process.Footnote 93

As we saw above, constitutional powers to lead the policy process in semi-presidential regimes may lie with the government, the president or both. Some scholars see the potential for flexible policy leadership by president or prime minister as a factor that is likely to enhance the performance of semi-presidential regimes,Footnote 94 others note the potential for gridlock between president and prime minister,Footnote 95 which may undermine the effective formulation and implementation of public policy. This latter Linzian perspective is not inconsistent with Tsebelis’s veto players approach: popularly elected presidents bring an element of the separation of powers and purpose to semi-presidential regimes and thereby increase the number and policy distance between institutional veto players.Footnote 96 This should, from a veto players perspective, decrease the ability of semi-presidential regimes to achieve policy change compared to parliamentarism, other things being equal.

As we shall see in this section, the ability to deliver policy change is one important dimension of regime performance in the policy arena – and research in this area has expanded beyond the Linzian concern with conflict within the dual executive, to investigate veto and agenda powers in semi-presidential regimes and their effects on legislative success and policy coherence. But the work available to date also highlights a second dimension of regime performance in the policy arena: the nature of the policies that a regime is likely to provide, specifically the balance between public and private goods.Footnote 97 This dimension of policy performance eludes a veto players approach because it requires reference to the mandate and electorate of politicians and their motivation to serve their democratic principal – issues, which are best understood from a principal–agent perspective.

Empirical work supports Linz’s concern about conflict between the two parts of the semi-presidential executive but offers no systematic evidence to suggest that policy gridlock is a general feature of semi-presidentialism. Two separate studies by Protsyk and Sedelius document frequent intra-executive conflict in Central and East European semi-presidential regimes since the early 1990s, and Protsyk concludes that presidential veto powers, cohabitation and minority status of the cabinet are among the factors that make such conflicts more likely.Footnote 98 The question of whether such conflicts result in a gridlocked policy process is addressed by Cheibub, who compares the average legislative effectiveness of all semi-presidential and parliamentary governments from 1946 to 2002 (measured as the proportion of government-initiated legislation that is enacted).Footnote 99 Cheibub finds no regime-type differences in the effectiveness of majority (coalition or single-party) governments. This is not a controlled analysis, though, and there is reason to expect that the diverse legislative agenda control mechanisms in semi-presidential regimes engender significant within-regime type variation in legislative effectiveness.

Both case study and comparative work suggests that the location of agenda control powers in semi-presidential regimes affects the co-ordination of the policy process and legislative success of governments. As is consistent with a veto players approach, semi-presidential regimes that concentrate agenda powers in the government and thus fuse legislative agenda control with control over the government, can have highly co-ordinated legislative processes. Huber’s work, for example, demonstrates that French governments deploy their constitutional powers to use restrictive legislative procedures in managing the legislative process in parliament, including the challenges of conflict within their coalitions and of minority status.Footnote 100 A governmental success rate of 99.5 per cent on final bills under Socialist minority government in France at the end of the 1980s is suggestive of the power of such constitutional rules.Footnote 101

By contrast, semi-presidential regimes that grant important agenda powers to the president can complicate policy change and compromise the coherence of policy. This, too, is consistent with a veto players approach. The case-oriented literature highlights that extensive presidential agenda powers potentially separate control over the government from control over the legislative agenda.Footnote 102 In such semi-presidential regimes, the president, the government and the assembly too may have powers to advance uncoordinated and competing policy initiatives. As Protsyk notes on Ukraine:

The presidential use of powers to issue decrees and executive orders results in the diffusion of decision-making responsibilities: a cabinet is no longer the only executive institution in the centre of government. Competing policy initiatives, parallel decision routes, [and an] excessive burden of bureaucratic coordination are all negative effects of the diffusion of executive powers [between the two parts of the executive].Footnote 103

However, performance in the policy arena is not simply a matter of the ease of policy change, it also concerns the substance of the policies themselves. As Shugart suggests, presidential agenda powers can contribute to the provision of public policy and public goods, and they can become critical when the party system is too weakly institutionalized to fulfil this role.Footnote 104 Presidential agenda powers create the option for governments, which lack stable parliamentary support, to rely on the president to advance an agenda. Moreover, because presidents represent a nationwide electorate, and are accountable to it, they have incentives to focus on the provision of policies, which benefit this broad constituency, that is, public goods. Work on Russia and Ukraine suggests that presidents can indeed be crucial in ensuring that governments pursue policies to provide public goods in the context of weakly institutionalized party systems.Footnote 105 Russian presidents have used a range of tools including ‘decrees and vetoes, their annual address, budget statements, and advisory commissions to outline and initiate policy’.Footnote 106 Between 1993 and 1999, when none of the Russian governments were supported by a governing coalition in the assembly, Yeltsin used these powers to ensure that governments made progress with mass privatization, control of inflation and legal reforms that provided the underpinnings of a market economy.Footnote 107

In sum, the empirical work lends support to the principal–agent theoretical expectation that the effectiveness of policy and legislative process in semi-presidential regimes varies with how constitutions structure the interaction over policy and legislation. While conflicts over policy can characterize semi-presidential regimes, policy gridlock is not a general feature of the regime type. Rather, the concentration of legislative agenda powers in the hands of the government appears to enhance the smooth passage of legislation, while a fragmentation of agenda powers between president, government and assembly seems to reduce it. These empirical patterns are consistent, too, with Tsebelis’s veto players perspective. A potentially countervailing strength of regimes with presidential agenda power, however, is highlighted only by the principal–agent approach and stems from the fact that presidents are agents of and accountable to the national median voter, which generates incentives for them to provide public goods. This regime type effect on policy may be especially critical to the performance of democracies with underdeveloped party systems, but much of the empirical work in this area remains to be done.

Semi-presidentialism: the changing research agenda

To conclude, scholars have made impressive progress in the study of semi-presidentialism over recent years. The last decade or so has seen advances towards the clarification of the conceptual debate about semi-presidentialism, a rapid broadening of the research agenda beyond Linz’s concern with the adverse effects of presidents on democratic stability and the pinpointing of regime type effects on the performance of semi-presidential democracies by an increasing number of controlled comparative studies. Our argument in this article has been that principal–agent theory is providing powerful conceptual tools to underpin this work, which serve to clarify three debates of importance in the study of semi-presidentialism and comparative politics more generally.

The first is the long-standing debate about the conceptual status of semi-presidentialism and the classification of democratic regime types. By providing a conceptual basis for the classification of democratic regimes, principal–agent theory advances this debate significantly beyond Duverger’s original work on the ‘new political system model’. It makes clear that it is the nature of the democratic delegation relationships that distinguishes different types of democratic regimes. From this perspective, semi-presidentialism is unequivocally defined as a distinct democratic regime type by the fact that it constitutionally entails the possibility that both agents of the electorate – president and assembly – can exercise some, although often asymmetrical, authority over the government.

The second area in which principal–agent theory clarifies the debate is the study of regime type effects on the performance of democracies. Work in this area since Linz has been characterized by an asymmetrical focus on the adverse effects of popularly elected presidents. By highlighting that the inclusion of presidents in semi-presidential regimes offers a means to mitigate some of the pitfalls of democratic delegation, but creates its own set of agency problems, principal–agent theory has opened the debate to a more balanced set of questions about the effects of presidents on democratic performance. This perspective not only makes intelligible the fact that semi-presidential regimes have a survival record, which rivals that of parliamentarism, which is baffling from a Linzian point of view, it also takes the analysis beyond the predominant concern with democratic stability. Principal–agent theory directs attention to the question how semi-presidential regimes perform as democracies and offers a framework to integrate research on a broad range of aspects of democratic performance from representation to the policy process that have often been studied in isolation. The picture that emerges when we draw together the findings of this research suggests powerful regime type effects. With respect to representation, the dual delegation regime of semi-presidentialism gives presidents (as well as assemblies) a role in representing the views of citizens – and presidents appear to reduce mandate representation, an effect that seems to become more pronounced as presidential powers grow. But the regime type has countervailing strengths in providing clear accountability. Likewise, semi-presidentialism affects the nature of governments and as presidential influence on the cabinet grows, the relationship between assembly parties and the government loosens, but there is no evidence that this complicates coalition formation. Serious government instability is not a feature of the regime type although semi-presidential regimes display shorter average government durations than other types of democracies. Finally, semi-presidentialism appears to affect the policy and legislative process: presidents may play a central role in securing policies that provide public goods, in particular in countries with underdeveloped party systems, but to the extent that they figure as an additional player in the legislative and policy process, they also seem to generate co-ordination problems and conflicts that can make it harder to achieve policy change. In sum, this work suggests that the Linzian view of presidents as solely detracting from the performance of semi-presidential regimes is misleading. Although the semi-presidential delegation chain creates its own set of problems, it has clear strengths in ensuring that voter choices translate into effective governments and policy.

The third question, which principal–agent theory clarifies, is whether regime-type effects on democratic performance can be adequately understood in terms of veto players and their impact on policy change as Tsebelis’s work suggests. We agree with Tsebelis that the number of institutional veto players affects the ability of a regime to deliver policy change, which is an important aspect of its performance. But the literature we have reviewed also demonstrates that the importance of presidents in semi-presidential constitutions does not primarily derive from their role as an additional institutional veto player or agenda setter in the policy process. Their importance derives primarily from their democratic role as executive actors who represent (and are accountable to) voters in a single national constituency. An understanding of this democratic delegation relationship is crucial in order to account for the motivations and political aims of presidents, which give rise to the tendency of presidents to privilege accountability over mandate representation, which motivate presidents to make choices in government composition that weaken the party–government relationship and which may encourage a presidential focus on public goods in policy choice. Presidential choices, which reflect these motivations, in turn shape the performance of semi-presidential democracies. In other words, the democratic role of presidents in semi-presidential regimes conditions how they use their veto and agenda powers. This makes the relationship between voters and elected politicians fundamental to evaluations of democratic performance and so analytically prior to questions of policy change. Principal–agent theory makes that clear.

We see the challenges ahead for scholars in this field as twofold. First, more work needs to be done to develop models that accurately specify the influence of within-regime type institutional variation on the performance of semi-presidential regimes. The second challenge ahead lies in establishing, theoretically and empirically, what the connections are across different links in the semi-presidential chain of delegation. For example, much work remains to be done on the effect that semi-presidential institutions have on policy implementation and the control of bureaucracies. As a field, studies of semi-presidentialism have advanced our understanding of this regime type significantly beyond the work of both Duverger and Linz over recent years – the research frontiers now lie in deepening and joining up the findings.

Footnotes

* Department of Politics and International Relations, University of Oxford; also St Hilda’s College, Oxford, and Keble College, Oxford, respectively (email: ). The authors thank the Editor, Albert Weale, and the anonymous referees for their very helpful comments on an earlier version of this article.

References

1 Maurice Duverger, ‘A New Political System Model: Semi-Presidential Government’, European Journal of Political Research, 8 (1980), 165–87; Robert Elgie, ‘The Politics of Semi-Presidentialism’, in Robert Elgie, ed., Semi-Presidentialism in Europe (Oxford: Oxford University Press, 1999), pp. 1–21.

2 José Antonio Cheibub, Presidentialism, Parliamentarism, and Democracy (New York: Cambridge University Press, 2007), p. 43.

3 Matthew Soberg Shugart, ‘Comparative Executive–Legislative Relations’, in R. A. Rhodes, Sarah Binder and Bert Rockman, eds, The Oxford Handbook of Political Institutions (Oxford: Oxford University Press, 2006), pp. 344–65; Robert Elgie, ‘What Is Semi-Presidentialism and Where Is It Found?’ in Robert Elgie and Sophia Moestrup, eds, Semi-Presidentialism Outside Europe (New York: Routledge, 2007), pp. 1–13, at p. 9.

4 Matthew Soberg Shugart, ‘Semi-Presidential Systems: Dual Executive and Mixed Authority Patterns’, French Politics, 3 (2005), 323–51, p. 344.

5 Duverger, ‘A New Political System Model’.

6 Juan Linz, ‘Presidential Versus Parliamentary Democracy: Does It Make a Difference?’ in Juan Linz and Arturo Valenzuela, eds, The Failure of Presidential Democracy (Baltimore, Md.: Johns Hopkins University Press, 1994), pp. 3–87; Juan Linz, ‘Introduction’, in Ray Taras, ed., Postcommunist Presidents (Cambridge: Cambridge University Press, 1997), pp. 1–14.

7 Arend Lijphart, Patterns of Democracy: Government Forms and Performance in Thirty-Six Countries (New Haven, Conn.: Yale University Press, 1999), pp. 121–3; Alan Siaroff, ‘Comparative Presidencies: The Inadequacy of the Presidential, Semi-Presidential and Parliamentary Distinction’, European Journal of Political Research, 42 (2003), 287–312.

8 Matthew Soberg Shugart and John M. Carey, Presidents and Assemblies: Constitutional Design and Electoral Dynamics (New York: Cambridge University Press, 1992); José Antonio Cheibub, ‘Mixed Systems and Democratic Performance: Do Popularly Elected Presidents in Parliamentary Systems Matter?’ (paper presented at the Annual Meeting of the American Political Science Association, Philadelphia, 2006).

9 Kent Eaton, ‘Parliamentarism Versus Presidentialism in the Policy Arena’, Comparative Politics, 32 (2000), 355–73; George Tsebelis, Veto Players: How Political Institutions Work (Princeton, N.J.: Princeton University Press, 2002).

10 Tsebelis, Veto Players, p. 67.

11 There is also an increasing interest in semi-presidential constitutions among legal scholars. Examples include Miroslaw Wyrzykowski and Agnieszka Cielen, ‘Presidential Elements in Government, Poland – Semi-Presidentialism or “Rationalised Parliamentarism”?’ European Constitutional Law Review, 2 (2006), 253–67; Vlad Constantinesco and Stephanie Pierre-Caps, ‘Presidential Elements of Government in France: The Quest for Political Responsibility of the President in the Fifth Republic’, European Constitutional Law Review, 2 (2006), 341–57; Ana Martins, ‘Presidential Elements in Government. The Portuguese Semi-Presidential System’, European Constitutional Law Review, 2 (2006), 81–100; Antero Jyränki, ‘Presidential Elements in Government. Finland, Foreign Affairs as the Last Stronghold of the President’, European Constitutional Law Review, 3 (2007), 285–306.

12 Shugart and Carey, Presidents and Assemblies; Arthur Lupia, ‘Delegation and Its Perils’, in Kaare Strøm, Wolfgang C. Müller and Torbjörn Bergman, eds, Delegation and Accountability in Parliamentary Democracies (Oxford: Oxford University Press, 2003), pp. 33–54; Kaare Strøm, ‘Parliamentary Democracy and Delegation’, in Strøm, Müller and Bergman, eds, Delegation and Accountability in Parliamentary Democracies, pp. 55–106; Shugart, ‘Semi-Presidential Systems’; Shugart, ‘Comparative Executive–Legislative Relations’.

13 Terry M. Moe, ‘The New Economics of Organization’, American Journal of Political Science, 28 (1984), 739–77; Lupia, ‘Delegation and Its Perils’; Strøm, ‘Parliamentary Democracy and Delegation’.

14 Duverger, ‘A New Political System Model’, p. 166.

15 Elgie, ‘The Politics of Semi-Presidentialism’.

16 Siaroff, ‘Comparative Presidencies’, p. 307.

17 Strøm, ‘Parliamentary Democracy and Delegation’; Shugart, ‘Semi-Presidential Systems’; David Samuels and Matthew S. Shugart, ‘Presidents, Prime Ministers, and Parties: A Neo-Madisonian Theory of Party Organization and Behavior’ (paper presented at the Annual Meeting of the American Political Science Association, Philadelphia, 2006).

18 Moe, ‘The New Economics of Organization’; Shugart and Carey, Presidents and Assemblies; Lupia, ‘Delegation and Its Perils’; Strøm, ‘Parliamentary Democracy and Delegation’; Shugart, ‘Semi-Presidential Systems’; Shugart, ‘Comparative Executive–Legislative Relations’.

19 Kaare Strøm, ‘Parliamentary Government and Legislative Organization’, in Herbert Döring, ed., Parliaments and Majority Rule in Western Europe (Frankfurt: Campus Verlag, 1995), pp. 51–82, at p. 74.

20 For a more detailed discussion of this point, see Petra Schleiter and Edward Morgan-Jones, ‘Party Control over European Cabinets?’, European Journal of Political Research, 48 (2009), 665–693.

21 Elgie, ‘The Politics of Semi-Presidentialism’, at p. 13.

22 This conceptualization of semi-presidentialism explicitly does not use Duverger’s criterion of considerable presidential powers to categorize regimes as semi-presidential, because of its inherent subjectivity. For a detailed discussion of this point, see Elgie, ‘What Is Semi-Presidentialism and Where Is It Found?’ at pp. 2–6.

23 Joel Hellman, ‘Constitutions and Economic Reform in Postcommunist Transitions’, East European Constitutional Review, 5 (1996), 46–56; Timothy Frye, ‘A Politics of Institutional Choice, Post-Communist Presidencies’, Comparative Political Studies, 30 (1997), 523–52; Lee Kendall Metcalf, ‘Measuring Presidential Power’, Comparative Political Studies, 33 (2000), 660–85, p. 657; Siaroff, ‘Comparative Presidencies’.

24 Siaroff, ‘Comparative Presidencies’.

25 Shugart, ‘Semi-Presidential Systems’, pp. 336–7.

26 Shugart, ‘Semi-Presidential Systems’, pp. 336–7; George Tsebelis and Tatiana P. Rizova, ‘Presidential Conditional Agenda Setting in the Former Communist Countries’, Comparative Political Studies, 40 (2007), 1155–82.

27 John D. Huber, ‘Restrictive Legislative Procedures in France and the United States’, American Political Science Review, 86 (1992), 675–87; John D. Huber, Rationalizing Parliament: Legislative Institutions and Party Politics in France (New York: Cambridge University Press, 1996).

28 Petra Schleiter and Edward Morgan-Jones, ‘Russia: The Benefits and Perils of Presidential Leadership’, in Robert Elgie and Sophia Moestrup, eds, Semi-Presidentialism in Central and Eastern Europe (Manchester: Manchester University Press, 2008), pp. 159–179.

29 Lijphart, Patterns of Democracy, pp. 121–2; François Frisson-Roche, ‘Semi-Presidentialism in a Post-Communist Context’, in Elgie and Moestrup, eds, Semi-Presidentialism Outside Europe, pp. 56–77, at pp. 67–73.

30 Duverger, ‘A New Political System Model’, p. 167.

31 Shugart, ‘Semi-Presidential Systems’; Octavio Amorim Neto and Kaare Strøm, ‘Breaking the Parliamentary Chain of Delegation: Presidents and Non-Partisan Cabinet Members in European Democracies’, British Journal of Political Science, 36 (2006), 619–43; Schleiter and Morgan-Jones, ‘Party Control over European Cabinets?’

32 Gianfranco Pasquino, ‘Semi-Presidentialism: A Political Model at Work’, European Journal of Political Research, 31 (1997), 128–37; Giovanni Sartori, Comparative Constitutional Engineering: An Inquiry into Structures, Incentives and Outcomes (Basingstoke, Hants.: Macmillan, 1997).

33 Oleh Protsyk, ‘Troubled Semi-Presidentialism: Stability of the Constitutional System and Cabinet in Ukraine’, Europe-Asia Studies, 55 (2003), 1077–95.

34 Jaakko Nousiainen, ‘From Semi-Presidentialism to Parliamentary Government: Political and Constitutional Developments in Finland’, Scandinavian Political Studies, 24 (2001), 95–109, p. 101.

35 Jean Blondel and Svetzosar A. Andreev, ‘Bulgaria’, in Jean Blondel and Ferdinand Müller-Rommel, eds, Cabinets in Eastern Europe (Basingstoke, Hants.: Palgrave, 2001), pp. 131–41, at p. 137.

36 Jan Herman Reestman, ‘Presidential Elements in Government’, European Constitutional Law Review, 2 (2006), 54–9, p. 58.

37 Elgie, ‘What Is Semi-Presidentialism and Where Is It Found?’ at pp. 10–11.

38 Linz, ‘Presidential Versus Parliamentary Democracy’, at p. 55.

39 Juan J. Linz and Alfred Stepan, Problems of Democratic Transition and Consolidation (Baltimore, Md.: Johns Hopkins University Press, 1996), pp. 278–9.

40 Cindy Skach, Borrowing Constitutional Designs: Constitutional Law in Weimar Germany and the French Fifth Republic (Princeton, N.J.: Princeton University Press, 2005), p. 14.

41 Skach, Borrowing Constitutional Designs, p. 124; Cindy Skach, ‘The “Newest” Separation of Powers: Semipresidentialism’, International Journal of Constitutional Law, 5 (2007), 93–121.

42 Pasquino, ‘Semi-Presidentialism’; Sartori, Comparative Constitutional Engineering.

43 Pasquino, ‘Semi-Presidentialism’, pp. 136–7.

44 Frisson-Roche, ‘Semi-Presidentialism in a Post-Communist Context’, at p. 75.

45 Shugart and Carey, Presidents and Assemblies, pp. 68–71.

46 Shugart and Carey, Presidents and Assemblies, p. 118.

47 Sophia Moestrup, ‘Semi-Presidentialism in Young Democracies: Help or Hindrance?’ in Elgie and Moestrup, eds, Semi-Presidentialism Outside Europe, pp. 30–55, at p. 41.

48 Petra Schleiter, ‘Mixed Constitutions and Political Instability: Russia 1991–1993’, Democratization, 10 (2003), 1–26, pp. 7–9.

49 Schleiter, ‘Mixed Constitutions and Political Instability’, p. 23.

50 Shugart, ‘Semi-Presidential Systems’, p. 340.

51 Cheibub, ‘Mixed Systems and Democratic Performance’; Moestrup, ‘Semi-Presidentialism in Young Democracies’; Robert Elgie, ‘The Perils of Semi-Presidentialism: Are They Exaggerated?’ Democratization, 15 (2008), 49–66.

52 Cheibub, ‘Mixed Systems and Democratic Performance’.

53 Moestrup, ‘Semi-Presidentialism in Young Democracies’, at p. 40.

54 Elgie, ‘The Perils of Semi-Presidentialism’, p. 63.

55 Elgie, ‘The Perils of Semi-Presidentialism’, p. 64.

56 David J. Samuels and Matthew Soberg Shugart, ‘Presidentialism, Elections and Representation’, Journal of Theoretical Politics, 15 (2003), 33–60, p. 33.

57 Stephen Whitefield, ‘Mind the Representation Gap’, Comparative Political Studies, 39 (2006), 733–58, pp. 733–4.

58 Linz, ‘Presidential Versus Parliamentary Democracy’, at pp. 8, 14, 19–20; Linz and Stepan, Problems of Democratic Transition and Consolidation, pp. 278–9; Skach, Borrowing Constitutional Designs, p. 14.

59 Shugart and Carey, Presidents and Assemblies, pp. 274–5.

60 John D. Huber and G. Bingham Powell, ‘Congruence between Citizens and Policymakers in Two Visions of Liberal Democracy’, World Politics, 46 (1994), 291–326; James Stimson, ‘Party Government and Responsiveness’, in Adam Przeworski, Susan C. Stokes and Bernard Manin, eds, Democracy, Accountability, and Representation (Cambridge: Cambridge University Press, 1999), pp. 197–221; Whitefield, ‘Mind the Representation Gap’.

61 Kaare Strøm, Wolfgang C. Müller and Torbjörn Bergman, ‘Challenges to Parliamentary Democracy’, in Kaare Strøm, Wolfgang C. Müller and Torbjörn Bergman, eds, Delegation and Accountability in Parliamentary Democracies (Oxford: Oxford University Press, 2003), pp. 707–50, at p. 714.

62 Kaare Strøm, Wolfgang C. Müller, Torbjörn Bergman and Benjamin Nyblade, ‘Dimensions of Citizen Control’, in Strøm, Müller and Bergman, eds, Delegation and Accountability in Parliamentary Democracies, pp. 651–706, at p. 677.

63 Whitefield, ‘Mind the Representation Gap’, p. 753.

64 Bernard Manin, Susan C. Stokes and Adam Przeworski, ‘Elections and Representation’, in Adam Przeworski, Susan C. Stokes and Bernard Manin, eds, Democracy, Accountability, and Representation (Cambridge: Cambridge University Press, 1999), pp. 29–54.

65 Samuels and Shugart, ‘Presidentialism, Elections and Representation’, p. 39.

66 Samuels and Shugart, ‘Presidentialism, Elections and Representation’, p. 41.

67 Shugart and Carey, Presidents and Assemblies, p. 44.

68 Timothy Hellwig and David J. Samuels, ‘Electoral Accountability and the Variety of Democratic Regimes’, British Journal of Political Science, 38 (2008), 65–90, p. 68.

69 Linz and Stepan, Problems of Democratic Transition and Consolidation, p. 279.

70 Skach, Borrowing Constitutional Designs, p. 125.

71 Skach, Borrowing Constitutional Designs, p. 126.

72 Michael S. Lewis-Beck, ‘Who’s the Chef? Economic Voting under a Dual Executive’, European Journal of Political Research, 31 (1997), 315–25.

73 Michael Lewis-Beck and Richard Nadeau, ‘French Electoral Institutions and the Economic Vote’, Electoral Studies, 19 (2000), 171–82, p. 176.

74 Lewis-Beck and Nadeau, ‘French Electoral Institutions and the Economic Vote’, p. 181.

75 Hellwig and Samuels, ‘Electoral Accountability and the Variety of Democratic Regimes’, p. 78.

76 Maurice Duverger, ‘The Political System of the European Union’, European Journal of Political Research, 31 (1997), 137–46, p. 137; Pasquino, ‘Semi-Presidentialism’, p. 136; Sartori, Comparative Constitutional Engineering.

77 Linz, ‘Presidential Versus Parliamentary Democracy’, at p. 58; Linz and Stepan, Problems of Democratic Transition and Consolidation; Skach, Borrowing Constitutional Designs.

78 Linz, ‘Introduction’, at p. 11.

79 Skach, Borrowing Constitutional Designs, p. 124.

80 Shugart and Carey, Presidents and Assemblies, pp. 122–3.

81 Shugart and Carey, Presidents and Assemblies, p. 121.

82 Oleh Protsyk, ‘Prime Ministers’ Identity in Semi-Presidential Regimes: Constitutional Norms and Cabinet Formation Outcomes’, European Journal of Political Research, 44 (2005), 721–8.

83 Amorim Neto and Strøm, ‘Breaking the Parliamentary Chain of Delegation’.

84 Schleiter and Morgan-Jones, ‘Party Control over European Cabinets?’

85 Amorim Neto and Strøm, ‘Breaking the Parliamentary Chain of Delegation’, p. 643.

86 Cheibub, ‘Mixed Systems and Democratic Performance’, p. 7.

87 Cheibub, Presidentialism, Parliamentarism, and Democracy, p. 80.

88 Gary King, James E. Alt, Nancy Elizabeth Burns and Michael Laver, ‘A Unified Model of Cabinet Dissolution in Parliamentary Democracies’, American Journal of Political Science, 34 (1990), 846–71, p. 847.

89 King, Alt, Burns and Laver, ‘A Unified Model of Cabinet Dissolution in Parliamentary Democracies’; Paul V. Warwick, Government Survival in Parliamentary Democracies (Cambridge: Cambridge University Press, 1994).

90 Shugart and Carey, Presidents and Assemblies.

91 Steven D. Roper, ‘Are All Semipresidential Regimes the Same? A Comparison of Premier-Presidential Regimes’, Comparative Politics, 34 (2002), 253–72.

92 Kaare Strøm and Stephen M. Swindle, ‘Strategic Parliamentary Dissolution’, American Political Science Review, 96 (2002), 575–91.

93 Torsten Persson and Guido Tabellini, The Economic Effects of Constitutions (Cambridge, Mass.: MIT Press, 2005), p. 6.

94 Pasquino, ‘Semi-Presidentialism’; Sartori, Comparative Constitutional Engineering.

95 Linz, ‘Introduction’, p. 4; Skach, Borrowing Constitutional Designs; Skach, ‘The “Newest” Separation of Powers’, pp. 96–7.

96 Tsebelis, Veto Players.

97 Matthew Soberg Shugart, ‘Presidentialism, Parliamentarism, and the Provision of Collective Goods in Less-Developed Countries’, Constitutional Political Economy, 10 (1999), 53–88.

98 Oleh Protsyk, ‘Politics of Intraexecutive Conflict in Semipresidential Regimes in Eastern Europe’, East European Politics and Societies, 19 (2005), 135–60; Thomas Sedelius (doctoral dissertation, ‘The Tug-of-War between Presidents and Prime Ministers, Semi-Presidentialism in Central and Eastern Europe’ (Statsventenskapliga Institutionen, Orebro: Orebro University, 2006), pp. 127–78.

99 Cheibub, ‘Mixed Systems and Democratic Performance’.

100 Huber, ‘Restrictive Legislative Procedures in France and the United States’.

101 Huber, ‘Restrictive Legislative Procedures in France and the United States’, p. 680.

102 Paul Chaisty and Petra Schleiter, ‘Productive but Not Valued: The Russian State Duma, 1994–2001’, Europe-Asia Studies, 54 (2002), 701–724; Oleh Protsyk, ‘Cabinet Decision-Making in Ukraine: The Dual Executive and the Diffusion of Policy-Making Authority’, in Allan Rosenbaum and Juraj Nemec, eds, Democratic Governance in Central and East European Countries: Challenges and Responses for the 21st Century (Bratislava: NISPAcee, 2006); Schleiter and Morgan-Jones, ‘Russia: The Benefits and Perils of Presidential Leadership’.

103 Protsyk, ‘Cabinet Decision-Making in Ukraine’, at p. 19.

104 Shugart, ‘Presidentialism, Parliamentarism, and the Provision of Collective Goods in Less-Developed Countries’.

105 Oleh Protsyk, ‘Ruling with Decrees: Presidential Decree Making in Russia and Ukraine’, Europe-Asia Studies, 56 (2004), 637–60.

106 Schleiter and Morgan-Jones, ‘Russia: The Benefits and Perils of Presidential Leadership’.

107 Andrei Shleifer and Daniel Treisman, Without a Map: Political Tactics and Economic Reform in Russia (Cambridge, Mass.: MIT Press, 2000).