Varieties of Presidentialism & of Leadership Outcomes

What is the relationship between institutional power and political leadership? What is the effect of presidential institutions on political, economic, and social outcomes? These are questions that scholars have debated for centuries. For many years, proponents of U.S.-style presidentialism were pitted against supporters of U.K.-style parliamentarism. Here, there were seminal contributions from Woodrow Wilson and Walter Bagehot in the late nineteenth century,¹ as well as an important exchange between Harold Laski and Don Price in the mid-twentieth century.² In the late 1980s and early 1990s, this debate was revived with the creation of many newly independent states and a wave of democratization. At this time, many scholars warned against what they saw as the potentially negative consequences of presidential leadership on young democracies. Over the last few decades, though, the terms of the debate have broadened and changed. In particular, scholars have begun to consider the pros and cons of what are known as semipresidential regimes. Scholars have also started to look at institutional variation within presidential and semipresidential regimes. This work has suggested that successful presidential leadership is a realistic possibility for countries in the process of democratic consolidation.

When scholars debate the pros and cons of different political regimes, the two baseline categories are well known. Presidential regimes have both a popularly elected, fixed-term president and a fixed-term legislature. The president nominates members of the Cabinet subject to legislative approval, but the government collectively cannot be dismissed by the legislature. By contrast, parliamentary regimes are headed either by a figurehead monarch or by a weak, indirectly elected president, who is often selected by the legislature. The legislature approves the choice of the prime minister, who is the central figure within the executive and who individually selects the members of the Cabinet. The prime minister and Cabinet, though, remain collectively responsible to the legislature. In presidential and parliamentary regimes, the institutional choices are very stark. For example, should there be a popularly elected president? Should the government be collectively accountable to the legislature? There has been a long scholarly debate about the effects of these different choices, particularly on the fate of young democracies.

In 1970, the French political scientist Maurice Duverger challenged the standard presidential/parliamentary dichotomy. He identified another type of institutional arrangement that he labeled semipresidential regimes.³ These are countries where there is both a popularly elected fixed-term president and a prime minister and Cabinet that are collectively responsible to the legislature. When Duverger first identified them, there were only a handful of semipresidential regimes in existence, most notably France. Now, though, this type of constitutional arrangement is much more common. Indeed, in the late 1980s, there were fewer than ten countries with a semipresidential constitution; now there are more than fifty.⁴ While Duverger's general label has persisted, scholars have further distinguished between two types of semipresidential regimes. These have the unwieldy names of president-parliamentary and premier-presidential regimes.⁵ Both have the basic features of semipresidentialism, but under president-parliamentarism, the government is responsible to both the legislature and the president, whereas under premier-presidentialism, the government is responsible solely to the legislature. The list of president-parliamentary countries includes Kyrgyzstan, Mozambique, Namibia, Russia, Sri Lanka, and Taiwan. The list of premier-presidential countries includes France, Georgia, Lithuania, Mongolia, Romania, and Turkey. It is safe to say that while scholars previously tended to confine their analyses of presidential leadership to presidential countries, they now invariably include consideration of semipresidential countries and, indeed, presidential leadership within the two subtypes of semipresidential regimes.

Since the wave of democratization in the late 1980s and early 1990s, scholars have debated the relative merits of these regime types in relation to the transition from authoritarianism to democratic consolidation. Here, there has been considerable focus on whether presidential institutions affect the likelihood of democratic consolidation. At the very beginning of the 1990s, Juan Linz framed the terms of the debate in this regard.⁶ He argued that presidentialism was a perilous choice for young democracies. At first glance, this recommendation seems highly contentious, because it flies in the face of the U.S. experience. After all, as the world's oldest presidential regime and also the world's oldest constitutional democracy, it might be tempting to conclude that presidentialism is well suited to democratic consolidation. However, Linz drew heavily on the experience of presidentialism in Latin America, where, at the time he was writing, democracy had not yet taken firm root. Subsequently, Scott Mainwaring built on Linz's work, arguing that the interaction of presidentialism and a multiparty system was a dangerous combination for young democracies.⁷ Again relying primarily on Latin America, he claimed that the difficulty involved in coalition-building in multiparty presidential systems could threaten the survival of such democracies.

By contrast, Matthew Shugart and John Carey argued that the popular election of the president was not necessarily problematic.⁸ Drawing on the distinction between president-parliamentary and premier-presidential regimes, they argued that the former should be avoided, but that there were merits to the latter because premier-presidentialism allowed a degree of presidential leadership, while also constraining it within certain limits. By and large, though, Shugart and Carey's recommendation was overlooked because premier-presidentialism can also exhibit what is known as cohabitation. This is where the presidency is supported by one political force and the legislature is controlled by an opposing force. This is similar to divided government in the United States. The difference is that in a premier-presidential system, the prime minister and the government are also independent of the president because they have the support of the legislature. The potential for conflict within the executive, and not merely between the president and the legislature, was usually enough for constitution-builders to recommend against premier-presidentialism and, indeed, semipresidentialism in general. On the strength of these debates, by the late 1990s, there was agreement that presidential leadership was likely to be problematic for new democracies.

In the background of this debate was the issue of political leadership itself. The institutional architecture of presidential and president-parliamentary regimes seemed to render the choices made by individual political leaders highly consequential. Were young democracies safe in the hands of the people who headed such regimes? Could they be trusted to exercise benign, never mind beneficial leadership over their country? Would leaders who were brought up under nondemocratic regimes have the requisite skills to exercise leadership safely, even if they wanted to? Scholars calculated that, on balance, it was more risky to introduce a presidential system in which idiosyncratic and potentially unpredictable leaders could exercise personal leadership than to establish a parliamentary system in which the prime minister was checked by party politics. They were also skeptical that placing checks upon presidential leadership would make a positive difference. After all, history showed that, outside the United States, frustrated presidents had a habit of calling in the military and/or ruling by decree, and even here, Watergate was still fresh in the collective memory.

This scholarly consensus against presidentialism had and continues to have concrete practical application. Many newly independent countries adopted their first ever constitution in the early 1990s. Meanwhile, other countries embarked upon a process of major constitutional reform. More than that, Tom Ginsburg, Zachary Elkins, and James Melton have shown that constitutional amendment is an ongoing process in many countries.⁹ Whatever the motivation, one of the issues that constitution-builders invariably have to address is how to organize both the executive and executive-legislative relations. In short, they have to make a basic choice among presidential, semipresidential, and parliamentary systems. Previously, constitution-builders tended to call upon individual experts to guide their choice. Over time, international organizations, such as the International Institute for Democracy and Electoral Assistance, have emerged with the expertise to provide general resources for constitution-builders across the world. The academic consensus against presidentialism has been influential in this context.

With the exception of Latin America, where there is a very long history of this institutional arrangement, and some parts of Anglophone Africa, where personalistic leaders found it convenient to centralize authority, most countries opted against presidentialism during the most recent wave of democratization. In Afghanistan, local constitution-builders did eventually opt for a presidential system, even though they were strongly warned against it by international advisers, including U.S. academics. The puzzle is why semipresidentialism has become so popular in recent decades. In general, constitutional experts warn against this form of government: it can generate strong presidential leadership when the president is backed by a secure legislative majority, but it also has the potential to generate confused lines of political authority between the president, prime minister, and legislature, which can result in cohabitation. The scholarly decks are truly stacked against this type of system. Even so, semipresidentialism has often suited local decision-makers. It provides a neat compromise between political forces that want presidentialism, usually because they calculate that their party will win the presidency, and those that want parliamentarism, usually because they believe that they are not strong enough to win the presidency, but stand a chance of entering a coalition government, thereby sharing in executive power. Mainly for this reason, semipresidentialism has emerged as the regime of choice for many young democracies. In this sense, contrary to the advice of much of the scholarly community, the opportunity for presidential leadership has spread around the world.

Nearly three decades on, though, the question is whether there is empirical evidence to support the academic consensus against presidential leadership. Here, the situation is much more confused. Initially, scholars such as Linz drew on in-depth regional knowledge to back up their arguments against presidentialism. They pointed to particular examples to show that democracy had collapsed in presidential countries. Other scholars, though, identified counter-examples in which presidentialism had survived. The same point applies to semipresidentialism. This system is said to have suited countries like Mongolia, but critics have pointed to the problems that it created in other countries, such as Niger. These examples merely show that the relationship between presidentialism and democratic collapse is not deterministic. The question, then, is whether presidential leadership increases the probability of democratic collapse. Here, the evidence is contradictory. The results are highly sensitive to the sample of countries under investigation, to the controls that are included in the equation, and to the statistical model that is used in the estimation. For example, based on a sample of 123 democratizations from 1960 to 2004, Ethan Kapstein and Nathan Converse found that parliamentarism was more dangerous for democracy than presidentialism.¹⁰ Taeko Hiroi and Sawa Omori, looking at 131 democracies from 1960 to 2006, also concluded that parliamentarism was more perilous than presidentialism.¹¹ By contrast, on the basis of 135 democratic periods from 1800 to 2004, Ko Maeda discovered that parliamentarism was better than presidentialism over the long run.¹² The value of the whole debate was called into question when Ming Sing, with a sample of 85 countries from 1946 to 2002, unearthed no relationship between either form of government and the collapse of democracy.¹³ For his part, José Cheibub argued that presidentialism is dangerous only if it is adopted in countries that were previously subject to military rule, effectively shifting the explanatory focus to the importance of history and away from institutions altogether.¹⁴

Overall, after a quarter-century of systematic study, it is still unclear whether presidentialism is more dangerous for young democracies than parliamentarism, or whether the choice of regime makes no difference. Indeed, the same can be said about semi-presidentialism. There is some supporting evidence for Shugart and Carey's claim about the perils of president-parliamentarism relative to premier-presidentialism,¹⁵ but there is no reliable, replicable evidence that premier-presidentialism is either more or less likely to be associated with democratic collapse than either presidentialism or parliamentarism. Thus, after so many studies, in terms of the empirical evidence at least, the jury is still out. In one sense, this is unsurprising. The success or failure of democracy is conditional upon many different factors. Context matters. More than that, the scholarly debate about presidentialism assumes it is an exogenous factor affecting leadership outcomes. Yet it is endogenous too. As Archie Brown points out in relation to semipresidentialism, there is a “chicken-and-egg question about whether leaders and political elites in countries with a tradition of authoritarian rule opt for a strongly presidentialized semi-presidentialism, leading to an excessive concentration of power in the hands of the chief executive.”¹⁶ Faced with such issues, scholars have started to unpack presidential and semipresidential institutions, and bring some consideration of both context and leadership back in.

One of the reasons why arguments about the supposed perils of presidential leadership have been so difficult to resolve relates to the concepts under investigation. The presidential system in the United States is so familiar that it is tempting to think only in terms of its example when discussing presidential leadership. Yet there is great variety of presidential leadership even within purely presidential regimes. For example, in Latin America, where the U.S. presidential model has been adopted wholesale, there are currently old-style caudillo presidents in Venezuela and Bolivia, whereas in other countries, including Ecuador, hostile congresses have resorted to presidential impeachment to divest themselves of unpopular political leaders. Indeed, some authors have seen such presidential “interruptions” as an increasingly common way of resolving political crises in the region.¹⁷ In Africa, where there is also a tradition of presidentialism, the “strong man” president is still a model of reference, though in some African countries, including Nigeria, presidents have recently struggled to assert their authority over the legislature. In Asia, too, there is considerable variation: in the Philippines, there is a form of hyperpresidentialism,¹⁸ while in South Korea and currently in Indonesia, presidents have had difficulty in passing their reform agenda through a divided legislature. What this all suggests is that while the United States may be the archetypal presidential regime, presidential leadership can take many forms. This is perhaps one reason why it has been difficult to identify a general association between presidentialism and democratic performance.

This point applies even more forcefully to presidential leadership under semipresidentialism. Here, scholars are obliged right from the start to make a basic distinction within semipresidentialism, typically between president-parliamentarism and premier-presidentialism. The difference between the two subtypes is purely constitutional, relating to whether or not the president has the power to dismiss the prime minister and government. While this is a constitutional distinction, it maps quite nicely on to the overall power of presidents in practice. Presidents in president-parliamentary systems do tend to be stronger than presidents in premier-presidential systems. Even so, the distinction masks important variation within each subtype. For example, Austria and Iceland both have purely ceremonial presidents, yet their presidents enjoy the constitutional power to dismiss the Cabinet. It is simply that, by convention, this power is never used. So they are president-parliamentary in constitutional terms, but their presidents act like figurehead presidents in parliamentary regimes.

There are also differences in presidential leadership between president-parliamentary countries such as Mozambique and Namibia, on the one hand, and Taiwan, on the other. In the former, presidents have been backed by dominant parties with a cohesive majority in the legislature. This ensures that the president enjoys the political resources to exercise great power. By contrast, in the latter, President Chen Shui-bian was without a legislative majority for much of his eight-year term, limiting his power considerably. There is also the well-known case of president-parliamentary Russia. When President Putin was term-limited in 2008, he simply moved to the premiership, dominating the system from there before returning to the presidency in 2012. Putin's example shows the limitations of an analysis focused solely on a consideration of basic separation-of-powers features.

Under premier-presidentialism there is also considerable variation in presidential leadership. The list of premier-presidential countries includes a number of cases in which the presidency is also a purely ceremonial office, even if the president is directly elected. In these countries, which include Croatia, Finland, Ireland, Macedonia, and Slovenia, the practice of politics is purely parliamentary. By contrast, in countries such as France, Turkey, and Ukraine, the president is usually the most important actor in the system. What is more, there is also considerable variation across time within individual premier-presidential countries. David Samuels and Matthew Shugart have shown that cohabitation is almost unheard of in president-parliamentary systems.¹⁹ However, it is not uncommon in premier-presidential systems. Here, when the legislative majority is opposed to the president, the president usually loses the opportunity to exercise leadership. For example, in France, power shifted from the president to the prime minister during all three periods of cohabitation (1986–1988, 1993–1995, and 1997–2002). In Romania, President Băsescu faced two periods of cohabitation (2007–2008 and 2012–2014). On both occasions, the legislative majority voted to suspend him from office and organized a popular referendum to decide whether or not he should be impeached. On both occasions, President Băsescu survived politically, but it was a sign that presidential leadership under premier-presidentialism cannot be taken for granted. In Portugal, by contrast, the opposite scenario occurs during cohabitation: the president becomes stronger. The Portuguese president is not the party leader. Instead, when the president and prime minister are from the same party, the latter has more party authority than the former. During cohabitation, though, the president is now the most senior party figure remaining within the executive. As a result, party opposition to the government expresses itself through the presidency, which becomes more powerful.

What this all suggests is that a typology of regimes based on institutional design is a relatively blunt conceptual instrument. It may be possible to identify some very general trends about presidential leadership under presidentialism compared with parliamentarism, and also, adjusting for countries such as Austria and Iceland, about the effect of president-parliamentarism relative to premier-presidentialism. Even if this were the case, though–and so far, as has been shown, the evidence is contradictory–it would still miss much of the effect of within-regime variation. Partly for these reasons, the research agenda has started to shift toward new questions about institutions and presidential leadership.

Some scholars have chosen to focus on the study of presidential power more specifically. For example, in a recent survey of articles in top-rated political science journals, David Doyle and I identified forty-nine studies that included an estimation of presidential power.²⁰ In forty-five of these, presidential power was operationalized as an explanatory variable, and in thirty of these forty-five studies, presidential power was confirmed to have had a significant effect on the outcome under investigation. What is more, the outcome of interest varied considerably across the set of studies. Some authors were indeed concerned with democratic consolidation. However, other scholars were interested in the relationship between presidential power and outcomes such as economic reform, economic growth, the level and timing of privatization, protectionism, corruption, human rights violations, Cabinet stability, ministerial portfolio allocation, the effective number of political parties, and voter turnout. Most of these studies were conducted in the last few years, suggesting that there is an increasing interest in the effect of presidential power.

This approach is consistent with some of the underlying logic of the more traditional regime-based inquiry. In general, presidents in presidential systems and, indeed, in president-parliamentary systems are stronger than their counterparts in premier-presidential regimes, who, in turn, are stronger than their head-of-state equivalents in parliamentary systems. In other words, regime-oriented studies can already be interpreted as studies of the relative impact of presidential power. However, as has been noted, there is considerable variation in presidential power within each of these regime types. Therefore, if presidential power really is the variable of interest, then scholars have argued that it needs to be operationalized much more carefully than is possible in regime-based analysis.

The question arises, then, as to how presidential power is best measured. Typically, measures are based on a set of individual indicators, such as whether a president has the power to issue decrees with the force of law. If a president enjoys a particular power, then a value of one may be assigned for that indicator. Otherwise, a value of zero is recorded. The total score for presidential power is invariably the simple aggregate of the scores for each indicator. This generates a set of cross-national presidential power scores for individual countries. This methodology, though, begs some important questions. What powers should be included in the set of indicators? Here, scholars make very different decisions. Some prefer a relatively small number of indicators, others include up to forty. Moreover, even if they include a similar number of indicators, they do not necessarily include exactly the same ones. In addition, whatever indicators are chosen, are the values assigned in each case determined by the wording of the constitution or by presidential leadership in practice? The Austrian and Icelandic cases demonstrate clearly that constitutions can sometimes be an imperfect guide to political life in reality. So there are serious concerns about the reliability of the measurement of presidential power. These issues have led some observers to question the validity of the exercise altogether.²¹ Recently, though, Doyle and I have tried to maximize the reliability of presidential power measures by, in effect, pooling the scores of scholars who have already come up with such measures.²² We drew upon the mass of information contained in twenty-eight existing presidential power measures, reducing them to a single score for each country's president. We adjusted for measures that appeared to produce idiosyncratic scores for particular countries and we provided some information as to whether there was general scholarly agreement on the presidential power score for any given country.

This exercise suggests that there is now the opportunity to engage in the study of the impact of presidential power more reliably than was previously the case. All the same, it leaves open the difficult issue of within-country variation over time. For example, on a scale from zero to one, the French president has a normalized presidential power score of 0.465.²³ This figure is in the right ballpark intuitively. There are plenty of executive presidents in Latin America with higher scores, as well as plenty of figurehead presidents with much lower scores. However, as has been indicated, presidential power has varied over time within France in the context of cohabitation. A single country score cannot capture this variation. Moreover, these scores are based on twenty-eight measures of the constitutional powers of presidents. This is a more reliable foundation upon which to base a study in the sense that no in-depth country knowledge is required. It is simply a matter of reading the words in a country's publicly available constitution. That said, it does leave the issue of the difference between constitutional powers and actual presidential leadership still unresolved.

More fundamentally, though, such presidential power scores can never capture the individual quality of political leadership. Working within the same institutional framework in the same country, political leaders can exercise leadership very differently. This personal element of political leadership is very difficult to capture. In fact, comparative scholars who engage in large-n statistical studies do not attempt to do so. For them, the impact of political leadership can be found somewhere in the error term of the equation. This is a natural consequence of the type of analysis in which they are engaging. They wish to make general statements about the impact of certain explanatory factors. Institutions, whether operationalized as different separation-of-powers regimes or presidential power scores, can be manipulated and the effect of institutional variation on various outcomes can be tested. By contrast, variation in individual political leadership cannot be investigated in the same way. This does not mean that political leadership does not matter. On the contrary, whether good or bad, competent or incompetent, honest or corrupt, political leadership will always make a difference in particular contexts. However, the study of both regime types and presidential power scores does not place the focus on such questions. Thus, while there is a growing interest in estimating the general effect of presidential power, and while there are now measures of presidential power that are more reliable than ever before, there are nonetheless still profound limitations to this exercise.

Partly in response to these issues, some scholars have placed the emphasis on the importance of more particular aspects of presidential leadership. This work addresses head-on one of the issues that has bedeviled regime-based inquiry. Why is there variation in outcomes within particular regime types? Specifically, why is presidential leadership more successful in some presidential countries than others? This question was first asked in relation to Latin America, where most countries have a presidential system. In this sense, the broad institutional context is constant. At the same time, and in contrast to the two-party polarization in the United States, most Latin American countries also have multiparty systems. Therefore, presidents often come to power without the backing of majority support in the legislature: they are minority presidents. Indeed, this was the difficult combination for democratic consolidation that Scott Mainwaring identified in the mid-1990s. More than that, and in this regard there are similarities with the United States, political parties in Latin America often lack cohesion in the legislature. Parties are not loyal to the president. There is party switching: deputies shift their allegiance from one party or coalition to another. In other words, even if the president has the nominal support of a particular party or coalition, such support cannot be taken for granted.

In this context, why have some minority presidents been more successful than others? For example, in Brazil, both President Fernando Henrique Cardoso and President Lula were able to pass reforms through Congress, even though their own party did not have a majority there. By contrast, in Ecuador, presidents have been stifled in their ambitions, with President Abdalá Bucaram even being dismissed from office by Congress in 1997 on the grounds that he was mentally unfit to rule. To put it another way, why have some presidents been more successful at building legislative coalitions than others? The attempt to find an answer to this question has generated a literature on so-called “coalitional presidentialism.”

The work on coalitional presidentialism (or presidencialismo de coalizão) has roots in the study of Brazil. Here, there were repeated periods of democracy followed by democratic collapse. However, since democracy was reinstated in the late 1980s, it has survived. At least in part, this success has been put down to the success of presidential coalition-building. According to Timothy Power, the “core insight of coalitional presidentialism is that presidents must behave like European prime ministers. Executives must fashion multiparty cabinets and voting blocs on the floor of the legislature.”²⁴ In this regard, President Cardoso wrote what has been described as a user's manual for other presidents to follow.²⁵

While the study of coalitional presidentialism is rooted in the Brazilian experience, it has struck a chord with scholars of the region generally. For example, Carlos Pereira and Marcus André Melo argue that the success of coalitional presidentialism can be attributed to three factors: 1) whether the president is constitutionally strong; 2) whether the president has “goods” to trade in order to attract and keep coalition partners; and 3) whether there are institutionalized and effective checks on presidential actions.²⁶ For them, it is important that presidents have the constitutional power to distribute political goods, such as Cabinet posts and budgetary resources. Presidents can use these goods in the form of “selective incentives” to reward and/or punish members of the legislature. In a form of politics that would be familiar to U.S. observers, coalitional presidentialism relies on the president's ability to distribute “pork” to members of Congress. At the same time, though, it is also important for there to be checks on the president's power, including an active and independent judiciary and a plural media. For his part, Steven Levitsky emphasizes a slightly different combination of factors to explain the success of presidential coalition-building.²⁷ For him, the three important aspects are 1) the sharing of executive power through the distribution of Cabinet seats to coalition parties; 2) pork, budgetary clientelism, and other discretionary side payments; and 3) the presence of oversized coalitions to compensate for the lack of party cohesion.²⁸ Levitsky thinks of these factors as informal institutional rules, taking the focus of the analysis even further away from the regime-based inquiry of the early 1990s.

The work on coalition presidentialism in Latin America has proved popular because it is potentially transferable to the study of presidential leadership in other regions. This has led to an interest in the so-called “executive toolbox,” or “presidential toolkit” approach.²⁹ For example, Paul Chaisty, Nic Cheeseman, and Timothy Power have extended the logic of coalitional presidentialism to countries in Africa and in the former Soviet Union. They argue that presidents have a range of tools that they can draw upon to engage in successful coalition-building, and that the particular tools they use will vary according to the local context. Specifically, they identify five key tools for constructing legislative coalitions: agenda power, budgetary authority, Cabinet management, partisan powers, and informal institutions, though they acknowledge that other tools might be appropriate in other contexts still. For example, they show that many African presidents have failed to command the support of a natural majority in the legislature. Faced with this problem and citing Benin as an example, they show how presidents there have had little choice but “to engage in complex processes of alliance formation, appointing representatives of opposition parties to the cabinet.”³⁰ This has meant, though, that presidents in Benin have been constrained in their exercise of power. For example, they have not always been able to monopolize control over economic rents and public policy. Instead, like President Kibaki in Kenya, presidents in Benin have “blended cabinet management, informal institutions, and agenda power into a single coherent strategy for coalition management.”³¹ Indeed, this example shows how the presidential toolkit does not simply manage itself. Skillful leaders have to decide on a strategy for manipulating it successfully. This opens up a space for the study of innovative and resourceful political leadership. It is reasonable to speculate that such leadership is in fact one of the reasons why Benin has had one of the more successful democratic experiments in Africa since the early 1990s. In other words, even though Benin has a presidential regime and presidents have lacked solid support in the legislature, the judicious use of tools in the presidential toolkit by successive leaders has perhaps helped maintain broad support for the regime, not least by bringing potentially oppositional forces into the decision-making process.

How should we sum up the long debate about the relative benefits of presidentialism and parliamentarism? Over the years, this simple distinction has become less relevant, first with the rise of semipresidential regimes across the world, and then with the scholarly focus on intraregime variation and the study of coalitional presidentialism, as well as the presidential toolkit. The development of this scholarship is important not least because it indicates the need to go beyond the standard archetype of presidential leadership in the United States. Looking to Latin America for lessons about presidential leadership, Juan Linz argued that a key problem with presidentialism was the potential for conflict between presidents who failed to enjoy majority support in the legislature and the legislature itself. This was exactly the type of scenario that he believed was likely either to lead to the intervention of the military in an attempt to restore stability to the regime (the golpe), or to see presidents abusing the rule of law and governing by decree (the autogolpe).

However, following on from work pointing out that coalitions are, in fact, relatively common in presidential regimes,³² the literature on coalitional presidentialism and the presidential toolkit has provided an explanation as to why presidential leadership in Latin America and elsewhere has been less destructive of democracy in recent times. Specifically, it has done so by shifting the emphasis away from blunt, regime-based inquiry, and, instead, has unpacked the concept of presidential leadership. This work is at once both consistent with and neglectful of the study of individual political leadership. Underlying the arguments about the perils of presidentialism was a distrust of individual leadership, or at least a skepticism that benign leadership was likely to be exercised in presidential regimes. At the same time, the debate about the relative effects of institutional structures on outcomes, including the debate about the effects of variation in presidential power generally, has been conducted largely without reference to leaders or leadership. There are signs, though, that the most recent scholarship is trying to address this issue more directly (and yet still systematically). The logic of the presidential toolkit approach is that presidents have to choose which tools are best suited to the specific context they face. Some presidents are likely to choose well and others less well. Here, in the interaction of institutions, leaders, and context, lies the eternal dilemma of the study of presidential leadership.