A Ballot Throne: does direct election matter for presidential power?

Part I: a discussion of Margit Tavits’ 2009 Presidents with Prime Ministers

From kings and archons to presidents and prime ministers, positions of leadership and authority form a crucial part of every state. Different states have institutionally different positions of executive and legislative authority and different rules structuring their respective powers and relations, known as executive type. In the study of these institutions in modern democracies, the different executive types have usually been categorised according to two main dimensions: constitutional origin, or how the executive is chosen, and constitutional survival: how the executive can be dismissed. My subject here is the part represented by origin. Effectively, the focus of this dimension is on whether or not the executive is popularly elected. A political system can go two ways: while presidentialism and semi-presidentialism feature a directly-elected executive leader, parliamentarism means the entire executive leadership is derived from the assembly or legislature.

In political science, popular election is widely seen as a crucial component of what has been labelled by some as the ‘Perils of Presidentialism’ (Linz 1990): having a directly-elected leader is argued to be perilous because, firstly, direct election means the executive has a separate origin from the assembly, so that the assembly and executive are far more likely to have clashing partisan and ideological complexions, and secondly, it confers a legitimacy and personal mandate on the leader, which is often held to exacerbate any actual clash between the executive and assembly over policy or appointments. Moreover, popular election of the chief executive is often argued to contribute to concentration of power in that position due to the increased legitimacy and perception of a personal policymaking mandate. The feared potential consequences of these effects range from poor governance outcomes such as policy deadlock, weaker parties, and poor clarity of responsibility for voters to an increased risk of democratic breakdown as a result of presidents that choose to overcome such deadlock by launching a self-coup (or, alternatively, a military intervention with a similar aim).

(Please note: for a more complete explanation of executive type the definitions of its categories, I recommend reading the accompanying blog post on executive type definitions). 

In her 2009 book, Presidents with Prime Ministers, Margit Tavits questions this line of argumentation from a perspective most of the literature neglects. If having an elected head of state makes such a big difference, especially as a result of the increased legitimacy, directly-elected presidents should tend to behave rather differently from indirectly-elected heads of state. Tavits’ contention is that they don’t. Not only can indirectly-elected presidents be every bit as partisan, activist, and political as elected ones, elected heads of state can also be very much the opposite. While Tavits’ keen analysis convincingly untethers important aspects of the scholarly narrative regarding executive type, there are two dimensions of the issue which she does not explore: the direction of political ambitions and parties’ nomination strategies. These latter dimensions are ultimately critical to understanding how direct election often does help structure equilibria which sharply contrast with most parliamentary regimes.

Tavits’ case against conventional models

To make her case that elected and unelected presidents behave similarly, Tavits provides evidence from parliamentary and semi-presidential republics. This case selection is an effective way of isolating the independent variable, as these types differ chiefly by one discontinuity: having a directly- or indirectly-elected head of state. While this is a by-definition difference between parliamentarism and semi-presidentialism, the debate over direct election tends to assume implicitly that this difference tends to correlate with (if not necessarily to cause) patterns of behaviour that structure the framework of authority in practice.

The conventional model of parliamentarism assumes a ‘figurehead’ heads of state. The role of a figurehead is essentially ceremonial; the ideal-typical figurehead uses his official powers only sparingly. Only when absolutely needed to mediate the choice and rejection of governments (executives) by the assembly might such a head of state intervene, and usually only to enforce the choice of the assembly (or to guide the system towards an early assembly election as a way of resolving a conflict). If the head of state has any formal veto or delay power, it is used rarely if ever.

A different model is assumed for semi-presidential systems. These presidents, being directly elected, are implicitly seen as political animals, with policy preferences and partisan interests expressed through strategic use of their powers in response to political incentives. The most important influence on those incentives is the partisan composition of the assembly and the coalition dynamics between the parties within it. If the assembly (and, therefore, the cabinet) is controlled by the president’s allies, the president would be expected to use his powers sparingly, because the president’s preferences align with the incumbent government, and because maintaining a united front is in the electoral interests of the president’s party. If, by contrast, the president faces an assembly controlled by opponents (a situation known as ‘cohabitation’), he would typically be expected to have an interest in undermining that government and trying to block their policy agenda, and to use his formal and informal powers with that aim in mind.

Tavits contends that, regardless of the mode of selection of the head of state, the latter model is usually more appropriate – i.e., it better explains behaviour in both semi-presidential and parliamentary systems.  Tavits convincingly shows that indirectly-elected heads of state use their powers and informal agenda-setting capacity according to incentives, dynamics and opportunities which are overwhelmingly political and partisan in nature. That is not to say that heads of state never act as ‘figureheads’. However, ‘figurehead’ patterns of behaviour are no more likely to result when the head of state is elected indirectly rather than directly.

Tavits’ evidence

The main thrust of the evidence Tavits presents comes in the form of seven case studies of parliamentary and semi-presidential countries in Europe. They show that the figurehead vs. political animal characterisation of heads of state in these systems does not fit the facts very well. Put simply, directly-elected presidents do not seem to be more active in the use of their powers. What kind of powers are we talking about? In almost all seven cases, presidents enjoy some version of each of the following formal powers: veto power (typically with override by an absolute majority of the assembly, although sometimes a higher threshold, making it a stronger veto) and the ability to refer a bill to judicial review before enactment, some role in the appointment of the government (typically including the power to appoint ministers on the prime minister’s advice, meaning that a president could refuse to appoint persons so nominated by the PM), some agenda powers (typically a formal power to propose legislation, although de facto all presidents have the informal power to make proposals through the media), and, lastly, often significant power in the formulation of foreign policy.

Indirect election did not necessarily produce apolitical figureheads. Tavits produces many examples of presidents elected by parliaments who actively involved themselves in politics to the extent that their powers allowed them to. Some refused to make some ministerial appointments suggested by prime ministers. Others made relatively frequent use of their power to delay legislation by vetoing it or referring it to the courts, which sometimes had the effect of entirely derailing the legislation in question. Many also insisted on using their constitutional prerogatives to not merely represent the government, but actually play a leading role in foreign policy. When were these powers most often used? Primarily when presidents came from different parties than their governments, or otherwise did not share their political preferences (in both semi-presidential and parliamentary systems, this most often happens as a result of the fact that the terms of the assembly and the executive don’t always overlap).

On the other hand, direct elections did not necessarily produce very political or very active heads of state. Often, presidential ‘activism’ was mainly a question of the president’s partisanship and the assembly’s partisan composition. Cohabitation was associated with activism, as was a highly fragmented assembly which made it difficult for the assembly to resist the president’s actions. Presidents whose party led the government or was part of it, by contrast, tended not to get as involved in daily politics. In some semi-presidential countries, such ‘figurehead’ behaviour was the rule rather than the exception. Austria, in particular, has been known for having highly apolitical presidents, despite their head of state having access to very strong powers, according to the country’s constitution. 

Interestingly, Tavits finds very few instances in which heads of state referred to any kind of personal mandate to justify their political actions. To the contrary – counterintuitively, some of those that did claim to ‘represent the people’ were actually not chosen in a direct election. Not only that, but Tavits gives multiple examples of indirect head of state elections (by national assemblies or special electoral colleges composed of national and subnational representatives) that were preceded by very public campaigns targeted at regular voters, who do not take part in these elections, though they were likely expected to hold some sway over their representatives in parliament).

The political importance and impact of indirectly-elected presidents is not just a matter of rhetoric. Tavits also shows that having one’s partisan in the office of head of state benefits a party in elections to the assembly, demonstrating that the same ‘coattails effect’ applies (and in similar measure) as in the case of directly-elected presidents. Consequently, it makes sense that parties would see the head of state position as a valuable political prize, and less surprising that both such a position and elections to it should often be quite politicised despite its relative powerlessness.

As a bit of an aside, there is one outcome on which Tavits does find a difference between parliamentary and semi-presidential executive types – election turnout. Holding direct presidential elections in addition to elections to the assembly reduces turnout in assembly elections by about 7%. And this is in countries where elections to the assembly are largely by far the most important type of election, given the assembly’s central role in making policy and choosing the government. So if there’s an outcome differential for constitutional designers to consider between direct and indirect election, it seems it isn’t really the conflict or power dynamics between the branches of government so much as the effect on voter fatigue. Certainly not an insignificant issue, but a far cry from the usual concerns of the literature.

Constitutional powers of presidents

Where does that leave us regarding the difference between executive types, on which so much ink has been spilt? Why might anyone have thought that indirect elections produce figureheads while direct elections make for active and influential political actors? Tavits suggests that one of the reasons is a selection effect: elected presidents are granted more powers. She shows this using a survey of constitutions and measures of presidential powers in different countries. Even among semi-presidential systems where the president has little control over the executive branch, the extent of formal power (in contexts such as government formation, the legislative process, military command and foreign affairs) is greater than comparable parliamentary regimes (where the head of state is by definition not elected by the voters). As a result of this pattern, directly-elected presidents (on average) may play a bigger, more central role in politics than their indirectly-elected counterparts – but this difference is the result of their relative powers, not the mode of election.

It seems to me that this analysis is probably accurate. Elected heads of state do tend to have more formal powers, and this fact may have coloured some of the perception of difference between semi-presidential and parliamentary systems. I do want to qualify this, since there are some clear exceptions. As I mentioned above, Austria stands out from Tavits’ own case studies as a particular example of a directly elected president with huge constitutional authority which has not been exercised much in practice.

In any case, even if the distribution of formal powers between elected and unelected heads of state have coloured their perception in political science, this would not necessarily obviate the importance of this distinction. Instead, it would change that distinction’s meaning and the causal process from direct elections to ‘presidential’ outcomes. Instead of the direct election changing the behaviour of heads of state (in particular, the extent of their ‘activism’), direct election influences the choices made by constitution framers and amenders regarding their powers. If there is an effect of the perceived legitimacy of direct election, then, it is not that direct election does not increase presidential power and influence directly, but instead that it plays some role in legitimising enhanced presidential power in constitutions that feature it. This effect need not be causal, at least not everywhere. There are probably cases where advocates of strong presidencies win out and also favour direct election as a means of strengthening such a president with a personal mandate.

The partisan dimension: political opportunities and incentives

Theorising about the origins of the confluence between presidential powers and direct election is important, and merits further study, but there is another factor which I think is far more important for understanding presidential power: the partisan dimension. Tavits clearly addresses one part of this, and the role of parties and partisanship is certainly central to her argument. In her case studies, Tavits finds that (regardless of how they had been elected) the extent of presidential activism depends in large part on an incumbent’s opportunities and incentives. The current shape of the party system (especially in the assembly), the president’s own partisanship, as well as where that puts him in relation to the parties, all interact to structure presidential opportunities and incentives. To recap, Tavits argues that situations of cohabitation (or a highly fragmented assembly) produce more presidential activism than situations approximating ‘unified government’. Her data suggest that this pattern holds regardless of how the president is chosen. These dynamics more or less summarise Tavits’ argument. But while they seem to hold fairly consistently across most European cases, there is a big fly in the ointment: the French Fifth Republic.

Unlike all the cases Tavits investigates, French presidential elections are not second-order elections – they are the main show in French politics. Moreover, French presidents have been far more powerful than in any of the cases studied by Tavits – probably more powerful than any head of state in the EU. Apart from being a major country in Europe, why is France so significant? It is perhaps the most prominent and well-known semi-presidential system – indeed, the term ‘semi-presidential’ was itself coined in reference to the French Fifth Republic, as was ‘cohabitation’.

Moreover, the terms of partisan dynamics and their influence on presidential incentives are very different in France from what Tavits would have us expect. In France, cohabitation isn’t known as a period when the president is more active, and certainly not more powerful, than periods of unified government – quite the opposite. This is because, when the president’s party is in the majority, the French president doesn’t fall into the background, but instead is the government’s main spokesman and agenda-setter. Neither has the assembly been very fragmented. Since the Fifth republic’s foundation in 1958, the National Assembly has mainly seen fairly solid single-party or coalition governments, usually supporting the president (at a few other times opposing him and forcing him to ‘cohabit’ with an opposing prime minister). So what explains these outcomes? Recalling the other part of Tavits’ argument, one might suggest they result from presidential powers under the French Constitution. But this is unlikely to be the answer, since the formal powers of the French President are about middle of the range among the cases Tavits studies.

Instead, to return to a point I made above, one answer is that activism and power are two different things. Tavits’ focus on activism may be a useful exercise, but it does not tell us much about the relative power of elected and unelected presidents, nor about their relative power given different situations of partisan control. In France, but also in the cases Tavits brings to bear, cohabitation is when the head of state is most constrained in these systems of government: the forces opposed to the president and/or his party control the assembly, and thereby typically also the cabinet and all the most important channels of policymaking. Under such circumstances, the president may have more reasons to use his powers actively, but it is not because the president is in a position of power (relative to a unified government).

In Tavits’ case studies, if the president faces a politically allied or friendly assembly, the president becomes less ‘active’. Does that mean such situations leave presidents less powerful? Not necessarily. On the one hand, it is hard to evaluate the power or influence of such heads of state, because much of it can easily take place behind the scenes. On the other hand, France’s presidency demonstrates that such circumstances allow for very overt exercise of power, as well, and everything suggests that French presidents are indeed very powerful in those situations, setting the legislative agenda through the prime minister and cabinet, who are very much his agents.

Why are French governments during unified government the agents of the president rather than the assembly? Partisan control is incredibly important, but it obviously can’t explain the contrast between French cases of unified control and similar cases in the rest of Europe. Instead, there is a missing element. The difference is this: in France, the president is not a secondary politician within his party, but its leader. This is the other partisan dimension: whereas, in many semi-presidential systems, party leaders typically direct their ambition towards the prime ministership rather than the presidency, French parties put forward candidates for the presidential contest which, if they win, become party leader.

Conclusion: the question of ambition

Returning to our initial theme, there are different positions of authority in government. The mode of election of leaders is one of the main dimensions which have been used to explain differing distributions of power and to distinguish different executive ‘types’. Tavits’ analysis questions this choice, comparing the behaviour of elected and unelected leaders and finding no clear pattern correlating with mode of election. As I have suggested, her findings may partly reflect an imperfect dependent variable (‘activism’). They also reflect her case selection. But there’s something more profound here than case selection. I want to highlight this profound question, which Tavits does not ask: why do politicians and parties’ ambitions focus on one position rather than another?

These ambitions ultimately structure the wider political game. Tavits is absolutely right that the rules of the game can be different (e.g. direct election of the head of state) without much difference in outcome. But outcomes can also differ among countries with the same basic rules. Semi-presidential constitutions can produce both president-centric and prime minister-centric power distributions. As I will show in my next post, pondering why this is the case (and also why, for example, this does not seem to be the case in parliamentary systems) easily leads us back to institutions, including the institution of direct election. And if it’s true that direct election affects politicians’ ambitions, well, they should matter to us.