The Comparative Constitutional Law of Presidential Impeachment

“Impeach Eisenhower. Impeach Nixon. Impeach Lyndon Johnson. Impeach Ronald Reagan.”

–Annie Hall (Charles H. Joffe, 1977)

Introduction

The president must go! Thus rings the call across many democracies, including our own. Political opposition and civil society movements have targeted elected leaders who have become politically unpopular, ineffective, or allegedly (and perhaps actually) corrupt. Impeachment discussions surged in the United States in 2016 even before President Donald Trump had taken his oath of office.1 They burst dramatically into the realm of political plausibility in September 2019 with the announcement of an inquiry in the House of Representatives, which precipitated the third presidential impeachment and Senate trial in U.S. history. Yet this specter of removal has not been distinctive to Trump. Impeachment talk also dogged his predecessors.2 Nor should Americans think their discontents unique. In France, the gilets jaunes protest movement has been candid in its “hatred” for President Emmanuel Macron and its desire to see him ousted from office.3 And in Venezuela, an opposition leader went so far as to declare himself “interim president” in a (so far, vain) attempt to accelerate the departure of a well-entrenched presidential incumbent.4 Regime change has yet to arrive in Caracas, Paris, or Washington, D.C. But presidents have no cause to rest easy. In democracies as diverse as Brazil, South Korea, and South Africa, presidents have been removed in the middle of their term in the past decade.5 Impeachment talk is not necessarily idle chatter. At least in some instances, it is a credible position that can attract sufficient political and popular support to be realized.

The removal of a president from office by a mechanism other than through the regular operation of elections, term limits, and the normal apparatus of political selection goes to the core of democratic governance. This is a moment of increasing popular discontent with established regimes, coupled with a growing polarization within the voting publics of many democracies.6 Under those conditions, it is perhaps unsurprising that elected tenures would prove to be fragile, and talk of preemptive removal and impeachment endemic.

Nevertheless, legal scholars and social scientists have until now lagged behind the roiling wave of popular sentiment. To be sure, there is a wealth of scholarship on the role of impeachment in the U.S. Constitution.7 That work—much of it excellent—starts from the Framers’ design, and then reasons from that design to present applications.8 As a result, it explores a relatively narrow compass within the space of possible constitutional design. It does not help that the “Constitution is surprisingly opaque as to how apex criminality should be addressed.”9 The U.S. Constitution’s text, for example, uses the ambiguous term “high Crimes and Misdemeanors”10 to define a threshold for presidential removal. It is no surprise that Trump, like his predecessors, insisted that this included only statutorily defined crimes.11 The Constitution also fails to specify a standard of proof for either impeachment or conviction. Again, it is no surprise that both the President’s defenders and his prosecutors each have asserted their own favored substantive standards of impeachability.12 Finally, the text conspicuously fails to specify clearly whether a sitting president can be indicted prior to the completion of impeachment proceedings.13 The result is a process of deeply uncertain scope and consequences.14 Arguments about many of these uncertainties—not just in the context of the Trump impeachment, but beyond—necessarily hinge on predictions about the consequences of presidential ouster.

With many other constitutional questions, our post-ratification history can provide clarity about consequences. Not so here. To date, there have been only three successful presidential impeachments; no sitting president has ever been removed.15 We thus simply have no basis for knowing whether impeachments tend to shore up democracy, or whether they undermine it.16 The impeachment language in Articles I and II is largely (if not wholly) general, extending beyond presidents to encompass judges and certain officials.17 But the history of nonpresidential removals is also of limited use. Presidential impeachments plainly raise empirical questions, legal problems, and normative concerns beyond those implicated by the removal of federal judges and other officials. Most obviously, the electoral mandate that presidents, unlike unelected actors, possess raises a distinctive question about the democratic legitimacy of impeachment-like removal mechanisms, such as criminal prosecution or declarations of incapacity, that bypass the people.18 There is a distinct and pressing question whether impeachment is consistent with the principle of popular sovereignty that underwrites democracy—or whether it is at odds with democracy as a going concern.

One analytic pathway, however, remains relatively uncharted. At the same time as the focus of U.S. scholars narrows, there remains a dearth of legal scholarship leveraging other countries’ experience with presidential removal.19 While some political scientists have documented the relatively low success rate of calls for removal globally,20 no one has systematically examined the design of presidential impeachment from a comparative perspective. This is not for want of relevant evidence. As we shall show, the design of removal procedures for chief executives is almost uniformly a matter of constitutional text, not exclusively statutory policy. This reflects a (perhaps undertheorized) assumption that the question is an important one to be insulated, to some extent, from transient politics. The sheer proliferation of presidential removal provisions also suggests that there is a common problem to which constitutional designers around the world are responding. It could well be that designers are responding to slightly different understandings of a general problem, and are doing so under very different conditions of democracy (or lack thereof). Wide variance in context and conceptualization of governance problems might exist. Yet the observation of a common design choice suggests that there is something to be learned through comparison. Certainly, there is no reason to assume that the United States is “exceptional” among presidential regimes in the functions played by impeachment. That is a kind of intellectual parochialism that we think wise to avoid from the get-go.21

Examination of impeachment provisions and practices globally is relevant to a number of questions fundamental to a democracy.22 At a minimum, it seems important to know whether the substantive and procedural elements of the U.S. system are distinctive, or outliers as a matter of constitutional design. Relatedly, a global view of impeachment can illuminate its potential function in a constitutional democracy, and hence suggest how its scope and mechanisms might be reconciled with electoral democracy. When should a democratic mandate be superseded because of the perceived costs of allowing the people’s choice to remain in power after some form of wrongdoing? If supersession is to be allowed, should it be through a political process (defined and judged according to partisan standards), or a more formalized, law-governed process (say, defined by the criminal law)? And what mechanisms, institutions, and procedures should be involved in the removal process? Should they be other elected actors, or nonelected, professional institutions? What should be the result of a presidential removal: a new election, or either an ally of the president or someone else taking control of the government?

This Article analyzes the problem of presidential impeachment or removal through a comparative lens. We present here the first comprehensive analysis of how constitutions globally have addressed this question, and what the consequences of different design choices are likely to be. Because actual removals of chief executives turn out to be rare (although calls to remove are much more frequent), we employ a twofold empirical strategy. We begin by developing five case studies, including the United States, of removals that occur through a range of procedures and under quite different political conditions. This granular approach helps pick out some of the variation in constitutional technologies of presidential removal. It also offers clues as to what legal and political factors matter in practice. Causal inference, to be sure, is perilous given the small number of observed outcomes and the endogeneity of observed outcomes to institutional choices. That is, because presidential behavior will be influenced by the choice of substantive and procedural impeachment rules, it is not feasible to isolate the effect of those rules on decisions to impeach. Because the law influences both the independent variable of impeachable acts and the outcome of observed impeachments, no crisp causal inference is possible. Rather, we tentatively view different structural arrangements as inducing different patterns of both underlying behavior and removal-related responses. Next, we zoom out to offer a comprehensive, large-N description and evaluation of the relevant constitutional design choices. Finally, we draw carefully nuanced conclusions about the normative stakes of varying design decisions in this domain.

Before summarizing our key descriptive findings and normative suggestions, we should clarify the universe of cases that we are considering. Removal of a chief executive is a necessary power in any political system, whether presidential, parliamentary, or otherwise. Even traditional monarchies had procedures for removing kings who were incapacitated or incompetent.23 Our focus, however, is primarily on fixed-term executives, who tend to be called presidents.24 Such officials are found in an array of political systems, including presidential systems, semi-presidential systems,25 and even some parliamentary systems.26 We include heads of state in parliamentary systems, who tend to have a more ceremonial role, but exclude prime ministers (who are typically disciplined instead through a parliamentary “vote of confidence” mechanism).27

We show first that impeachment does not always focus on the criminal behavior or bad acts of an individual president. Rather, it also serves as a response to a particular kind of political crisis in a presidential system, commonly in which public support for the leader has collapsed. In some recent impeachments, such as in South Korea, crisis combined with evidence of criminality to oust a president from office.28 But in other cases, such as in Brazil and Paraguay, there was scant evidence of apex criminality.29 Removal was rather used to push out weak presidents who had lost the ability to govern. Consistent with this practice, many constitutions around the world include a textual standard for removal that explicitly goes beyond criminality to include governance failures or poor performance in office, while others enable such an approach through ambiguity. Generalizing from textual evidence and case studies, we suggest that impeachment globally is, in practice, a device to mitigate the risk of paralyzing political gridlock, rather than simply a way to deal with individual malfeasance. A second important empirical conclusion follows. Examining measures of democratic quality in impeachment’s wake, we find no evidence (at least in the small sample of extant cases) that impeachment of a president reduces the quality of democracy in countries where it is carried out.30 The same holds true when removal through impeachment is attempted, but not completed. The fear that a more political impeachment process would necessarily be destabilizing has no empirical support in the recent comparative experience. Rather than being a way of undermining or circumventing democracy, we suggest that in fact impeachment may play an important role in its stabilization.

Although we tread carefully in drawing normative conclusions given the limited pool of available data and endogeneity concerns, our analysis nevertheless has implications for the design and practice of impeachment, particularly in the United States. We argue that a model of impeachment focused only on the individual culpability of chief executives—what we call a “bad actor” model—is likely incomplete and undesirable as a functional matter. Instead, impeachment processes should be attentive to the broader political context, which we call a “political reset” model. Impeachment can be useful to ameliorate one of the major weaknesses of presidentialism—rigidity31 —by removing poorly performing presidents when their support has collapsed.

Professor Stephen Griffin has recently tracked the history of impeachment discourse in the United States to show that partisan dynamics forced it into a Procrustean bed of “indictable crimes” and nothing more.32 Consistent with Griffin’s careful analysis, Trump’s legal team argued in the Senate that the House’s articles were deficient in part because impeachment was only appropriate in the event of a violation of “established law” and, likely, “criminal law.”33 We think, to the contrary, that the comparative evidence suggests that such a narrow interpretation of the term “high Crimes and Misdemeanors” may well be problematic. A broader, more political meaning of this notoriously cryptic standard may make more functional sense as an element of a well-functioning democracy.

Aside from shedding new light on the well-studied issue of “high Crimes and Misdemeanors,” our analysis also critiques a range of crucial but less studied features of impeachment in the United States. Some are a product of judicial or political practice; others would require a constitutional amendment to fix. All are taken as givens—in quite problematic ways. For example, we highlight the striking fact that the impeachment standard in the United States is uniform across different types of actors, such as presidents, judges, and cabinet members, rather than varying as in many other countries. We think a more differentiated approach makes more sense. Impeachment of different kinds of actors serves different purposes, and it makes little sense to use a one-size-fits-all approach. We also highlight the ways in which actors other than legislatures contribute fact-finding, legitimacy, and other benefits to impeachment processes in some contexts. In particular, and contrary to the settled understanding in the United States and the leading precedent,34 we suggest that a more robust role for courts in impeachment processes may be consistent with a political, regime-centered model of impeachment. In some contexts, courts can lend credibility to factual and legal determinations made during impeachment—a credibility that has been in short supply during recent processes in the United States. Finally, our analysis suggests that impeachment design in the United States fails to maximize its value by having the vice president (or a similar actor) automatically succeed to office, rather than calling new elections. We think that calling new elections after a successful impeachment is a superior option because it increases impeachment’s ability to serve as a reset for a crisis-laden system.

We recognize that this topic is of great current interest in the United States, largely because of the recent impeachment trial and acquittal of Trump. Indeed, there is a growing academic and nonacademic literature on the topic of his impeachment.35 Some of these contributions confront Trump’s actions in light of the relevant standard; others are more abstract treatments not limited to the particulars of his case. We place ourselves in the latter camp, abstracting away from the current presidency, and avoiding inevitably partisan implications in the hope of generating more durable insights. At the same time, we also recognize that the occasion of the Trump impeachment and acquittal seems to be a particularly good moment to stimulate careful reflection on an important constitutional issue.

Our analysis is organized as follows. Part I motivates our analysis by presenting case studies of recent instances of presidential removals from around the world: South Korea, Brazil, Paraguay, and South Africa. We also briefly survey U.S. law and experience to benchmark domestic experience. Part II draws on large-N empirical evidence to describe and analyze the history, rules, and practice of presidential removal globally. We find that impeachment is often a response to governance problems related to waning public support for a fixed-term leader. It thus extends beyond the standard bad actor model that dominates much of the American legal discourse. Systems vary in terms of both the predicate acts that can trigger impeachment along with the process, including both the actors involved and the various rules governing time and consequence. Finally, Part III draws on this evidence to theorize better impeachment institutions, focusing on implications for the United States. We conclude by suggesting that at least as a prima facie matter a more frequent, systemic use of impeachment in presidential democracies, including our own, should not be feared. It is likely to do more good than harm.

I. The Irresistible Rise of Impeachment: Snapshots from the World of Presidential Removal

We begin by considering the three most recent cases of successful removal by impeachment—in South Korea, Brazil, and Paraguay—along with the removal of President Jacob Zuma midway through his second term in South Africa as a consequence of a protracted corruption-related investigation. These case studies are useful for “clarifying previously obscure theoretical relationships” and as a step toward “richer models” than would be enabled by purely large-N analysis.36 The case study approach is especially appropriate here because, as we demonstrate in Part II, the rate of successful impeachments in the past half century or so turns out to be small in comparison to the denominator of elected chief executives holding office, or even the number of proposals for impeachment.37 Impeachment is often proposed and rarely realized. A case study approach allows a thick account of most of the relevant positive instances of impeachment or removal that would be missed by a large-N analysis alone. Finally, by way of counterpoint (and to tee up our normative analysis in Part III), we recapitulate briefly the historical framing and practice of impeachment in the United States as a point of reference and contrast.

In each of our first three case studies, directly elected presidents did not finish their terms, albeit for different reasons. South Korea’s President Park Geun-hye was removed from office in 2017 after an impeachment confirmed by the Constitutional Court. Brazil’s President Dilma Rousseff was removed in 2016 shortly after her reelection to a second term in relation to an alleged fraud scheme. And Paraguay’s President Fernando Lugo was removed from office in 2012, primarily on the grounds that he had botched policy decisions prior to and after a massacre involving a land invasion. In each of these cases, the ousted presidents were extremely unpopular. Their ousters constituted a political opening, consequently, for political opponents, who gained new access to the levers of power. In South Africa, in contrast, where presidents are selected by the Parliament rather than directly elected, Zuma was replaced by a leader of his own party, after losing support from within the party.

In our view, all of these removals had normative justifications, albeit ones not necessarily anchored in the specific criminal acts of a given leader. But the political outcomes they produced were radically different. For example, after removing the incumbent, South Koreans elected a left-wing candidate, President Moon Jae-in, while Brazilians chose a fiery right-wing populist, President Jair Bolsonaro. His tenure is still too new to evaluate, but concerns about democratic backsliding and state violence have deepened. In contrast, Zuma was replaced by his copartisan President Cyril Ramaphosa, who went on to lead the African National Congress (ANC) party to a close election win. In most of these cases, the system has found a new equilibrium, and democracy has not fallen.

A. South Korea: The Park Impeachment

The South Korean Constitution allows impeachment for a “violat[ion of] the Constitution or other laws in the performance of official duties.”38 A majority of members of the National Assembly can propose an impeachment bill for the president, which must then be approved by a two-thirds vote.39 The president is immediately suspended from serving; his or her duties pass on to the Prime Minister.40 In a second stage, the impeachment motion then goes to the Constitutional Court for final approval.41

In the first Korean impeachment of the twenty-first century, this last step proved dispositive. In 2004, President Roh Moo-hyun was impeached.42 Before the Constitutional Court could decide on the question of removal, an intervening parliamentary election gave Roh’s party a slim parliamentary majority.43 The court, perhaps in a move of political pragmatism, decided that the charges against Roh were not sufficient to warrant removal.44 Roh went on to serve to the end of his term, though he eventually committed suicide during a corruption probe.45 The Constitutional Court’s decision was systemically important for clarifying many of the relevant rules.46 Most importantly, it held that even if charges against a president were well-founded, removal should only occur if there was a grave violation of law and if removal was “necessary to rehabilitate the damaged constitutional order.”47 The court also explained the division of labor in impeachment cases, holding that the Assembly had a political and fact-finding role, while the bench itself was the ultimate judge of whether the facts presented met the legal threshold for removal.48

A decade later, a second South Korean president faced defenestration. This time the court ratified some of the grounds for impeachment. President Park Geun-hye, like most Korean presidents, found her popularity dropping precipitously after her 2012 election.49 In 2016, it was revealed that she had been taking instruction from, and acting on behalf of, a close confidant, Choi Soon-sil.50 Choi’s father had been the head of a secretive cult and an associate of Park’s father, President Park Chung-hee. Choi had been extorting money from Korea’s large business corporations. When these facts were revealed, massive public demonstrations ensued and the opposition party filed impeachment motions against Park.51 The charges included seven counts, including, inter alia, abuse of power, violating the duty of confidentiality by sharing government documents with Choi, and violation of the right to life in the Sewol ferry disaster, which had taken the lives of hundreds of high school students in 2014. Several members of Park’s own Saenuri Party joined in passing the motion by the required two-thirds vote, and Park was suspended as president.52 As in Roh’s case, the Constitutional Court then initiated its proceedings.

On March 10, 2017, the court delivered a verdict upholding Park’s impeachment on three of the seven counts: violation of the obligation to serve the public interest, infringement upon private property rights, and violation of confidentiality.53 Her interactions with the “shaman or medium” Choi were central to this finding, as they were to the growing tide of public anger at her administration’s corruption.54 The court did not accept three other grounds for impeachment, including one based on allegations related to the Sewol ferry disaster, and it found a final charge—the “obligation to faithfully execute the duties of the President”—to be nonjusticiable.55 The court then held that these charges met the test for seriousness laid out during the Roh impeachment case because they gave a private citizen influence over the office of the presidency.56 Park was subsequently convicted in criminal court. She is currently serving a twenty-five-year prison term.57

Under the South Korean Constitution, an impeached president is replaced by the prime minister, a weak vice presidential figure without independent executive authority. Moreover, the prime minister assumes presidential duties as soon as the impeachment charge is approved by the National Assembly, while the Constitutional Court conducts its trial. Importantly, though, the Acting Presidency lasts only until a new presidential election can be held, a period of no more than sixty days.58 After Park’s removal, Prime Minister Hwang Kyo-ahn remained in office until new elections in May 2017 brought in Moon Jae-in.59

In our view, the removal of Park before her five-year term ended was a model of procedural integrity. The impeachment decision by the Constitutional Court laid out in depth the extent to which Park had given over the public trust to a private individual, with no official position or relevant experience. It resolved a major political crisis in which hundreds of thousands of people were demonstrating in Seoul. The court’s judgment, moreover, provides a model of sober evaluation of the evidence, rejecting superfluous charges while upholding those for which the evidence was clear. At the same time, the court’s careful election of some impeachment grounds over others seems to have tracked the nature of public discontent at the perceived dysfunctionality of the Park government.

B. Brazil: The Rousseff Ouster

Shortly after President Dilma Rousseff had been elected to her second term in office as Brazil’s president, a scandal known as “Operation Car Wash” revealed massive corruption tied to Brazil’s state-owned oil company during the period she had been in charge of it before becoming president.60 Though no evidence emerged that she was personally involved, Rousseff was held politically responsible for the failings of her party’s (the Worker’s Party, or “PT”) long period in governance. With public discontent at PT’s perceived corruption rising, opponents began to look for a hook to remove her. In late 2015, Rousseff was charged with a violation of article 85 of the constitution, which details the grounds for impeachment.61 Just like previous presidencies, Rousseff’s administration had engaged in an accounting maneuver to try to make it look as if the government had more assets than it did. The maneuver allowed it to allocate funds to social programs without direct allocation from the Congress. A tax court held the maneuver to be illegal, opening the door to an impeachment that many analysts believed to be primarily partisan.62

The substantive grounds for impeachment in the Brazilian Constitution are ambiguous. Article 85 states:

Acts of the President of the Republic that are attempts against the Federal Constitution are impeachable offenses, especially those against the: I. existence of the Union; II. free exercise of the powers of the Legislature, Judiciary, Public Ministry and constitutional powers of the units of the Federation; III. exercise of political, individual and social rights; IV. internal security of the Country; V. probity in administration; VI. the budget law; [and] VII. compliance with the laws and court decisions.63

Article 85 of the Brazilian Constitution thus lays out a fairly broad and reasonably political—as opposed to strictly legal—standard for impeachment, which seems to reach well beyond criminality. It also includes a “by law” clause giving legislation the power to further define both the standards and process for impeachment. The relevant law, Law 1079, was passed in 1950, and so predates the current constitution of 1988, although the law was amended more recently.64 The law, oddly, conflicts with the constitutional text in certain key respects. Some commentators have suggested the law may play a bigger influence on impeachment in practice than the constitution itself.65 The law fleshes out the broader categories found in article 85, but still maintains a definition of those terms that is highly political in nature.

The allegations against Rousseff focused on crimes against the administration and the budget, chiefly (as noted above) that she disbursed public money without congressional authorization.66 The allegations also linked Rousseff to the Operation Car Wash scandal, albeit indirectly. More specifically, it was argued that she had failed to act with sufficient vigor against participants in the scandal.67 This latter allegation, however, did not become the basis for impeachment, which instead focused (at least formally) solely on the alleged illegal appropriations.68

Article 86 of the constitution fleshes out the bare bones of the process of impeachment of the president, which again is regulated more closely in Law 1079 and in internal congressional bylaws.69 Under that process, the lower House investigates accusations and decides whether to impeach the president, by a two-thirds vote. Cases then proceed either to the Senate (in cases of “impeachable offenses” defined in article 85) or to the Supreme Court (in cases of “common criminal offenses”), for the final trial.70 Once the Senate begins removal proceedings, the president is suspended for up to 180 days during the trial. A two-thirds vote of the Senate is required to remove officials from office for commission of an “impeachable offense.” As in the United States, the president of the Supreme Court must be present and must preside over the trial that occurs in the Senate.71

In 2016, Rousseff was formally impeached by the required two-thirds vote in the lower house on a vote of 367–13, and trial commenced in the Senate.72 When the Senate voted to initiate removal proceedings, Rousseff was suspended and Vice President Michel Temer took over as acting president. Temer retained this position after the Senate voted on August 31 to remove Rousseff from office, again by a two-thirds vote of 61–20, from August 2016 until the end of 2018.73 But at the same time, the Senate failed to reach a two-thirds supermajority to deprive Rousseff of her political rights for eight years. As a result, she retained the ability to run for future office (and indeed ran unsuccessfully for a Senate seat in 2018).74

The Supreme Court played a complex, multilayered role throughout the episode as an agenda setter and adjudicator of key procedural choices. Unlike its South Korean analogue, however, it exercised no ex post review once the legislative part of the impeachment process had come to its conclusion. Actors on all sides of the political spectrum bombarded the bench with a series of challenges and requests throughout the impeachment process. The court’s response was mixed. On the one hand, the court generally avoided judging the substantive question whether the allegations against Rousseff were sufficient for impeachment, demurring to the legislature.75 On the other hand, it issued some judgments that impacted the process in meaningful ways. For example, the court issued a ruling in December 2015, when the impeachment process was just beginning, that allowed the process to go forward but held that the committee investigating Rousseff needed to be reconstituted because it had previously been stacked with proponents of impeachment, in violation of the relevant laws and regulations.76 Membership in the committee, directed the court, needed to be proportional to the composition of the House.77 The court also held that the Senate, as well as the House, should issue a preliminary vote on whether to accept the impeachment allegation against Rousseff.78

It is worth noting that, as in South Korea, the recent Brazilian impeachment had a historical precursor: President Fernando Collor de Mello’s ouster in 1992, shortly after Brazil’s transition to democracy.79 The latter shared key features with Rousseff’s removal. As with Rousseff, political context rendered Collor vulnerable to impeachment. He was an outsider president without strong ties to existing parties; he hence had great difficulty building a governing legislative coalition. Collor was forced to resort aggressively to unilateral decree powers because of his lack of partisan support, often reissuing provisional decrees before they could expire.80 Opponents alleged that this practice was abusive. It was eventually restricted by the Supreme Court and then by a constitutional amendment.81

The immediate triggers for Collor’s impeachment were corruption allegations. The House’s charges did not allege any specific crimes, but rather facilitating “the breach of law and order” and behaving in a way that was inconsistent with the “dignity” of the presidential office.82 Collor argued that noncriminal acts could not be the basis for impeachment. But the House and Senate proceeded to impeach him regardless. Collor technically resigned before the impeachment was completed, but the Congress nonetheless finished the process, with the Senate voting in favor by an overwhelming 76–3 vote. As in the Rousseff impeachment, judges played a major role in Collor’s: the president of the Supreme Court, in his role presiding over the Senate trial, crafted special rules that simplified and streamlined some of the procedures found in Law 1079.83

What lessons does the Rousseff impeachment (and its echoes in the Collor impeachment) hold for the comparative study of presidential removal? To begin with, unlike the Park ouster in South Korea, it is hard to conceptualize Rousseff’s impeachment as being about criminal behavior, or even serious moral wrongs, of the President herself. The acts that formed the basis of her impeachment—basically, accounting tricks and related devices to authorize additional social spending, allegedly with the intent of helping the PT retain power—had been engaged in by presidents prior to Rousseff. Even the broader context for the allegations and impeachment, which revolved around alleged involvement with the Operation Car Wash investigation, did not yield much evidence inculpating Rousseff herself. Rather, she was accused of negligence in handling accusations and being connected to involved actors. But these accusations did not meaningfully distinguish her from the larger political class. So it is perhaps unsurprising that Rousseff’s impeachment prompted outcry in some quarters and was described by her and her allies as a coup.84

The political framing of the impeachment resonates even more when Brazil’s recent political history is brought into the analysis. Rousseff’s 2014 reelection campaign had been fought in a context where the revelations of the Operation Car Wash investigation started to discredit the country’s political class as a whole. When she won reelection in 2014, it was by a much smaller margin than in 2010.85 Indeed, her PT party lost support in Congress. In consequence, she was forced to rely on a more fluid pattern of support without a clear majority coalition to legislate. The president of the House, Congressman Eduardo Cunha, was never an ally of the PT and became strongly opposed to it in mid-July 2015; his party (the second largest in the House) turned against Rousseff during the impeachment process, depriving her of needed support. The theory of the case against Rousseff also “echoed the street protests” against the PT more generally.86 At the time, the economy in Brazil was experiencing an extended period of stagnation.87

But, crucially, the impeachment did not reset the political system. New presidential elections did not occur until 2018. Instead, Temer took over the chief executive’s role. As a result, PT allies saw the impeachment, as well as related actions like the jailing of former President Luiz Inácio Lula da Silva, as an attempt by more traditional and conservative actors to take down the country’s most organized progressive force.88 Temer served for about two and a half years after Rousseff’s suspension, but was a weak and unpopular president. He had already been implicated in corruption more directly than Rousseff, as were many of those who remained in Congress.89 The discrediting of Brazil’s political class en masse continued; space thus opened for self-styled outsider and right-wing populist Jair Bolsonaro to win election in 2018.

Bolsonaro has not been immune from impeachment talk, either. Notorious for consistently dismissing the COVID-19 virus as the “little flu,”90 Bolsonaro’s erratic performance and authoritarian rhetoric have led to calls for his impeachment from Lula, among others.91 At the time of writing, some forty-eight petitions for impeachment are before the Speaker of the House, though it is not clear whether he will let them advance.92 This is in part because Vice President Hamilton Mourão, whose popularity exceeds that of Bolsonaro, is considered a wild card himself with possible authoritarian leanings. The lack of a reset option has worked to Bolsonaro’s advantage.

C. Paraguay: The Removal of Lugo

The removal of Paraguayan President Fernando Lugo by Congress in 2012 is another case which is difficult to interpret as the removal of a criminal or morally depraved leader. A former Catholic bishop and political outsider, Lugo won the presidency in 2008 on the ticket of a small party and in alliance with seven other political parties, ending over sixty years of rule by the Colorado Party.93 In return for the support of the largest opposition party, the Liberal Party, he picked an insider vice president, Federico Franco, with Liberal bona fides.94 Lugo and his vice president were not close. There were rumors from early in Lugo’s term that the Liberals were seeking to supplant him with Franco.95 Further, Lugo was unsuccessful at carrying out most of his initially ambitious political and economic programs, especially on his signature issue of land reform, and over time his popularity fell sharply.96 He was unable to pass any significant legislation in a deeply divided Congress. His own coalition remained highly factionalized.97 There was considerable instability during Lugo’s term, with other impeachment attempts prior to the successful one. A failed military coup led to Lugo’s replacement of the entire military leadership in 2009.98

The proximate cause for the Lugo impeachment was an incident on June 15, 2012, where seventeen people (six police officers and eleven farmers) were killed.99 Landless farmers occupied land estates that they alleged had been unlawfully acquired, leading to the clashes. The impeachment charges laid against Lugo focused on this incident, as well as four others,100 and complained in general terms of “bad performance in office.”101 Referring to the ki-llings, the charging document also stated sweepingly that Lugo had exercised power in an “inappropriate, negligent and irresponsible way . . . generating constant confrontation and war between social classes.”102 It did not accuse Lugo, though, of committing a crime. Like the Brazilian organic law, the Paraguayan Constitution explicitly allowed impeachment for poor political performance.103

A lightning-fast process of impeachment began and ended within the space of mere days. On June 21, 2012, the Chamber of Deputies voted to impeach Lugo by a 76–1 vote; the next day, the Senate voted to remove him from office by a 39–4 vote.104 The rules required a two-thirds vote of those present in the Chamber of Deputies for impeachment and a two-thirds absolute majority of members of the Senate for removal. Both thresholds were easily met.105 Under the constitutional framework in force, the vice president and Liberal Party member Franco, who had become a manifest opponent of Lugo, then became president.106

Lugo and his allies complained of a lack of due process in his impeachment. They pointed to the breathtaking speed of the impeachment and the fact that he was offered only two hours to appear before the Senate to present his defense. Like Rousseff and her allies, regional leaders condemned the removal as an “institutional coup.”107 The leaders of many other countries in the region agreed.108 Paraguay was in fact suspended from regional organizations Mercosur and Unasur until the next set of elections were held in the country in 2013.109 The Inter-American Commission on Human Rights issued a statement calling the speed with which the removal was carried out “unacceptable” and stating that it was “highly questionable” that the removal of a Head of State could be “done within 24 hours while still respecting the due process guarantees necessary for an impartial trial.”110 It concluded that the speed of the procedure raised “profound questions as to its integrity.”111

It is hard to see the Paraguayan example, with its extraordinary speed and resulting lack of deliberation, as a model of how impeachment should be done. At the same time, the case shows how impeachment can work more as an attempted exit from a political crisis rather than a judgment of criminal behavior (or serious wrongdoing) by the incumbent. Like the Rousseff removal, but even more clearly, the impeachment of Lugo did not focus on his culpable status as a “bad actor.” The opponents of Lugo did not argue that he had committed a statutory crime. Instead, they relied on his “poor performance of duties” (mal desempeño de sus funciones) in office, a noncriminal ground of impeachment expressly contemplated in the Paraguayan Constitution.112

As in the Park and the Rousseff cases, it appears that a decisive factor in Lugo’s impeachment was the fragility of his political support. Lugo was removed because he had lost the support of nearly the entire political class, including most of his own coalition, and was deeply unpopular. The Liberal Party, for example, resigned en masse from Lugo’s cabinet just before the impeachment began.113 Lugo appealed his removal to the Supreme Court, but the court summarily dismissed the petition in a brief order, using reasoning similar to that of the U.S. Supreme Court when confronted with challenges to impeachment procedures. It held that the process of impeachment was delegated to the legislature and that the court had no basis to intervene.114

In effect, then, the Paraguayan impeachment process operated as a (supermajoritarian) vote of no confidence in the president. There are similar regime dynamics in the South Korean and (especially) Brazilian contexts as well, where the criminal allegations sometimes seem to be used as cover to remove unpopular presidents who had lost an enormous amount of congressional support. The Paraguayan impeachment is the clearest case of removal operating to address political deadlock rather than particular individualistic flaws.

D. South Africa: The Ouster of Zuma

We now turn to a case in which a president was in effect removed, albeit in the end through a resignation rather than the culmination of a formal process of removal: the ejection of Jacob Zuma from office in the middle of his term as South Africa’s president in early 2018. Although South Africa has a president with a substantive rather than a ceremonial role, the 1996 South Africa Constitution is more akin to a parliamentary rather than presidential system. The president is not directly elected by the public, but chosen by the Parliament. Moreover, as with prime ministers in parliamentary systems, the Parliament has the ability to force the resignation of the president by voting no confidence in him or her at any time.115 Since 1996, the position has always gone to the head of the dominant African National Congress. Under conditions of ANC hegemony, the president will continue in office so long as he or she can maintain the support of members of the party.

But, under section 89 of the constitution, the president can also be removed by a two-thirds parliamentary supermajority via an impeachment-like procedure, in the event of a serious violation of the constitution or law, serious misconduct, or an inability to perform the functions of the office, and a figure removed in this way may not receive any benefits of the office or serve in any public office in the future.116 Even though the position of president in South Africa is more like that of a prime minister in other systems, we discuss the case briefly here because it highlights some mechanisms that may facilitate a successful removal process, especially the involvement of other, nonparliamentary institutions.

The Zuma presidency was characterized by an acute crisis of corruption. During the tenure of his predecessor President Thabo Mbeki, an “ANC party-state” developed in which party loyalists were assigned to high posts in public office, parastatals came under party control rather than state control, and ANC elites increasingly dominated the “commanding heights” of the private economy.117 During the Zuma presidency, the state was captured by a small group of private actors, who steered public contracts to preferred businesses in exchange for kickbacks.118 Ministers who declined to cooperate were quickly relieved of their duties and office.119 As a result of ineffectual or corrupt presidential leadership, a raft of structural, macroeconomic problems accumulated.120

Zuma did not keep his hands clean. His country residence “Nkandla homestead” in KwaZulu-Natal became an epicenter of public controversy as a result of a publicly funded security upgrade ultimately costing some R246 million.121 At least initially, the ANC resisted attempts to hold him accountable. Without an internal check from his party, and with that party playing a dominant role in the country’s politics, there was a real risk of the erosion of democracy itself. But the prosecuting and investigating institutions of the state were not particularly active in seeking to hold Zuma accountable. Only the Public Protector, an ombudsman-like body with relatively weak powers, seemed to be willing to challenge Zuma’s corrupt behavior and the larger problem of state capture.

In this context, the Constitutional Court intervened several times to both protect opposition rights within the Parliament, and also to require Parliament itself to maintain and use mechanisms for presidential accountability. Hence, the court strongly suggested that votes on no confidence in the president had to be secret.122 It also insisted that minority rights in Parliament not be squelched.123 It then held that the Speaker of the House could not simply ignore a motion of no confidence challenging Zuma’s continued tenure.124 Parliament had a duty to hear such motions, the court instructed.125 In a particularly critical decision, the court empowered the Public Protector, whose findings were given legal force.126 The Public Protector had issued a report that followed an investigation into the use of public funds for the improvement of the President’s Nklanda residence. The report concluded that money misspent on portions of the upgrades should be repaid by Zuma. The President failed to comply with the findings, claiming that they constituted mere “recommendations.”127 The court, however, held that such findings were legally binding and that the President was not entitled to disregard them. It also held that Parliament had to come up with a mechanism to hold the president accountable. Importantly, the Public Protector’s report concluded that in receiving undue benefits from the state, the President had breached “his constitutional obligations.”128 Many regarded this statement, now imbued with the force of law, as fulfilling the criteria for impeachment set forth in section 89(1) of the constitution.129

Despite this, Zuma subsequently survived a secret ballot of no confidence in August 2017.130 The narrowness of the vote margin, though, demonstrated the extent to which Zuma and his allies had lost support within the parliamentary ANC party. “It thus anticipated, and rendered more likely, Zuma’s ultimate February 2018 ouster.”131 The ANC effectuating a removal of its own leader is a remarkable instance of an intraparty check on power. Such intraparty checks are quite rare in true presidential systems and are likely to reflect the strategic calculation of party insiders of how to minimize electoral losses due to an unpopular elected figurehead.

In short, the South African Constitutional Court forced the political system to act. It did not directly remove the President, but it ensured that the processes of democratic accountability could not be ignored. The Public Protector also played the vital role of documenting “state capture” in a form that Zuma could not easily ignore. At least formally, the Zuma case is a “near miss” rather than an impeachment.132 But it illustrates how institutional processes can cause a collapse in public support for a leader, which can make their continuance in office untenable. Across all these cases, the formal processes of removal operated in tandem with, and were entangled in, changing public sentiment with respect to the presence of not just personal malfeasance, but also a systemic crisis of governance. The South African case thus confirms that presidential removal operates as a way of expressing concern about systemic crisis, even if the causal relationship of legal censure mechanisms to public disapproval varies from the earlier cases.

E. Impeachment in the United States

With the recent cases of South Korea, Brazil, Paraguay, and South Africa in hand, it is useful to return to the United States.133 Removing a sitting president in the United States through impeachment has been described as “the most powerful weapon in the political armoury, short of civil war.”134 Yet this is in some tension with the thinking at the Philadelphia Convention, where there is evidence of a rather more capacious concept. The delegates to that Convention borrowed the institution of impeachment from English law, where it had been a device to discipline and remove the king’s ministers.135 Indeed, over the centuries, it provided a central power of parliamentary accountability in the United Kingdom, but was not limited to serious crimes.136 Even while the debates about the Constitution were ongoing, for example, Edmund Burke was spearheading an effort to impeach Warren Hastings, the first Governor-General of India, for “high crimes and misdemeanors” in the form of gross maladministration.137

The formation of the constitutional text on impeachment followed from one of those exchanges between two delegates that admits of speculation, inference, and endless conjecture: one of the early iterations of the impeachment mechanism considered by the 1787 Constitutional Convention limited impeachment only to cases of treason or bribery.138 But George Mason of Virginia worried that those bases would be insufficient to remove a president who committed no crime but was inclined toward tyranny.139 Mason proposed adding “maladministration” as a basis for impeachment and removal from office, which would have made our system more like a parliamentary one.140 When James Madison objected that maladministration was a vague term, Mason then proposed the usage “treason, bribery, or other high crimes and misdemeanors.”141 It was that language that was ultimately adopted in the Constitution.142 The Mason-Madison exchange suggests that a narrow “bad actor” model fails to exhaust impeachment’s purpose. Yet it also allows different inferences about how far beyond that model the text ought to extend.

As a congressional report issued during the impeachment of President Richard Nixon recounts, the phrase “high Crimes and Misdemeanors” had been first used in 1386 during a procedure to remove Michael de la Pole, the first Earl of Suffolk.143 The Earl’s failures included negligence in office and embezzlement. He had failed to follow parliamentary instructions for improvements to the king’s estate and had failed to deliver the king’s ransom for the town of Ghent, letting it fall to the French as a result. For these failures, Suffolk became the first official in English history to lose his office through impeachment.144 Impeachment was subsequently used episodically throughout English history,145 before falling into desuetude with the creation of modern parties and the emergence of the “ministerial responsibility” principle.146 Under ministerial responsibility, a minister can be removed simply on a lack of confidence, which makes removal a purely political matter without need for a legal proceeding. Impeachment was last used in the United Kingdom in 1806.147 Drawing on this history, the Nixon-era congressional report concludes that “the scope of impeachment was not viewed narrowly.”148

The ratification debates contain further evidence of this “political” understanding. Hence, Alexander Hamilton wrote in Federalist 65 that impeachment would be addressed at “those offences which proceed from the misconduct of public men, or, in other words, from the abuse or violation of some public trust. They are of a nature which may with peculiar propriety be denominated political, as they relate chiefly to injuries done immediately to the society itself.”149 Subsequently, Madison, speaking at the Virginia ratification convention for the Constitution, intimated a fundamentally political purpose to impeachment. When Mason raised concerns about the breadth of the pardon power and the possibility that a president would use it to establish tyranny, noting that a president could use it to pardon crimes that “were advised by himself,” Madison responded that impeachment would be the appropriate remedy in such a case:

There is one security in this case to which gentlemen may not have adverted: If the President be connected in any suspicious manner with any persons, and there be grounds to believe he will shelter himself; the house of representatives can impeach him: They can remove him if found guilty: They can suspend him when suspected, and the power will devolve on the vice-president.150

Consistent with this evidence, most impeachment scholars in the United States have argued that the substantive standard reaches beyond crimes, although there are debates over exactly how broad the standard is.151 Scholars tend to conclude, consistent with that sense of the original understanding, that impeachable offenses must be “abuses against the state” that are analogous, in injury and intention, to those that concerned the Founders.152

But in the practice of impeachment, original understanding has not been destiny. Professor Griffin’s examination of the historical record of presidential impeachments shows that “the historical reality of the Johnson, Nixon, and Clinton impeachments is quite different.”153 Rather than hewing to the broader “Hamiltonian” reading of impeachment, as Griffin calls it, presidents and their supporters have since the early nineteenth century articulated an unsurprisingly narrower alternative—and have largely prevailed. On this more constrained view, presidents could be impeached “only for committing indictable crimes, or at least significant violations of law.”154 As noted in the Introduction, debates during the Trump impeachment reflected and deepened this conflation between serious crime and impeachment.

During the impeachment of President Andrew Johnson, for example, “Congress wanted to impeach Johnson for abusing his constitutional powers to obstruct the enforcement of federal laws.”155 But the actual process centered mostly around Johnson’s supposed violation of the Tenure in Office Act by dismissing Edward Stanton from his post as Secretary of War.156 Since this was not really a crime in any conventional sense, but rather something more akin to an abuse of power, the tension between different models of impeachment was apparent. In contrast, during the impeachment of President Bill Clinton, the House of Representatives seemed to proceed under a more legalistic conception of the impeachment power. Three of the charges formulated by the House spoke directly to alleged crimes committed by Clinton: two counts of perjury and obstruction of justice. Two of these three counts passed the House and formed the basis on which Clinton was impeached; the third narrowly failed. In contrast, a single count of abuse of power failed overwhelmingly in a 148–285 vote.157 Similarly, during the weeks leading up the impeachment vote of Donald Trump, many possible charges were put forward, but the final charges were two: abuse of power and obstruction of Congress. The abuse of power count was criticized by Trump’s team as legally deficient on the grounds that it did not allege the violation of clearly established law, and particularly of a crime.158

Another reason for the dominance of a narrow, criminally focused understanding of impeachment (one not stressed by Griffin) may be the manner in which the text is formulated. The Constitution is normally read to create a unified impeachment standard that includes judges, high political officials, and chief executives.159 Removing only bad actors, essentially convicted criminals, makes good sense in the removal of judges as a way to protect judicial independence. Yet the same standard applied to chief executives may inhibit impeachment from facilitating exit during political crises, or at least may force actors to make disingenuous statements during impeachment processes. If so, this would be an example of drafting choices having unanticipated, even pernicious, effects on major elements of constitutional operation—a point to which we return in Part III.

Apart from the question of impeachment’s substantive threshold, the law and the historical record are sparse. Since the Founding, there have been many resolutions of impeachment brought against federal officials. Twenty were formally impe­ached in the House of Representatives.160 Of these, fifteen were federal judges, one was a senator, one a cabinet member, and three—Andrew Johnson in 1868, Bill Clinton in 1999, and Donald Trump in 2019—were sitting presidents.161 Of these, eight were convicted after a trial in the Senate, and removed from office. No chief executive has ever been removed from power following a Senate trial. The Clinton impeachment failed to achieve the requisite two-thirds vote by a significant margin; the Johnson removal failed by a single vote, 35–19; and Trump was acquitted by a vote of 48 in favor of conviction and 52 against on the closest charge, abuse of power.162

The difficulty, and resulting infrequency, of impeachment generates a perhaps troubling dynamic: it elicits a surfeit of impeachment talk, and arguably improper invocations of the procedure. Because impeachment attempts require a supermajority of two-third of senators for removal, there is a moral hazard dynamic inducing individual members in the House to introduce resolutions of impeachment. Members can claim credit without having to take responsibility for the subsequent costs of an impeachment that will almost certainly not proceed. As a result, almost every president has faced an effort by members of Congress to use impeachment as a way to paint them as a bad actor. In particular, in an increasingly polarized era, motions of impeachment have become somewhat routine, even if the process has rarely advanced beyond the stage of introduction. (In the post-Watergate era, President Jimmy Carter is the only president not to have had such a motion introduced.) The Clinton impeachment, in fact, was marred by such problems. Republicans wielded the report of special counsel Kenneth Starr as a way to paint Clinton as a bad actor. The crux of the debate focused on whether the acts that Clinton was accused of (essentially, lying under oath as part of a civil case about his sexual conduct) were sufficient to warrant impeachment. What got lost in this focus on the conduct of one man were broader issues of political context: Republicans controlled the House and thus were able to push through articles of impeachment, but they had nowhere close to the two-thirds majority in the Senate needed to remove Clinton from office without substantial Democratic party votes. The prospect of Democrats turning on Clinton was remote, given that his popularity remained high throughout the impeachment process.

Beyond this, one of the most striking regularities of historical practice in the United States is the absence—especially notable in comparison to the South Korean, Brazilian, and South African examples—of any real role for the courts.163 The Supreme Court has identified impeachment as the quintessential political question that precludes virtually all judicial review.164 The Court has found issues related to impeachment nonjusticiable, mostly bec­ause the text of the Constitution committed them “sole[ly]” to the two houses of Congress.165 Since no constitutional text clearly prohibits the Court from supervising legislative action in this area, this decision is perhaps better explained by pragmatic factors, such as the chaos that could ensue if there was a constitutional challenge to the removal of the president, and by the difficulty of crafting standards to figure out what terms like “try” mean in the context of an impeachment. One implication of this relatively light judicial touch is that there has been no “overlegalization” of impeachment procedure. This at least leaves open the possibility of impeachment being deployed as a way of removing a deeply unpopular leader.

In summary, impeachment in the U.S. context is marked by the gap between original expectations and incentive-compatible practice. Instead of a serious tool of accountability to remove a president in moments of systemic risk, impeachment talk has become an instrument of political harassment. On one view, therefore, it is possible to characterize the U.S. system of impeachment as marked by the worst of both worlds—an ineffective tool that nonetheless has become highly politicized.

F. Conclusion

Except for the United States—where the impeachment of chief executives has largely fallen into desuetude beyond the context of partisan cheap talk—there is a tight connection between removal mechanisms for chief executives and the presence of a crisis of popularity. Where both political elites and the public perceive a regime as unable to operate effectively (for whatever reason), they are inclined to support removal. Removal in the global context is not a matter of individual malfeasance. Rather, these case studies suggest, impeachment can additionally work as a systemic means of political reset triggered in moments of deep confidence crises among the public. Whether this conclusion can be sustained by a broader consideration of large-N comparative evidence is the question to which we turn next.

II. The Dynamics of Impeachment in Global Perspective

The case studies presented in Part I suggest that the term “impeachment” is in practice a catchall for a range of different practices. In this Part, we ask how frequently one observes different substantive and procedural versions of impeachment across different jurisdictions in different periods. As noted in the Introduction, we focus on the removal of fixed-term presidents. The most important examples of these are in presidential systems like the United States, where a chief executive who selects the government and has at least some constitutional lawmaking authority is selected by direct elections and survives for a fixed term of years,166 or in semi-presidential systems like France, where a fixed-term president coexists with a prime minister and both figures may have substantial power.167 But some parliamentary systems (such as Austria) also have fixed-term presidents who serve as heads of state with no real governmental power; we include impeachment of these figures as well in our dataset, though the cases are rare. In appropriate instances, we provide separate statistics for subsets, such as presidential and semi-presidential systems. We draw many of the statistics and analyses that follow from the Comparative Constitutions Project, a comprehensive inventory of the provisions of written constitutions for all independent states between 1789 and 2006, with data updated through 2017.168

A. Impeachment from Text to Practice

It is very common for democratic constitutions to provide for removal of the head of state under some conditions. As of 2017, 90% of presidential and semi-presidential regimes had constitutional rules that laid out a process for removal, either for incompetence, criminal action or some other basis.169 The procedures differ widely on such issues as the basis for dismissal, the process of proposal for dismissal, the process of approval, the period of the term of office within which the president’s mandate can be revoked, and the various timing of different steps. But they are matters of constitutional text, not of statutory enactment. Yet as the case of Brazil shows, the fact of constitutional entrenchment does not necessarily preclude the enactment of statutes with important effects on the process.170 We focus here, however, mainly on constitutional text. As a result, due caution should be exercised in drawing inferences about how that text interacts with statutory supplements or institutional cultures. In this Section, we first provide some basic empirics about the frequency of impeac­hment, and then lay out some examples of the range of provisions.

The ubiquity of constitutional text on impeachment is matched by a similar pervasiveness of attempts to remove presidents. Although attempts are not rare, they are rarely successful. One scholar, Professor Young Hun Kim, notes that some 45% of new presidential democracies faced an impeachment attempt in the period 1974–2003, and that nearly a quarter of presidents who served in this period were subjected to an attempt.171 Such attempts can vary in seriousness, ranging from mere calls by some set of legislators for impeachment to full formal votes in the parliament. Defined as a mere proposal in the legislature (that is, the first two rows of Table 1), attempts are exceedingly common. Supplementing Kim’s data, we gathered data on all such attempts between 1990 and 2018, and found at least 210 proposals in 61 countries, against 128 different heads of state. Using Kim’s fourfold framework for level of attempt, we identified the highest level of seriousness in each attempt, and report these in Table 1. We add the first two rows to get the total number of proposals, though acknowledge there is some difficulty distinguishing different attempts.

Table 1: Frequency of Impeachment Attempts 1990–2018

Level	1990–1999	2000– 2009	2010–2018	Total
1 = proposal by some deputies to impeach	34	80	30	144
2 = unsuccessful attempt to place the question on the parliamentary agenda	22	37	10	69
3 = parliament votes on impeachment but motion fails	3	11	6	20
4 = parliament passes an impeachment vote	8	8	6	22
Head of state leaves office before process complete172	3	3	3	9
Removal through impeachment	3	4	3	10

These attempts are not uniformly distributed. Impeachment is quite common in some countries: Ukraine, for example, has featured 25 different proposals in the 28-year period we examine. Other countries with frequent calls include Nigeria (17), South Korea (13), Ecuador (10), the Philippines (9), and Brazil (10). Russia had 13 attempts in the tumultuous 1990s, but none since President Vladimir Putin came to power. These are countries, one might speculate, where ordinary processes of political bargaining have broken down, leading parties to escalate quickly to the ultimate weapon in the political arsenal. Preliminary evidence also indicates that such motions become more likely after the first deployment.

As the last row in Table 1 demonstrates, successful removal by impeachment is a rarity. We identify a total of 10 cases since 1990, listed in Table 2 below.173 Close inspection of these cases suggests that successful removal typically involves a situation in which the opposition has control of the parliament and is also able to convince some members of the president’s party to defect. Both attempts and removals are more frequent when the president is unpopular and does not have a majority of support in the legislature. They often occur in the context of structural shifts in the larger party system.174

Convincing a president’s copartisans to defect is difficult. Presidential systems are characterized by single individuals who enjoy popular appeal but may not necessarily have strong roles within their own parties.175 Party leaders may have a good deal of trouble controlling their presidential candidate once in office (and so the occasionally rocky relations between President Trump and congressional leaders of the Republican Party are less atypical than one might expect). While one might assume that this would lead to parties turning against their presidents on occasion, the linked electoral fates of parties in the legislative and executive branches mean that they have relatively weak incentives to do so (even if they do control the levers of impeachment or removal).176 At the very least, to impeach one’s own party leader implies that the party was incompetent in choosing the person as a candidate. Worse, it might catalyze a fragmentation of the party, as the spurned leader breaks off with his or her own political coalition.

To illustrate why it is that removing presidents is so hard even when their party turns against them, consider the attempt to dismiss President Ranasinghe Premadasa of Sri Lanka. In 1991, a motion to impeach Premadasa was raised in the Parliament, and was supported by some members of his own party.177 Premadasa was able to expel dissident members from the party, which meant, in accordance with the text of the Sri Lankan Constitution, that they lost their seats in Parliament. Other instances of failed attempts in presidential systems to use impeachment for intraparty conflict include the case of South Korea’s Roh Moo-hyun, as discussed above in Part I.A. Recall that Roh was impeached after a split in his party, but not removed by the country’s Constitutional Court, as it found that the violations were insufficiently severe to justify a removal from office.178 Again, because Roh maintained public support, and his party was faring well at the polls, there was a close alignment of interests between chief executive and party. Under those circumstances, impeachment will rarely occur.

In the context of pure presidential systems, we have been able to locate one case of a party’s legislative majority voting to remove its own president. That was President Raúl Cubas Grau in Paraguay in 1999, who resigned after his impeachment by the Chamber of Deputies and just before a Senate vote that would have completed his removal from office.179 Cubas had won the party’s nomination only because the party leader had been jailed for a coup attempt. After a political assassination, another faction in his party attempted to impeach him in favor of its preferred candidate. This was successful after a period of political turmoil. Professors David Samuels and Matthew Shugart attribute the successful removal to a rare instance in which the party in question truly dominates the political scene and all levers of power.180 Intraparty fights thus substitute for the party-against-party competition that typically characterizes general elections.

At the same time, it is sometimes the case that a handful of members of a president’s party will join with others in an impeachment motion or threat. Such was the case when Richard Nixon resigned under threat of impeachment in 1974. Other examples involving impeachment or related mechanisms include Ecuador’s President Abdalá Bucaram in 1997 and President Jamil Mahuad in 2000, Venezuela’s President Carlos Andrés Pérez in 1993, and Guatemala’s President Otto Pérez Molina in 2015.181 In 2005, Ecuador’s Congress deposed President Lucio Gutiérrez from office for abandoning his duties, though it did not have to complete the impeachment process because of his resignation.182 In these cases, individual legislators’ interests plainly diverged from those of the party, perhaps because of differences in the consistencies represented by different legislators within the same party, or perhaps because of ideological divisions within the party.

Table 2 presents all the cases of successful removal of directly elected presidents through impeachment since 1990. It shows that the phenomenon is not unknown. But it is also not particularly common. It represents well less than half of 1% of all country-years in which impeachment might have occurred. The final column of Table 2 also offers a threshold piece of evidence of the impact of impeachment on the political system. It does so by tracking whether the country’s level of democracy improves or declines as a result of impeachment. To measure democracy, we use the widely utilized Polity2 index, which rates democratic quality on a 21-point scale ranging from –10 (total autocracy) to +10 (total democracy). By convention, scores of 6 or higher are considered full democracies. In the column on the far right, we track the change in the Polity2 rating from two years prior to impeachment to two years after.

Table 2: Successful Presidential Removals Involving Impeachment 1990–2017183

Country	Year	President	Polity Score Two Years Before	Polity Score Two Years After	Change in Polity Score from t-2 to t+2
Brazil	1992	Fernando Collor	8	8	0
Venezuela	1993	Carlos Andres Pérez184	9	8	-1
Madagascar	1996	Albert Zafy	9	7	-2
Peru	2000	Alberto Fujimori185	1	9	+8
Philippines	2001	Joseph Estrada	8	8	0
Indonesia	2001	Abdurrahman Wahid	6	6	0
Lithuania	2004	Rolandas Paksas186	10	10	0
Paraguay	2012	Fernando Lugo	8	9	+1
Brazil	2016	Dilma Rouseff	8	8	0
South Korea	2017	Park Geun-hye	8	8187	0

It is first worth noting that every country that successfully impeached a president remained a full democracy thereafter, in most cases without any change in the level of democracy. Even Madagascar, where President Albert Zafy was impeached in 1996, was still a full democracy a few years later. Peru’s impeachment of President Alberto Fujimori occurred as part of the restoration of democracy after his period of autocratic rule, and hence we see a significant positive jump in that case.

To this list could be added several instances in which impeachment occurred but the president was not removed, either because he or she was not convicted or because of extraconstitutional action. Of course, U.S. Presidents Bill Clinton and Donald Trump were examples of the former. Russian President Boris Yeltsin was impeached in the early 1990s but dissolved Parliament to stay in office.188 Similarly, Alberto Fujimori’s “self-coup” in 1992 was followed by a vote to remove him, but Fujimori had already dissolved Congress.189 Only the Russian case, which occurred when Russia could be characterized as a semidemocracy in the midst of a tenuous (and ultimately failed) transition from authoritarianism, led to a significant decline in the Polity score. Finally, we note that the ultimate results of the Brazilian case are still ambiguous: although Temer’s rocky tenure was followed by a competitive election, it remains to be seen whether, or to what extent, Jair Bolsonaro damages Brazil’s democratic structure.190 Early signs suggest that he may be effectively constrained by the legislature from implementing his most authoritarian plans, and his coronavirus strategy of denial has generated significant institutional pushback.191

There are also cases in which some kind of removal vote was held and the president departed, but not through impeachment.192 Only two of these led to a country’s level of democracy being eroded to fall outside the category. These were the 2014 impeac­hment of President Viktor Yanukovich, which led to him fleeing to Russia, in which the country dropped from a score of 6 two years before impeachment to 4 two years after, and the 2005 removal of Ecuador’s Lucio Gutiérrez, in which he fled the country before the legal proceeding was complete, and led the country to drop from a score of 6 to 5 in the Polity scale. In addition, there have been at least twenty formal impeachment attempts that did not reach the required threshold in the legislature since 1990. Of the cases, only one, the 2002 attempt against President Ange-Félix Patassé in the Central African Republic, led to a significant decline in the Polity scale, from a score of 5 (just below the conventional cutoff) to –1. In short, impeachment, whether or not it leads to removal, does not seem to negatively impact the level of democracy in a country.

How have these instances of removal, as well as the calls for removal that inevitably precede and surround them, emerged over time? Has there been a global moment of impeachment? Figure 1 provides a visual representation of the frequency of removal attempts since 1990, distinguishing calls by a party in parliament from formal motions of impeachment. The data shows a rather constant frequency of calls and removals around the world: Intriguingly, there is no uptick in the wake of the 2008–09 financial crisis, which is generally thought to have triggered a surge of po­pulist discontent and antidemocratic moves.193 Our prior was that this might have been an inflection point, triggering a wave of calls to remove elected leaders who had been forced by economic exigency to make unpopular decisions. Contrary to our expectations, however, there is no concentrated moment of global impeachments. We rather find a constant background drone of calls for impeachment.

Figure 1: Frequency of Calls and Removals194

It is instructive to set this alongside Figure 2, which des­cribes the relative frequency of democracies, autocracies, and hybrid regimes in the same period.

Figure 2: The Relative Frequency of Democracies, Autocracies, and Hybrid Regimes195

Comparison of these statistics and figures suggests that, in general and at least in terms of average effects, there is little evidence that either talk of impeachment or impeachment itself is unhealthy for a democratic political system. While there is one instance in which a president used the attempt at impeachment to overthrow the parliament, few would argue that Russia in the early 1990s was a democracy in any real sense; Yeltsin’s parliamentary opponents, moreover, were largely unreconstructed communists.196 In virtually every other case, impeachment was used to remove an unpopular leader and to recalibrate the political system. The relative ease of doing so, of course, depends on the substantive basis for removal and procedural aspects. We turn now to these topics.

B. The Global Grounds for Removal and Impeachment

This Section presents data on the formal rules invoked in removal. The first necessary step here is to map out the predicate conditions for removal. Table 3 summarizes the bases for removal of heads of state as of 2017, as set forth in national constitutional texts. (Note that many constitutions provide for multiple alternative grounds for removal and so there is no reason we would expect the percentages to sum to one.) We first look at the universe of the 149 constitutional systems that provide for some such procedure, and then examine a subset of presidential and semi-presidential democracies only. The vast majority of serious att­empts at impeachment have taken place in such countries.

Table 3: Basis for Removing Heads of State as of 2017197

Basis	Number of All Constitutions Providing for Removal (n = 149)	% of Constitutions	Presidential & Semi-presidential Democracies Only (n = 68)	% of Constitutions
Crimes	88	59%	43	63%
Violations of the Constitution	69	46%	19	28%
Incapacity	55	37%	19	28%
Treason	51	34%	19	28%
General Dissatisfaction	20	13%	7	10%
Other	29	19%	10	15%

As Table 3 illustrates, the most common basis for head of state removal is criminal misconduct. But apart from the United States, constitutions generally do not stipulate a requirement that crimes be “high.” Indeed, the phrase “high crimes” seems to be limited to constitutions directly influenced by the U.S. one, including those of Palau, the Marshall Islands, and the Philipp­ines. Of these, only the Philippines is a true presidential system.198 Its formulation is that the president and other high officials can be removed “on impeachment for, and conviction of, culpable violation of the Constitution, treason, bribery, graft and corruption, other high crimes, or betrayal of public trust.”199 At first glance, this seems quite similar to the language of the U.S Constitution. But the Philippine model of impeachment sweeps beyond the domain of criminal offenses to cover constitutional wrongs, as well as “corruption,” which might include but not be exhausted by formal criminal offenses. In this regard, even the Philippine model may sweep beyond the focus on individual criminality.

Beyond criminal offenses, violations of the constitution or the president’s oath of office are also common predicates for removal. A violation of the constitution may or may not be a crime in a particular political system, but it can obviously cut to the core of the constitutional order. Several countries in Africa stipulate that the violation must be “wilful.”200 As Professor Griffin has demonstrated, this possibility has gradually fallen out of constitutional practice in the United States (although the Johnson impeachment contains traces of the idea).201 That said, the “cheap talk” of impeachment echoing through Capitol Hill, today as bef­ore, contains the idea that removal of a president can be grounded on his or her constitutional infidelity.202

For our purposes, the most interesting category is what we label “general dissatisfaction” in Table 3, which covers a variety of situations. In many countries, more general grounds for removal blur the canonical distinction between presidential and parliamentary systems. For example, the Constitution of Ghana allows the president to be removed by a two-thirds vote in the legislative assembly for conducting himself in a manner “i. which brings or is likely to bring the high office of President into disrepute, ridicule or contempt; or ii. prejudicial or inimical to the economy or the security of the State,” as well as for reasons of incapacity or “violation of the oath of allegiance and the presidential oath.”203 This formulation blends two different grounds for removal: policy dissatisfaction and misconduct. Similarly, in Tanz­ania, the president can be removed if he “has conducted himself in a manner which lowers the esteem of the office of President of the United Republic.”204 Uganda’s Constitution allows the presid­ent to be removed for conduct that “bring[s] the office of President into hatred, ridicule, contempt or disrepute.”205 Honduras allows impeachment to proceed against “actions contrary to the Constitution of the Republic or the national interest and for manifest negligence, inability, or incompetence in the exercise of office.”206 These standards seem to spill over into the distinctly political bases of removal that characterize the parliamentary system, in which the head of government is dependent on the parliament for continued tenure. And like parliamentary systems, in many cases a legislature in a presidential system can remove the executive under relatively broad criteria.

In short, the implication of the case studies—that formal impeachment operates in practice as a vessel for the implementation of broad discontent with a particular regime—thus carries through in the text of many constitutions.

C. The Procedural Apparatus of Presidential Removal

Processes of removal typically involve multiple phases and different institutions. They are also characterized by different voting thresholds (sometimes within the same document) and time limits. These procedural details also sometimes vary along with the basis of the removal charge. This means that there is a good deal of complexity and variation. Table 4 provides the most common thresholds and actors for all independent countries as of 2017, ranked with the most frequent choice in each category at the top.

Table 4: Most Common Removal Procedures as of 2017 for All Constitutions (n = 194)

Rank Within Category	Who Can Propose? (n)	Legislative Threshold to Propose (n)207	Who Approves? (n)	Legislative Threshold to Approve (n)208
1	Lower house (100)	2/3 (53)	Court/constitutional council (61)	2/3 (54)
2	Both houses required (19)	Majority (20)	Lower house (50)	3/4 (10)
3	Court/constitutional council (9)	3/4 (7)	Upper house (17)	Majority (7)
4	Upper house (6)	3/5 (3)	Both housesrequired (17)	Other (3)
5	Cabinet (5)	Other (30)	Public through referendum (12)	-
6	Prime minister (4)	-	Cabinet (2)	-
	Public through recall (4)	-	-	-

Because of the complexity of the procedures, we organize our discussion by examining the roles of distinct constitutional actors in the proposal, approval, and confirmation of decisions to remove a president.

1. Legislatures.

Impeachment is, as Hamilton noted long ago, a predominantly legislative procedure. This means that it requires the aggregation of votes in one or more houses of a legislative body. Even if not called impeachment, head of state removal typically begins with action in the legislature, either in the lower house, the upper house, or both houses acting jointly. The most common vote threshold is a two-thirds rule. Whether or not the legislature proposes removal, it often has a role in approving the process. Again, the modal threshold is a two-thirds vote. There are some interesting variations. When the legislature is bicameral, for example, it is quite common for an upper house or the two chambers acting jointly to be the body to approve the motion to remove a leader. In Ireland, which has a nonexecutive president, two-thirds of either house can propose an impeachment, in which case the other house tries the case and can remove with a two-thirds vote.209 In a small number of cases, however, the legislative role is nondiscretionary. For example, in Fiji, the prime minister can propose the removal of the president. Whether removal occurs in the case of allegations of misbehavior is then determined by a tr­ibunal of three judges.210 Parliament is required to accept the judgment of the panel.211

Legislative procedures sometimes involve constitutionally mandated actions by legislative committees or other subparts of the chamber.212 In Tanzania, a written notice must be signed and backed by at least 20% of members of Parliament to be submitted to the speaker of Parliament at least 30 days prior to the sitting at which the motion of dismissal is to be moved.213 The next stage entails a Special Committee of Inquiry, whose membership is to be voted upon by members of Parliament.214 This is formed to investigate the charges levied against the president. During this period of inquiry, the office of president is deemed vacant. After receiving a report from the Special Committee, the National Assembly discusses the report, and can approve the charges by a two-thirds supermajority vote of all members of Parliament, in which case the president is removed.215

2. Courts.

The role of the judiciary in impeachment processes is complex and varied. At one end of the spectrum is the United States, where the Constitution gives no role to the courts beyond the chief justice’s function of presiding at trial of the president, and where the Supreme Court has signaled that the national judiciary should play essentially no role in impeachment procedures.216 On the other end of the scale is Honduras. There, until a 2013 amen­dment, the only body with the power to remove high officials such as the president was the country’s Supreme Court.

Most constitutions steer a middle course between these poles. More in keeping with a quasi-legal conception of impeachment, courts in many countries have a role in approving the removal of the president. But the judicial role in impeachment varies quite widely. In some cases, courts may be limited to ensuring that impeachment procedures are being carried out using the proper procedures by political actors. In others, such as the South Korean Constitution,217 courts may become involved at the final, trial-like stage of impeachment, after the legislature has made an initial decision as to whether impeachment is justified.218 A few constitutions also have multiple tracks for impeachment, some dominated by the courts and some by legislators. For example, the Colombian Constitution provides that if the president is impeached for “crimes committed in the exercise of his/her functions” or “unworth[iness] to serve because of a misdemeanor” the House impeaches and the final trial for removal is before the Senate.219 But where a president is impeached for a common crime, the final trial instead occurs before the Criminal Chamber of the Supreme Court.220

The Honduran case, as noted, is especially interesting bec­ause removal, before 2013, was concentrated only in judicial hands. High officials had the right to be criminally tried only by the Supreme Court; the court had the power to suspend them during the pendency of the trial and could remove them permanently upon conviction.221 The legislature had no textual removal power.222 These provisions were important during the constitutional crisis involving President Manuel Zelaya in 2009, which ended with a military intervention that deposed Zelaya.223 Most of the Congress and other political officials clashed with Zelaya over his plans to hold a referendum on a potential Constituent Assembly to replace the constitution; they alleged that his plans violated that law and constitution, and that he was disobeying judicial orders.224 Zelaya initially had a sizable amount of support from his own Liberal Party (one of the two major parties in the Congress at the time), but his intraparty support eroded sharply after his proposal for a constituent assembly and his forging of an alliance with Venezuelan President Hugo Chávez.225 However, the Congress was powerless to remove Zelaya from power directly, despite his loss of elite support.

Early one morning shortly before Zelaya had planned a “nonbinding” consultation on his constituent assembly proposal, the heads of the branches of the military arrived at his home and put him on a plane to Costa Rica.226 They later produced a supposed charging document and arrest warrant issued by the Supreme Court for his detention. Critics charged that it may have been backdated. At any rate it would not explain why Zelaya was put on a plane to Costa Rica, rather than being brought before the Supreme Court.227 The Congress met later that same day and declared the presidency to be “vacant”; following the rules in the constitution, it voted then to ratify the vice president to serve as president for the rest of Zelaya’s term. Most of the rest of the world deemed the incident a coup—for example, Honduras was suspended from the Organization of American States because of an “unconstitutional interruption” in the democratic order,228 a suspension that was lifted only after the next set of presidential elections in 2011.229

The highly legalistic nature of the Honduran impeachment process likely contributed to the problems experienced during the removal of Zelaya. First, the process required an indictment and conviction for an actual crime. It did not hinge, either formally or de facto, upon a broad and durable loss of support or very poor political performance on Zelaya’s part. Second, the process was technically in the hands of a court, rather than the legislature (although in fact, the final step in the removal was provided by the military). The country subsequently amended its constitution to create a legislative impeachment procedure in 2013, after Congress had (illegally) removed several members of the Constitutional Chamber of the Supreme Court.230 This suggests that reposing impeachment exclusively in the hands of a judicial body can present risks of elite capture and can squeeze out considerations of system-wide stability, preventing an exit even in situations where a system desperately needs one.

3. Public involvement.

The public has a role in impeachment in several countries. In some cases the public can approve the removal of the president by referendum. For example, in Gambia, the constitution allows a vote of no confidence by the legislature, proposed by one-third of members and approved by a two-thirds majority, in which case a referendum is called for the public to endorse or reject the decision.231 In the Austrian semi-presidential system, the legislature can call a referendum on the president’s impeachment, requiring a two-thirds vote of the upper house; if the referendum fails, the upper house is disbanded.232 In Colombia, members the public may file complaints against the president or other officials to the House of Representatives, which must then assess as the basis for any impeachment resolution before the Senate.233 A two-thirds vote in the Senate is also required.

In keeping with their populist rhetorical emphasis on the “people,”234 several of the so-called Bolivarian constitutions of Latin America also give the public a role in a recall procedure that shares some features with impeachment. In Bolivia and Ecuador, the public can initiate the revocation of the mandate of the president with 15% of registered voters proposing it.235 There are temporal restrictions: in Bolivia it can only be invoked after at least half the term has elapsed, while in Ecuador after the first year (and in both countries so long as at least one year remains in the term).236 Similarly, in Venezuela, 20% of registered voters can petition for a referendum to dismiss the president, after at least half the term has elapsed.237 Only one petition to remove the president can be filed during his or her term of office.238 The absolute number of voters in favor of dismissal must be equal to or greater than the number of voters who elected the president, and voters in favor of the dismissal must be equal to or greater than 25% of the total number of registered voters.239

Interestingly, nineteen U.S. states allow recall of elected governors. The procedure remains rare, having been used only three times in U.S. history, of which two led to successful removals.240 In 2012, for instance, Governor Scott Walker of Wisconsin was subjected to a recall election, but he retained office.241

Our case studies indicate another, more informal, mode of public involvement, namely mass protest. When large numbers of citizens come out into the streets, as happened in Brazil and South Korea, it can inform politicians about the depth of opposition to a leader, and in fact can itself be the crisis of governability to which impeachment responds.

D. Substitutes for Impeachment

We have focused so far on impeachment and cognate removal devices. But some constitutions contain other provisions that might be taken to be a substitute for the impeachment and removal of a president under certain circumstances. A censure procedure is one example (and in fact there have been four resolutions of censure against presidents in U.S. history242 ). For removal, the main alternative mechanisms are recall and removal for incapacity. In the United States, the latter is covered by the Twenty-Fifth Amendment, which gives “the Vice President and a majority of either the principal officers of the executive departments or of such other body as Congress may by law provide” the ability to certify to Congress “their written declaration that the Presid­ent is unable to discharge the powers and duties of his office.”243 When such a declaration is made, the president is removed; the vice president then assumes the powers of the presidency.244

The most obvious application of the Twenty-Fifth Amendment is in cases where the president is physically incapable of performing his or her duties because of complete incapacitation, say following a catastrophic stroke.245 But some recent commentary has suggested applying it on broader grounds like mental instability or obvious unfitness to hold office, arguing further that these grounds might apply to Trump.246 This broader application of the Twenty-Fifth Amendment (which remains as of this writing hypothetical) may render it a partial substitute for impeachment.

Ecuador offers a cautionary example of how a similar substitute for impeachment might be used to remove an incumbent president from office. The populist Abdalá Bucaram was elected to the presidency and took office in August 1996. His term would be a short one. His party was not the largest party in Congress and in Ecuador’s highly fragmented party system, did not hold anywhere near a majority of seats, making it hard for Bucaram to govern.247 In addition, he took office in the midst of serious economic problems, and shifted from his prior populist stance to propose highly unpopular neoliberal austerity measures to deal with the crisis.248 Many of his former allies, such as Ecuador’s indigenous parties and movements, abandoned him after he made these proposals.249

Bucaram nonetheless retained sufficient support to avoid impeachment and removal, which would have needed a two-thirds supermajority in the Congress. Faced with this problem, opponents of Bucaram turned to another constitutional provision providing that the president would “cease to perform his/her duties and shall leave office” for “physical or mental disability . . . so declared by the National Assembly.”250 The key point is that the “incapacity” clause could be activated by a majority of Congress, rather than the two-thirds supermajority needed for impeachment.251 By a vote of 44–34, the Congress declared Bucaram “mentally incapacitated” and removed him from power in February 1997, only about six months after he had taken office. Congress initially ignored the constitutional article governing succession and appointed the president of Congress, Fabián Alarcón, rather than the vice president, as the new national president, before technically complying with it and having the vice president serve as president for two days before resigning to make way for Alarcón.252

Bucaram was a colorful and unstable figure, who led a populist party with no clear ideology. He even embraced the seemingly derogatory nickname “the crazy one” (el loco).253 But he was not mentally incapacitated by any reasonable definition. His dubious removal deepened the political crisis in Ecuador and ushered in a period of extraordinary instability.254 Between 1997 and 2007, the country had seven distinct presidents, none of whom served a full constitutional term of four years. The incident may thus suggest concerns about the use of substitute mechanisms such as incapacity clauses to evade the normal rules and voting thresholds of impeachment. It suggests that those clauses may best be limited to a narrow set of circumstances in which their criteria are clearly met. Broader interpretations may destabilize the constitutional order because of the deep contestability and malleability of the category of mental incapacity. Furthermore, impeachment itself may need to be constructed in such a way that it is usable during a significant crisis, so as to avoid political actors from turning to either dubious alternatives such as in Ecuador, or clearly illegal steps such as the military intervention in Honduras.

E. The Consequences of Successful and Failed Removal Efforts

A successful impeachment process will typically lead to the immediate removal of the chief executive. Sometimes the presid­ent is suspended from serving after the initial vote, until the complete resolution of the process. Failed procedures can also have formal and informal consequences, however. For example, Tanzania also involves a feature of removal procedures that looks parliamentary in character.255 If at the end of the process the vote for removal fails, no new motion can be brought for twenty months. This means the president can be somewhat insulated from repeated abuse of the legislative procedures, an institutional design that resembles parliamentary systems, which typically protect prime ministers from votes of no confidence for a period after a failed attempt.

On the other hand, when an impeachment does go through, ouster may not be its sole effect. In addition to removal from offi­ce, constitutional impeachment provisions also envisage lifetime (or more limited) bans on holding public office, criminal punishment, and new elections. Consider these in turn.

A first important constitutional choice concerns whether an impeached executive may run again. Some constitutions ban a convicted president from ever running again for the presidency.256 In 2004, shortly after being impeached, Rolandas Paksas of Lithuania made clear his desire to run again in the next presidential election. In anticipation, the legislature passed a constitutional amendment prohibiting an impeached leader from competing again for office.257 Other constitutions impose shorter prohibitions. In Brazil, for instance, the constitutional text states an eight-year ban from office upon removal.258 This ban was imposed after Collor was removed. During the impeachment of Rousseff, the Congress was allowed to hold two separate votes, and ended up removing her from office but not imposing a ban on future runs.259

A second question concerns how impeachment relates to criminal prosecution and punishment. As in the United States, the process of prosecution is often separated from that of removal from office. For example, in Colombia, although the Senate cannot impose criminal charges, it can refer the matter to a court for prosecution after removal.260 Indeed, many constitutions allow for prosecution after leaving office. Collor, for example, was tried for corruption in Brazil after he was out of office but acquitted in 1994 by the Supreme Court for lack of evidence.261

A third important design decision about removal relates to whether or not it triggers a new election. In the United States, of course, removal leads to the vice president assuming the office of the presidency for the remainder of the term. However, it is worth noting that this is neither necessary nor particularly common. For any political system in which the president is indirectly elected, for example by parliament, the removal of the president will typically trigger a new selection process.262 But remarkably, it is far more common for presidential and semi-presidential systems to respond to the removal of a president with new elections rather than to allow a substitute to serve out the remainder of the term. Our analysis shows that, of presidential or semi-presidential constitutions which speak to the issue, 74% (51 out of 69) provide for new elections within a short period, while the remainder provide for a vice president or other official completing the term.263 In other words, the South Korean model described in Part I.A is more common than the U.S. one described in Part I.E. We consider the normative benefits of this design in the next Part.

F. Conclusion

Our large-N analysis of constitutional provisions supports three broad conclusions. First, most constitutions allow impeachment for the commission of crimes, although many sweep beyond this to allow removal for a range of grounds including violations of the constitution or poor performance in office. In many systems, impeachment is not just about removing criminals, but also has broader purposes such as removing politically weak presidents who would otherwise be unable to govern effectively. There is also variation in the process of removal. Legislatures are the modal vehicle for removing a president, but courts often have a (limited) additional role, especially in approving findings of other institutions.

Second, there are some empirical regularities in the use of impeachment: (1) impeachment is exceedingly rare; (2) the risk of misuse of “maladministration” as a ground of impeachment seems to be quite small; (3) impeachment is almost always channeled through partisan politics; and (4) impeachment is usually a response to systemic problems rather than, or in addition to, ind­ividual presidential malfeasance. These patterns do not appear to have changed over time (although the universe of cases is also small, and hence care must be taken in extrapolating beyond those cases). They also appear unaffected by exogenous shocks such as the 2008–09 economic crisis and the austerity regimes that followed it.

Third, the substantive predicates for removal and the choice between different procedures likely interact. In criminal law, it is generally recognized that regulators can choose between substantive and procedural law as levers to make convictions either easier or harder.264 A simple index capturing their interaction is presented in Table 5. We separate out two dimensions: the substantive standard required for removal and procedural difficulty. The substantive standard is coded as high, medium, or low depending on whether there is no basis for removal other than illness (high), removal is restricted to serious constitutional violations or crimes (medium), or, alternatively, the constitution allows for more general removal (low). We code silence on the substantive standard as equaling the most difficult level of removal. To calculate the difficulty of the procedure to remove, we draw on the idea of institutional “veto players,” or “individual or collective actors whose agreement is necessary for a change in the status quo.”265 We code an impeachment as “easy” if it requires fewer than the modal number of decision-makers to effectuate (two), “intermediate” if it has two decision-makers with no higher than a two-thirds vote threshold in one house, and “difficult” if it inv­olves more than two decision-makers.266 In addition, if two decision-makers are involved, the process is considered difficult if it involves more than the modal legislative super majority of two-thirds, or two-thirds majorities in more than one house of parliament.

Table 5: Index of the Difficulty of Impeachment

	Substantive Standard	Procedural Difficulty
Low	35	47
Medium	92	60
High	9	29

These two margins of impeachment difficulty are positively correlated at a level of 0.27. This means that, in general, countries that have lower thresholds for predicate acts also tend to make the process of removal easier, although the correlation is not perfect. Figure 3 below presents the range of different countries in sequence on the horizontal axis in terms of the level of predicate and procedural difficulty, with the vertical axis measuring difficulty on our index.

Figure 3: Distribution of Index of Impeachment Difficulty

As the figure demonstrates, most countries tend to have similar levels for the two variables. Moreover, there is a clumping of countries in the center of our index. The United States and South Korea would fall in the center of the figure; Brazil in turn has relatively lax standards but procedures that are in the middle category. Overall, this analysis suggests that constitutional design at the global level has converged on a moderate level of difficulty for the removal of chief executives, with a few countries to be found at each of the extremes.

III. Theorizing Impeachment Design: Improvements and Pitfalls in the United States and Beyond

The analysis so far has developed an empirical account of the design and practice of impeachment in constitutions around the globe. In this Part, we turn to normative implications of our analysis. What role should impeachment of a chief executive play in a presidential system? And given that role, what implications follow for constitutional design, either in terms of the substantive standard for removal or in terms of its procedural channels? We focus here largely on ways in which the design and practice of impeachment in the United States might be improved in light of comparative experience. We hence bear in mind normative values such as democratic governance and the rule of law that should be widely accepted across the political spectrum. Some of our suggestions (like broadening the substantive standard for impeachment or giving some role to the judiciary) might be carried out through changes in practice. Others would probably or certainly require a constitutional amendment. In either case, we aim to use comparative evidence to contribute to the ongoing conversation about how presidential impeachment should be operationalized in the United States, as well as globally.

A caveat: We are mindful of the limited state of knowledge, the small sample of cases, and the endogeneity of outcomes to the ex ante choice of legal rules. For instance, we have largely analyzed textual provisions from constitutions in Part II, although our discussion of Brazil and other cases in Part I revealed that statutory frameworks can also matter. And, as we noted in the Introduction, the choice of impeachment-related rules influences both the rate of removal-worthy actions and events, and also the tendency of legislatures (and other actors) to engage in impeachment. There is no clean causal arrow running from constitutional design to impeachments. A focus on formal law also brackets a host of considerations related to the political environment and socioeconomic considerations. Indeed, we are skeptical of the idea of a single ideal or optimal design.267 Given variation in political, social, and economic conditions, we doubt that there is one “right” way of doing things when it comes to constitutional design. Institutions must fit their political and social context. At the same time, it would be bizarre to suggest that nothing could be learned from the global history of constitutional design and practice. Some institutional solutions are likely to incentivize disruptive behavior. Perhaps the best we can offer is how to avoid bad choices,268 and to infer likely downstream consequences from what is known of past practice. At the same time, there is probably a domain of easy cases where the dominance of impeachment is clear. In this spirit we proceed to assess the costs and benefits of the various institutional dimensions we have laid out, beginning with the overall conceptualization of the purpose of impeachment.

A. Conceptualizing Impeachment: Bad Actor Versus Political Reset Models

One might usefully distinguish two ideal types of impeachment following the analysis above. The first is what we call the “bad actor” model. Here impeachment is about removing serious criminals from office; elections ought to settle everything else. This is the model, as we have indicated above, that seems to inf­orm most modern U.S. rhetoric on impeachment.269 A second model one might call a “political reset” model. Here impeachment is not really about the individual criminality or unfitness of the chief executive, but instead a response to features of the contemporaneous political context. In this second model, impeachment can provide an exit from a situation of ungovernability, such as when a president has lost a massive amount of popularity and no longer has anything close to a governing coalition in Congress. The case studies of Part I provide some examples of how this can occur in the wake of exogenous economic and social shocks.

One of the major lessons of the case studies and empirical evidence reviewed above is that impeachment is not, or at least not only, about the bad actor model. Many of the crimes committed by impeached presidents are rather minor, and we doubt that the ten presidents removed since 1990 were the only ones engaged in criminal behavior during this period. Thus, theories of impeachment, such as those common in the United States, that focus exclusively on individual wrongdoing may obfuscate some of the core functions played by impeachment in constitutional democracies. Impeachment will always be about systemic problems in the political environment, either in addition to, or instead of, evidence of serious individual wrongdoing by the chief executive.

Electoral politics typically forms a hard constraint on executive removal. As we have seen in our case studies, even a bad actor will not be removed if he or she has sufficient support among legislators and the voting public.270 Indeed, without a very high level of opposition, presidents tend to survive in office regardless of the individually culpable act they have committed. Typically, a successful removal involves not just attack by the opposition party or coalition, but that the president’s former party, coalition, or allies turn against him or her. Consider the recent case of Trump as an example—he was virtually uniformly opposed by Democratic legislators but supported by virtually all Republicans—and his ability to maintain almost monolithic support from his own party allowed him to easily defeat removal in the Senate. All of this suggests that a chief executive is most likely to be successfully ousted when he or she is perceived to be linked to a governance situation perceived as fundamentally unacceptable across the partisan spectrum, rather than as a function of individual foibles. In such cases, the formal basis for impeachment may appear to be somewhat minor, but the real driving force is the loss of political support for the leader.

In Part I, we saw a number of different ways in which “fundamentally unacceptable” can be understood: In South Korea, the president’s reliance on a “shaman” and fortune teller was perceived to be inconsistent with minimally acceptable forms of lawful government.271 In Brazil272 and South Africa,273 the central question was the systemic corruption of the entire ruling class—and the need for some kind of “fresh start” in which law-abiding actors would putatively have a chance to mitigate corruption and graft. No doubt the way in which systemic problems are perceived, described, and evaluated will vary: the important point here is that absent a sufficiently shared sense of such a crisis, impeachment is unlikely to occur in practice—even if the formal terms of constitutional text sound more in the “bad actor” model. The United States, it should be noted, is no exception to this trend: efforts to impeach Presidents Bill Clinton, Barack Obama, and Donald Trump all failed in part because there was an insufficient consensus on the systemic nature of the problems associated with their presidencies.

In some cases, of course, individual wrongdoing formed a key predicate for impeachment. But even then, there were also significant problems in the political system that made removal of the chief executive likely. South Korea offers the best example. Park Geun-hye was implicated in serious criminal wrongdoing that resulted in a lengthy prison sentence. But impeachment was also facilitated by a political context in which she had become deeply unpopular and had lost support from members of her own party.274 South Africa, although again not technically an impeachment, is another instance where individual wrongdoing by Zuma underpinned a forced resignation that was made possible because of fissures in the ruling ANC over systemic problems of state capture.275 In these cases, the identification of the president as a bad actor is at the core of an ouster, although a troubled political context must still exist for the removal to occur.

In contrast, in some other cases and constitutional designs, impeachment does not respond to serious individual failings of chief executives. It is almost exclusively about the political context. Consider Brazil and Paraguay: In the former, Rousseff was implicated at most in failing to suppress a corruption scandal engulfing the entire political class, and more directly in budgetary accounting “tricks” engaged in by administrations before her.276 In the latter, the allegations against Lugo were aimed squarely at his performance in office, not at individual wrongdoing. Both constitutions have broad, political standards for impeachment, and removal occurred because of weaknesses in the chief executives that made it possible.277 In these cases, in other words, individual wrongdoing or the removal of bad actors is at most incidental to a process driven by broader concerns.

Is the broader model of impeachment that we present, focused on systemic rather than individual wrongdoing, a good or a bad thing from a normative perspective? This is a difficult question to answer. But we are inclined to answer the question, at least tentatively, in the affirmative.

It is useful to develop the case for a political reset conception of impeachment by situating that conception within the contrast between presidential and parliamentary systems of government.278 Recognition of the political reset paradigm, in effect, is a way of seeing how the two forms of governance can converge toward each other in practice, even as they remain formally distinct. Parliamentarism, according to one fairly representative definition is “a system of government in which the executive is chosen by, and responsible to, an elective body (the legislature), thus creating a single locus of sovereignty at the national level.”279 The essence of parliamentarism is a logic of mutual dependence between the legislative and executive branches: either institution has the ability to bring down the other.280 The government can dissolve the legislature. Likewise, a legislature can bring down the government by voting no confidence in it. In contrast, presidentialism has a logic of mutual independence, where the president and the legislature are separately elected for fixed terms. Under ordinary conditions neither has the ability to curtail the term of the other.281

Impeachment is an exception to this rule of independent and durable electoral mandates. Correlatively, it is conceptualized as a rare and exceptional measure, one that violates the usual structural independence of the two institutions. The opposite is supposed to be true in a parliamentary system. Indeed, the very fact that in parliamentary systems the legislature may generally vote no confidence in the government for any reason at all is indicative of the very different conception of legislative/executive relations as mediated through removal protocols. The latter, of course, are quite distinct from appointment-related arrangements. Arrangements for executive removal are a core element of the distinction between presidential and parliamentary systems. Interestingly, although some prior work has explored various ways in which presidential systems can evolve parliamentary features (and vice versa), this line of inquiry has not focused on removal of the chief executive, which is seen as a canonical distinction between the two types.282

The contest between presidentialism and parliamentarism has spurred an enormous literature with few clear conclusions. At minimum, the performance of each regime type depends on many other variables, including the nature of the political party system in which the regime is embedded.283 That said, one of the core arguments against presidentialism rests on the personalization and centralization of power in a single individual, the president. Some work has argued that this may pose a heightened risk of moves toward authoritarianism.284 Others have pointed out that especially when the president and legislature are dominated by different parties or movements, presidential systems may calcify into policy gridlock.285 Gridlock may feed perceptions that government is ineffective, or stimulate expansions in executive power that spark moves toward authoritarianism. A well-known example is Chile in 1973, where the administration of President Salvador Allende faced a hostile Congress, navigated around that Congress through increasingly aggressive decree powers, and amidst rising tensions was removed in a military coup that led to a brutal dictatorship.286 In a well-functioning parliamentary system, a government that lacked at least implicit parliamentary support would likely fall in short order, leading either to a new government that had such support, or new legislative elections.287

Nevertheless, some of the criticisms of presidential systems can be blunted by tweaking the design and practice of impeachment to avoid or mitigate the kind of deep crisis to which presidentialism sometimes seems to succumb. The case studies highlighted in Part I suggest that impeachment can play instead the same sort of resetting function that is played by votes of no confidence, or dissolutions, in well-functioning parliamentary systems. It does not follow that impeachment should be easy, or become routinized. Indeed, we think it is likely to remain an off-the-equilibrium-path outcome in most systems. In comparative terms, impeachment is a relatively rare, and potentially traumatic, event in essentially all presidential democracies. But so too are no-confidence motions in parliamentary democracies, as they tend to be deployed with “great discretion.”288 Rather than thinking of impeachment as distinct and more infrequent than a no-confidence motion, impeachment can be conceptualized as a similar tool for navigating between the rigid and undesirable extremes of a strict rule of fixed-term electoral independence for the executive and the complete reliance on legislative confidence. At least in certain kinds of governance crises, permitting the legislature to remove the executive may ameliorate some of the most problematic features of a presidential system of government. Exactly which such crises should trigger use of impeachment is primarily a question for constitutional designers and practitioners in individual countries. But the core point here is that the political reset conception of impeachment should be recognized as a useful adaption that may ameliorate one of the weaknesses of presidentialism.

Here is an example of how such a political reset might work. Impeachment may make outsider presidents who are weakly tied to the existing party system in a country especially vulnerable to removal. These kinds of figures may be more likely to lose the support of a coalition in Congress that is sufficient to ward off impeachment, or to have support erode from within their own nominal party. Several of the presidents removed under threat of impeachment or similar mechanisms over the past several decades—Lugo in Paraguay, Gutierrez in Ecuador, Zelaya in Honduras, and Collor in Brazil—constituted such figures. But notice that these kinds of actors may be especially problematic for the health of a presidential system. Because of their weak ties to existing parties, they may be less willing and able to get things done through ordinary political routes and may hence turn to more problematic paths as alternatives.289 Outsiders may also be more likely to use populist modes of governance that undermine the democratic order.290 Perhaps then, the greater vulnerability of political outsiders to impeachment is a feature, not a bug, of the model.

Of course, moving toward a political reset conception of impeachment is not without certain costs. One is that impeachment may exacerbate rather than defuse political crises. Consider Brazil, where a number of commentators have argued that the removal of Rousseff drew Brazil deeper into a crisis of political distrust and corruption.291 The removal of Rousseff further destabilized the political system, leaving the country with a weak, corrupt, and unelected successor, and creating a vacuum in which the hard-right populism of Jair Bolsonaro could take power in 2019. We recognize the force of this point, although we argue (as emphasized below) that it can be partially dealt with through other procedural designs, such as requiring that impeachment trigger new elections immediately rather than allowing automatic accession by a preset successor like a vice president.

A related problem is that a broader use of impeachment could increase political polarization, thus begetting cycles of ever more-frequent removals. Some commentators have suggested such a risk of the “normalization of impeachment” in the United States.292 But, as shown in Part II, while countries do seem to differ in the frequency with which they resort to impeachment, successful removals are fairly rare everywhere. Even where a more flexible standard is employed, impeachment has not been succe­ssfully used with great frequency. As shown above, impeachment can be initiated as a purely partisan exercise supported by one party or movement, but it rarely succeeds unless it has substantial cross-partisan support, often from within the president’s own party. In contrast, there are countries where irregular removals of presidents are commonplace and highly destabilizing.293 Encouraging the use of impeachment as a removal tool may in fact lessen reliance on these alternatives, whether dubious legal substitutes or extralegal maneuvers such as military coups. But there is no reason to think that successful impeachments beget a destabilizing dynamic over the long term; at least tentatively, we think that this also may imply that the current global rate of impeachment is too low.

B. The Substantive Standard for Impeachment

Understanding impeachment as an exit from political crisis suggests that the standard for impeachment should be framed in terms that are more political than legal, or at least which leave room for ambiguity. The danger of conceptualizing impeachment in purely legal terms, say by tying it to a finding of criminality by the president, is that this may stop political actors from being able to impeach in some cases where there is truly a situation of gridlock with an unpopular leader, but legislators struggle to identify a clear crime committed by a president. If legislators respond by stretching the meaning of the criminal law, they may undermine public confidence in the process. If they fail to take action because of legalistic doubts or because of the threat of judicial intervention, they may prolong the crisis. The Honduran case explored above perhaps best illustrates the risk.294 Substantively, a president in Honduras could only be removed from power for committing crimes. Procedurally, the legislature played no role in removal, which was delegated entirely to the Supreme Court. The result was a process that was too rigid to remove an exceptionally crisis-ridden and ineffective president who had lost the support of his own party. This in turn led to a military removal. In effect, the opposition to Zelaya struggled to identify prosecutable crimes that he had committed, and had to make awkwardly framed arguments to square their purpose with the available legal tools.

The U.S. standard for impeachment, “Treason, Bribery, or other high Crimes and Misdemeanors,” is notoriously ambiguous, as we have noted, and debate continues to rage about whether the term should be limited to certain classes of prosecutable crimes, or should take on a broader meaning.295 As practiced in the modern period, however, it is relatively focused on crimes.296 So read, the U.S. standard is subject to the same critique as the Honduran model. As one of us has argued in another context, there is a risk that the policy disagreements that are endemic to a polity will be treated as points of legal infidelity. Rather than domesticating the polity’s endogenous conflict, the law’s decision to treat policy disagreements as a justification for punishment might escalate the stakes of political disagreement.297 As the Johnson impeachment and the Clinton impeachment respectively illustrate, it inexorably impels a president’s political opponents to reframe minor legal disagreements as matters of deep infidelity or to manufacture criminal offenses about the sexual veniality and vanity of the president. To paraphrase Raymond Carver, politics—and the deep politics of perceived structural crisis—is what we are really talking about when we talk about impeachment.298

In contrast to the Honduran and U.S. cases, the Brazilian and Paraguayan Constitutions (as well as many other constitutions around the world) supply the relevant institutional actors with a broader and more flexible concept of impeachable offenses. The Paraguayan text, which explicitly envisages impeachment for “bad performance” in office as well as common or high crimes, is perhaps the best example.299 The Brazilian formulation, which differentiates common crimes from vaguer and more highly political acts “against the Federal Constitution,” gets at similar ideas.300 The advantage of these formulations is that they may make it easier for impeachment to serve as a reset during a deep political crisis, even if evidence of individual criminality is scarce.

A relatively broad reading of “high Crimes and Misdemeanors” is of course plausible. In fact, it finds substantial support from both original understanding and scholarship. Such a reading would more easily sweep beyond criminal acts to include presidents who engaged in conduct that created systemic risks.301 Readings of the clause that focus on concepts such as grave “abuses against the state”302 or the “abuse or violation of some public trust”303 would do the job tolerably well. Whether a broad reading would be available given present partisan dynamics, though, is another question.

A similar analysis illuminates the appropriate voting threshold for impeachment. It is, to be sure, difficult to generalize about this issue. The consequences of any given voting threshold are very sensitive to context. But if a key function of impeachment is to serve as an extreme form of a no-confidence vote in situations of crisis, then allowing removal by a demanding (but not impossible) supermajority makes sense. In particular, actors probably should become vulnerable to impeachment when they lose high levels of support from their own parties and coalitions, something that comparative experience bears out. Not all presidents who lose such support are impeached, of course, but that is the kind of context in which impeachment becomes a realistic option. All this is to say that we think that most constitutions have answered the design question properly by using demanding (but not insurmountable) supermajority rules.304

C. One Standard or Many?

Our argument also has implications for the uniformity of impeachment standards across different types of elected officials. Consider the U.S. case. As normally glossed,305 the U.S. Constitution establishes the same standard for impeachment for several different types of actors—the president and vice president, lesser executive officials like cabinet secretaries, and federal judges.306 Some other constitutions around the world adopt the same uniform approach. But others, like that of Brazil, adopt different substantive standards (and different procedures or institutions) for the impeachment of different kinds of actors.307

The differentiated approach adopted by Brazil seems to us the superior one, and the uniform U.S. approach deeply problematic. A single impeachment standard bundles together several different types of actors who have different constitutional functions, distinct democratic mandates, are subject to different alternative forms of accountability, and whose removal will precipitate radically divergent repercussions. The president is the sole head of a branch of government and generally remains in place at least until the next fixed election is held. Cabinet secretaries and similar officials often have much more fluidity, since they can often be removed at will by the president. Judges, of course, also serve fixed terms (for life in the United States), but generally have no electoral accountability and serve in positions where political independence is often deemed essential. Lumping all these different actors together makes little sense. The standard for impeachment should be tailored to the function played by each actor, and not automatically set the same for all officials.308

For example, we have argued that removal of presidents will sometimes be desirable to allow a reset during a deep political crisis. This suggests a relatively broad, ambiguous standard for removal of presidents, perhaps incorporating poor performance in office, abuse of power, or similar notions. In contrast, allowing removal of judges on similarly broad grounds may give the political branches too much power to retaliate against the judiciary. For this reason, it may make sense to tether judicial removal to a narrower set of grounds tied to serious criminality. Furthermore, in the United States, judges are subject to other sanctions for wrongdoing, including judicial discipline and criminal prosecution. Cabinet officers too can be criminally prosecuted. At least under current understandings put forth by the Office of Legal Counsel, the president is not subject to criminal prosecution while in office, which in our view weighs toward a lower threshold for impeachment, as it is the only available mechanism for accountability in between elections.

D. The Process of Impeachment

It is even harder to generalize about the process of impeachment, for which our case studies and empirical evidence show great variation. However, one core point that we draw from the evidence is that process should follow from the purpose of impeachment. The set of considerations that may be dominant where the core purpose of a removal is cleansing a bad actor may be different from the core purpose where the impeachment responds to a systemic failure. Relatedly, different institutions may usefully play different roles during an impeachment.

Contrary to the U.S. process for impeachment, our analysis in Part II demonstrated that many constitutions involve actors other than the legislature in presidential removal. Some go so far as to adopt different kinds of impeachment procedures for different offenses. In some countries, for example, allegations of criminal wrongdoing involve the courts, while those alleging poor performance in office or similar political allegations involve only the legislature.309 This represents a rough sorting of cases in which the bad actor model is dominant, and those in which the removal is mainly about political reset.

It seems to us very difficult to take a firm normative position on this issue of differentiated standards, which may provide some benefits but also may create new problems, such as determining how an allegation should be routed between processes. Still, comparative exploration of process helps to show how impeachment may help to build or undermine the credibility of allegations, and thus how process and substance interact. The South Korean and South African removals were greatly aided by the presence of independent institutions that would investigate facts and weigh the credibility of allegations—the Constitutional Court and Public Protector, respectively.310 In South Africa, the Constitutional Court helped to lend additional credibility to the Public Protector by ruling that its report was legally binding on the political branches.311 At any rate, the independence and reputation of both institutions seemingly helped to enhance the credibility of the removals.

A comparison to the contemporary U.S. context is instructive. Here impeachment investigations are often left to Congress itself, which may undermine confidence in the findings. Two recent impeachment attempts, of course, flowed directly out of independent investigations, the Starr investigation into Clinton and the Mueller investigation into Trump. Special counsel Robert Mueller was not well-insulated from the president, raising concerns about potential interference or firing, and was not free to make recommendations free of constraining Justice Department legal positions.312 Similar concerns materialized during the investigations of President Nixon.313 Thus, one problem is that the U.S. constitutional design has a dearth of constitutionally insulated institutions analogous to the Public Protector in South Africa. Of course, even seemingly independent investigations that have been involved in recent impeachments have not been trusted and instead have been portrayed as politicized. The Mueller investigation, for example, has been widely derided by the right (not least by the President himself) as a partisan “witch hunt.”314 The Starr investigation, which was carried out by a statutorily independent Special Counsel,315 received a similar reception on the left.316

The broad point is that U.S. constitutional design and scholarship could benefit from thinking of the ways in which other institutions might play a useful role in carrying out specialized functions (such as fact-finding) or in enhancing the credibility of removals, especially in cases where they are tied to the finding of criminal wrongdoing (or something similar) by a sitting president. Similarly, it may be worth thinking of ways in which institutions might be used to spur the political branches to take their responsibilities seriously when confronted with the fruits of independent investigations, as the Constitutional Court did in South Africa.

E. The Role of Courts and Due Process

The role of courts is an especially interesting issue in impeachment processes. As we surveyed above, the U.S. constitutional text is silent on the role of the courts during impeachment, with the exception of noting that the chief justice presides over Senate trial of the president of the United States, a role that was understood by both Chief Justice William Rehnquist during the impeachment of Clinton and Chief Justice John Roberts during the impeachment of Trump to be essentially ceremonial.317 U.S. courts have stayed out of impeachment processes.318 The United States is not alone in taking such a position; the Paraguayan Supreme Court, for example, took a similar stance after the impeachment of Lugo.319 But the comparative evidence shows that the posture of no judicial involvement is one end of a broad spectrum. In some cases, such as South Korea, courts play a formal role in the impeachment process, often as the final step in the process after an initial political determination has been made.320 In other cases, like Brazil, courts may accept some role of judicial review, for example to determine whether the procedure for impeachment has been followed or the substantive standard has been met.321 In the rare extreme, as in Honduras prior to 2013, courts may be imbued with the sole power of removal.322

The comparative evidence is too thin to establish exactly how to fix the best point on this spectrum for any given polity. This likely depends on context. As we have noted, the Honduran solution of placing removal power exclusively in the hands of the courts seems deeply problematic, because it ignores the essentially political nature of removal. It required that the president be charged with a crime before impeachment proceedings could even begin. It may even have allowed Zelaya to cling to power for a long time after he had lost the support of virtually the entire political elite, including his own party. The legislature lacked any way to initiate removal proceedings against him, even though they complained repeatedly about his conduct.

Aside from this extreme position, though, a range of forms of judicial involvement may work at least tolerably well. In the South Korean case, the role of the Constitutional Court in confirming the removal of the President may have helped to build confidence in the outcome by showing that the removal was not merely the continuance of ordinary politics by other means. There is an obvious danger in a court playing this kind of confirmation role: it may stymie removals that are politically necessary but harder to justify legally. The countervailing benefit of models like the South Korean one is that direct judicial involvement of this type may bolster the credibility of impeachment processes and make it harder to argue that they are just a politically motivated, “constitutional coup,” as in Brazil and Paraguay.

The Brazilian Supreme Court was heavily criticized for its various interventions into the Rousseff impeachment.323 But the Court’s interventions, as well as those of the South African high court, may still illuminate the ways in which a judiciary could potentially shape impeachment without outstepping their reach. The Brazilian Court did not adjudicate any direct attacks on the impeachment process. Rather, it issued several rulings that shaped its procedures. The President of the Court as presiding official of the impeachment trial in the Senate also issued rulings that shaped the process. More powerfully, the South African Constitutional Court’s interventions had the effect of keeping the channels of political redress for corruption open, and ensuring that Zuma could not bury charges against him.324 It provides a salutary model of a high court effectively and deftly defending constitutional democracy under the rule of law, even though the court there made no substantive decisions on the merits of Zuma’s removal.

In short, there are forms of judicial involvement in impeachment that do not immediately risk the over-legalization trap that makes impeachment unduly rigid. In this sense, the strict U.S. position of permitting virtually no impeachment controversy to be justiciable may be unnecessary, and perhaps even undesirable.

Relatedly, our analysis has implications for due process arguments of the kind made during the removal of Lugo of Paraguay325 or the trial of Trump in the U.S. Senate. From the perspective of the individual official, these seem reasonable claims because the transparency of a process, and its perceived fairness, seem potentially important to popular acceptance and legitimacy of the result. But at the same time, invocations of due process, or similar concepts, during impeachment procedures should be used with great care. It is not just, as the U.S. Supreme Court has suggested in Nixon v. United States,326 that an impeachment trial in the Senate is by necessity quite different from a standard criminal trial. It is also that it may serve a purpose of political reset that goes well beyond the character of the individual president, and instead goes more to the political context within which that president is working. In such structural debates, individual claims to due process ought to have less weight.

F. Impeachment and the Hard “Political Reset” of a Democratic System

In many systems, impeachment works as a hard political reset of the democratic process by triggering new elections upon removal. The South Korean system provides an example: it requires a new election within sixty days of removal, resetting the schedule of presidencies.327 Indeed, we emphasized above in Part II.E that in most systems, impeachment triggers a new election. This design avoids one of the key problems with the U.S., Paraguayan, and Brazilian systems (among many others): namely that removal of a president means he or she will be replaced by his or her own vice president, usually of the same party and political persuasion, who then completes the full term. Restarting with a new election is closer to the design of a parliamentary system, and allows the system to avoid gridlock, which as noted above is one of the risks of a presidential system.328 Allowing the constitutional order to hit the reset button in this fashion seems to us like a useful tool.

In contrast, allowing the vice president to ascend to power once the president is removed, as in the United States, seems a problematic design. Allowing a preselected official of the same political coalition to ascend to power for the remainder of the ousted president’s term invites abuse of impeachment by allies of the presumptive heir to the throne, and it may at any rate prolong the crisis by preventing a true political reset. The vice presidential succession model raises an obvious possibility of manipulation, where vice presidents or their allies seek to engineer the removal of presidents knowing that they will then ascend to power. This is not just a theoretical risk, but rather a likely description of dynamics in Brazil and Paraguay. In both countries, the successors (Michel Temer and Federico Franco) were affiliated with a different party than the president. In both, there were credible rumors that the vice presidents were plotting to remove presidents long before the impeachment.329 The description of events across both countries as a “constitutional coup,” despite the fact that formal impeachment procedures were followed, depended in large part on the fact that the movements appear to have been engineered by supporters of the two vice presidents as a way to gain political advantage, and as “reactionary movements” by conservative forces against progressive presidents.330

Further, allowing the vice president to ascend to power for the remainder of an ousted president’s term does not allow for a political reset. If the vice president is still somewhat close politically to the deposed president, impeachment may do little to resolve the political crisis. Imagine, for example, if Al Gore had succeeded Bill Clinton in 1999, or if Mike Pence had succeeded Donald Trump. In both cases, the new leaders would likely have continued many of the same political dynamics as the old. Even in cases where the vice president is distant from the old president politically (as in both Paraguay and Brazil), the successor is fairly likely to be embroiled in similar scandals as the old president. Temer, for example, was embroiled in a series of corruption scandals during his two-and-a-half-year interim presidency. Indeed, months following the end of his term in December 2018, he was arrested for alleged involvement in a corruption enterprise.331 In Paraguay, Franco similarly was embroiled in corruption-related controversies during and after his roughly one-year term in offi­ce.332 Furthermore, neither Temer nor Franco was popular: neither was likely to have won an election.

What should happen instead? We think the case studies of Part I suggest the superiority of the South Korean design, which allows impeachment to play a hard-reset function in cases of political crisis. Holding a new election shortly after an impeachment reduces the possibility of strategic initiation of a removal process. The relevant players will have more uncertainty about who will benefit from the impeachment. In particular, supporters of impeachment will need to worry that backers of the deposed president may win the subsequent election, especially if there is a perception that impeachment was undertaken abusively or for a narrow agenda. A new election is also more likely to create an exit from a political crisis, since a new president will be able to claim a renewed popular mandate.333

In this way, impeachment followed by new elections helps to ease the much-criticized rigidity of presidentialism by giving it a bit of the flavor of parliamentarism. In parliamentary systems, governmental crises and drastic losses of governmental support by the legislature are often, albeit not inevitably, resolved not just through a change in the executive cabinet, but through new popular elections. Even if new elections do not occur after a change in government, the new government is reliant on at least implicit legislative support. In contrast, the fixed electoral calendar of presidentialism generally prevents the holding of new elections as an escape valve, even during a deep crisis. Indeed, this fixed calendar is often seen as one of the biggest vulnerabilities of presidentialism, sometimes feeding deadlock and even leading to breakdown. The removal of a chief executive through an extraordinary process like impeachment seems to us to be a strong candidate for an exception to the general rule of a fixed calendar: it allows a new election to help provide an exit from a crisis, but at the same time, impeachment is too rare an event to lead to very frequent elections that might themselves destabilize the system.

Our argument for a new election rather than vice presidential succession following a successful impeachment leaves many important questions of constitutional design open. One is who should serve as interim president for the period of time before the new election is held. Elevating a relatively weak figure as in South Korea (or even an outsider such as a judge) may make sense in such a context; designers may also want to consider whether this caretaker should be eligible to run in the special election, particularly given its emphasis on resetting the political system. Another key question is how quickly a new election should be held. Again, the Korean solution of sixty days seems like a fairly reasonable solution. It gives political groups some time to organize, while ensuring that a reset happens quickly and limiting time for the new incumbent and his or her allies to consolidate their position.334

A third question, perhaps the most interesting, is whether a successful impeachment should trigger new elections just for the president, or for the legislature as well. Having an impeachment trigger legislative elections in addition to presidential ones may risk deterring even meritorious impeachments. And perhaps it seems illogical to “punish” the legislature for removing a corrupt, criminal, or incompetent chief executive. However, having impeachment trigger mutual dissolution may help to facilitate exit from a crisis by allowing voters to weigh in on the composition of both institutions that were involved. There are at least a few examples of presidential constitutions allowing the kind of mutual dissolution that is usually a hallmark of parliamentarism,335 and impeachment may again be a strong case for this kind of design. Moreover, triggering mutual elections may help to avoid abuse of impeachment by making legislators think long and hard about the consequences of presidential removal.336 Finally, it would avoid unintended consequences in terms of the political rhythm of the constitutional order, in that it would not lead to asymmetric terms as between presidencies and legislatures.

Conclusion

Based on a broad range of comparative evidence, we have argued that presidential impeachment in practice is about far more than removing criminals or other bad actors; it often serves as an exit from the crises that presidential (and semi-presidential) systems of government sometimes undergo when a leader has lost any semblance of a popular, democratic mandate but still has time to serve. We have also argued that such a conceptualization of impeachment is not only descriptively accurate in comparative terms, but also normatively desirable.

Our analysis has important normative implications for the debate and design of impeachment in the United States by clarifying the function of impeachment. Some of our findings shed new light on old problems. For example, we argue for a broader and more political understanding of “high Crimes and Misdemeanors” on consequentialist grounds. Others highlight overlooked problems in U.S. impeachment, which could be fixed through reinterpretations or constitutional amendment: that judicial abdication of any role during impeachment might be neither necessary nor desirable; that impeachment standards arguably should not be uniform across types of political actors; and that successful impeachments should trigger new elections, rather than simply allowing the vice president to succeed to the presidency for the remainder of an ousted chief executive’s term.

Following our normative recommendations could make impeachments more frequent, both in the United States and elsewhere around the world. Would this be desirable? As noted above, Brazil is one of the few countries in the world to have made fairly frequent use of impeachment in modern times, removing Collor through this route in 1992 and then Rousseff in 2016.337 While there are certainly many problems in modern Brazilian democracy, impeachment as an occasional tool to remove weak and ineffective presidents unable to forge a governing coalition in a fragmented Congress may sometimes ameliorate crisis, rather than exacerbating it. This would be truer, of course, if the design of the impeachment mechanism allowed for new elections and thus a full reset following impeachment, rather than automatic succession of the vice president.

We have also shown that there is no evidence to date that impeachment or attempted impeachment generates immediate destabilizing consequences, or is correlated with reductions in democratic quality. Increasing the availability of impeachments for systemic problems (although not for bad actors) thus holds the prospect of mitigating some of the worst aspects of presidential democracy without generating new costs. It is a constitutional possibility, in short, that seems well worth exploring.

• 1See, e.g., Emily Jane Fox, Democrats Are Paving the Way to Impeach Donald Trump, Vanity Fair (Dec. 15, 2016), https://perma.cc/GG2Q-5VU8.
• 2See, e.g., Alexander Bolton, White House Taking Impeachment Seriously, The Hill (July 25, 2014), https://perma.cc/FB4Z-64GX; Charlotte Dennett, The People v. Bush: One Lawyer’s Campaign to Bring the President to Justice and the National Grassroots Movement She Encounters Along the Way 12 (2010).
• 3See Arthur Goldhammer, The Yellow Vests Protests and the Tragedy of Emmanuel Macron, Foreign Affs. (Dec. 12, 2018), https://www.foreignaffairs.com/articles/france/2018-12-12/yellow-vest-protests-and-tragedy-emmanuel-macron.
• 4See Alan Taylor, Photos: A Venezuelan Opposition Leader Declares Himself ‘Interim President’, The Atlantic (Jan. 24, 2019), https://perma.cc/K8GW-PMSE.
• 5See infra Part I.
• 6On the relation of polarization to democratic crisis, see Jennifer McCoy, Tahmina Rahman & Murat Somer, Polarization and the Global Crisis of Democracy: Common Patterns, Dynamics, and Pernicious Consequences for Democratic Polities, 62 Am. Behav. Sci. 16, 16 (2018) (showing how popular polarization can lead to “gridlock and careening,” “democratic erosion or collapse under new elites and dominant groups,” or “democratic erosion or collapse with old elites and dominant groups”).
• 7See generally Cass R. Sunstein, Impeachment: A Citizen’s Guide (2017); Laurence H. Tribe & Joshua Matz, To End a Presidency: The Power of Impeachment (2018); Frank O. Bowman III, High Crimes and Misdemeanors: A History of Impeachment for the Age of Trump (2019); Allan J. Lichtman, The Case for Impeachment (2017); Elizabeth Holtzman, The Case for Impeaching Trump (2018). Earlier treatments include Charles L. Black, Jr., Impeachment: A Handbook (1974); Raoul Berger, Impeachment: The Constitutional Problems (1973); Laurence H. Tribe, Defining High Crimes and Misdemeanors: Basic Principles, 67 Geo. Wash. L. Rev. 712 (1999).
• 8Correlatively, much of the criticism of that work focuses on the “strategic” nature of the analysis. See Michael Stokes Paulsen, To End a (Republican) Presidency, 132 Harv. L. Rev. 689, 711–12 (2018) (reviewing Tribe & Matz, supra note 7). For a vigorous response to this criticism, see Laurence H. Tribe & Joshua Matz, To (Pretend to) Review Our Book, 132 Harv. L. Rev. F. 78, 79 (2018) (accusing Professor Paulsen of “repeatedly and egregiously misdescrib[ing] our thesis, reasoning, and conclusions”). The vehemence of this debate is indicative of how difficult scholarly discussion of impeachment can be.
• 9See Aziz Z. Huq, Legal or Political Checks on Apex Criminality: An Essay on Constitutional Design, 65 UCLA L. Rev. 1506, 1508 (2018) [hereinafter Huq, Legal or Political Checks] (discussing sources of ambiguity).
• 10U.S. Const. art. 2, § 4.
• 11See generally Peter Baker, Trump Team, Opening Defense, Accuses Democrats of Plot to Subvert Election, N.Y. Times (Jan. 25, 2020), https://perma.cc/Z2GH-ZUDZ; Nikolas Bowie, High Crimes Without Law, 132 Harv. L. Rev. F. 59 (2018).
• 12See Baker, supra note 11.
• 13For the Justice Department’s view, see generally A Sitting President’s Amenability to Indictment and Criminal Prosecution, 24 Op. O.L.C. 222 (2000).
• 14See, e.g., Nicholas Fandos, Republicans Block Subpoenas for New Evidence as Impeachment Trial Begins, N.Y. Times (Jan. 21, 2020), https://perma.cc/KP5B-3K9C.
• 15See Sunstein, supra note 7, at 88–116 (providing a characteristically incisive account of the Johnson and Clinton impeachments).
• 16See Baker, supra note 11.
• 17See Michael J. Gerhardt, The Federal Impeachment Process: A Constitutional and Historical Analysis 77–83 (3d ed. 1996) (discussing the scope of the impeachment remedy under the U.S. Constitution).
• 18See Daniel J. Hemel & Eric A. Posner, Presidential Obstruction of Justice, 106 Calif. L. Rev. 1277, 1279–80 (2018).
• 19But see generally, e.g., John K.M. Ohnesorge, Comparing Impeachment Regimes (Univ. of Wis. L. Sch. Legal Stud. Rsch. Paper Series, Paper No. 1468, 2020).
• 20See infra notes 171, 173.
• 21See Tom Ginsburg & Aziz Z. Huq, How to Save a Constitutional Democracy 4–5, 245 (2018) (discussing American Exceptionalism).
• 22Some of these questions are also likely to matter to authoritarian constitutions, which are also designed with the aim of minimizing agency costs. See Tom Ginsburg & Alberto Simpser, Introduction, in Constitutions in Authoritarian Regimes 1, 6 (Tom Ginsburg & Alberto Simpser eds., 2013).
• 23Indeed, during the Middle Ages the question of monarchical removal became a central problem for English constitutional theory; between 1327 and 1485, five English monarchs were deposed. See William Huse Dunham, Jr. & Charles T. Wood, The Right to Rule in England: Depositions and the Kingdom’s Authority, 1327–1485, 81 Am. Hist. Rev. 738, 760–61 (1976). Two centuries later, regicide was hedged with numerous defenses. See Amos Tubb, Printing the Regicide of Charles I, 89 Hist. 500, 509–10 (2004).
• 24The main exception is the South African example that we look at briefly in Part I below. The South African president is the executive in South Africa’s parliamentary system—he or she is selected by the Parliament and can be removed by the Parliament at any time via a vote of no confidence, as well as through an impeachment-like mechanism. We nonetheless include the recent example of President Jacob Zuma’s resignation because it sheds light on the ways in which nonlegislative institutions might facilitate presidential removal.
• 25See Robert Elgie, A Fresh Look at Semipresidentialism: Variations on a Theme, 16 J. Democracy 98, 99–101 (2005).
• 26For example, the Czech and Slovak states have nonelected presidents coexisting with elected parliaments. See Matthew S. Shugart, Of Presidents and Parliaments, 2 E. Eur. Const. Rev. 30, 31 (1993). On the increasing similarity between presidential and parliamentary systems, see Oren Tamir, Governing by Chief Executives 3–6 (unpublished manuscript) (on file with authors); José Antonio Cheibub, Zachary Elkins & Tom Ginsburg, Beyond Presidentialism and Parliamentarism, 44 Brit. J. Pol. Sci. 515, 537 (2014); Richard Albert, The Fusion of Presidentialism and Parliamentarism, 57 Am. J. Compar. L. 531, 549–55 (2009); Paul Webb & Thomas Poguntke, The Presidentialization of Contemporary Democratic Politics: Evidence, Causes and Consequences, in The Presidentialization of Politics: A Comparative Study of Modern Democracies 336, 340–41 (Paul Webb & Thomas Poguntke eds., 2005).
• 27See John D. Huber, The Vote of Confidence in Parliamentary Democracies, 90 Am. Pol. Sci. Rev. 269, 270–72 (1996) (describing vote of confidence mechanisms in eighteen democracies).
• 28See infra Part I.A.
• 29See infra Parts I.B and I.C.
• 30See infra Table 2.
• 31A point, of course, made in the classic essay by Professor Juan Linz. See Juan J. Linz, The Perils of Presidentialism, 1 J. Democracy 51, 56–57 (1990).
• 32See Stephen M. Griffin, Presidential Impeachment in Tribal Times: The Historical Logic of Informal Constitutional Change, 51 Conn. L. Rev. 413, 423–25 (2019).
• 33See Trial Memorandum of President Donald J. Trump at 1, In re Impeachment of President Donald J. Trump (U.S. Senate 2020) (quotations marks omitted), https://perma.cc/42Z4-WNDD (arguing that for this reason, “abuse of power” was not an impeachable offense); cf. Bowie, supra note 11 (canvassing arguments that crimes are not required).
• 34See Nixon v. United States, 506 U.S. 224, 228–29, 233–34 (1993) (rejecting a constitutional challenge to impeachment procedures of a federal judge as a “political question” beyond judicial competence).
• 35See generally, e.g., Lichtman, The Case for Impeachment, supra note 7; cf. generally Alan Dershowitz, The Case Against Impeaching Trump (2018). At the very least, this tide of books provides evidence of the impoverished imagination of book publishers when it comes to titles.
• 36Timothy J. McKeown, Case Studies and the Limits of the Quantitative Worldview, in Rethinking Social Inquiry: Diverse Tools, Shared Standards 139, 153 (Henry E. Brady & David Collier eds., 2004).
• 37See infra Table 1 (finding 10 removals in 213 attempts since 1990); see also infra Part II.A.
• 38S. Kor. Const. art. 65.
• 39S. Kor. Const. art. 65.
• 40S. Kor. Const. arts. 65, 71.
• 41S. Kor. Const. art. 111.
• 42Youngjae Lee, Law, Politics, and Impeachment: The Impeachment of Roh Moo-hyun from a Comparative Constitutional Perspective, 53 Am. J. Compar. L. 403, 411–12 (2005).
• 43See id. at 412.
• 44See id. at 418–19.
• 45See Martin Fackler, Recriminations and Regrets Follow Suicide of South Korean, N.Y. Times (May 24, 2009), https://perma.cc/CQH7-UDTY.
• 46See Lee, supra note 42, at 415–20.
• 47See id. at 419 (quotation marks and citation omitted) (translating the Constitutional Court’s opinion).
• 48See Hunbeobjaepanso [Const. Ct.], May 14, 2004, 2004Hun-Na1 (Hungong at 157) (S. Kor.); Chun-Yuan Lin, Court in Political Conflict: Note on South Korean Impeachment Case, 4 Nat’l Taiwan U. L. Rev. 249, 260 (2009).
• 49See Yul Sohn & Won-Taek Kang, South Korea in 2012: An Election Year Under Rebalancing Challenges, 53 Asian Surv. 198, 201 (2013).
• 50See South Korea’s President Fights Impeachment and Other Demons, The Economist (Dec. 17, 2016), https://perma.cc/4JEW-TX4E.
• 51See Gi-Wook Shin & Rennie J. Moon, South Korea After Impeachment, 28 J. Democracy 117, 119 (2017).
• 52See id. at 119–22.
• 53See Hunbeobjaepanso [Const. Ct.], Mar. 10, 2017, 2016Hun-Na1 (Hungong 1, 46–52) (S. Kor.).
• 54See Shin & Moon, supra note 51, at 119.
• 55See 2016Hun-Na1 at 59–60.
• 56See id. at 48.
• 57See Choe Sang-Hun, Park Geun-hye, Ex-South Korean Leader, Gets 25 Years in Prison, N.Y. Times (Aug. 24, 2018), https://perma.cc/42A8-GRCC.
• 58See S. Kor. Const. art. 68.
• 59See Shin & Moon, supra note 51, at 122, 124.
• 60See Marcus André Melo, Latin America’s New Turbulence: Crisis and Integrity in Brazil, 27 J. Democracy 50, 60 (2016).
• 61See Braz. Const. art. 85.
• 62For a particularly pugnacious account in these terms, see Teun A. van Dijk, How Globo Media Manipulated the Impeachment of Brazilian President Dilma Rousseff, 11 Discourse & Commc’n 199, 202 (2017).
• 63Braz. Const. 1988, art. 85.
• 64See Lei No. 1.079, de 10 de Abril de 1950, Diário Oficial da União [D.O.U.] de 12.04.1950 (Braz.).
• 65For example, the law is broader than the constitution in terms of which officials are subject to impeachment, and it imposes a different term—five years rather than eight—of potential disqualification from public office in the event of a successful impeachment and removal. See Lei No. 1.079 art. 68.
• 66See Melo, supra note 60, at 50–51.
• 67See id. at 60.
• 68Cultural expectations about women’s appropriate role in public life may also have played a role. See Omar G. Encarnación, The Patriarchy’s Revenge: How Retro-Macho Politics Doomed Dilma Rousseff, 34 World Pol’y J. 82, 83 (2017).
• 69See Braz. Const. 1988, art. 86.
• 70See Braz. Const. 1988, art. 86.
• 71See Braz. Const. 1988, art. 86.
• 72Andrew Jacobs, Brazil’s Lower House of Congress Votes for Impeachment of Dilma Rousseff, N.Y. Times (Apr. 17, 2016), https://perma.cc/57KR-JK4C.
• 73Simon Romero, Dilma Rousseff Is Ousted as Brazil’s President in Impeachment Vote, N.Y. Times (Aug. 31, 2016), https://perma.cc/74JJ-64WK.
• 74It is unclear under the constitution whether the Senate has the power to split the impeachment vote into two issues, one of removal and one of loss of political rights, since the text of the constitution seems to state that loss of political rights for eight years is an automatic consequence of impeachment and removal, although the text of Law 1079 contemplates two distinct votes. See Alexandra Rattinger, The Impeachment Process of Brazil: A Comparative Look at Impeachment in Brazil and the United States, 49 U. Mia. Inter-Am. L. Rev. 129, 155 (2018).
• 75See Juliano Zaiden Benvindo, Abusive Impeachment? Brazilian Political Turmoil and the Judicialization of Mega-Politics, Int’l J. Const. L. Blog (Apr. 23, 2016), https://perma.cc/2R66-ACHH.
• 76See Steven Wildberger, Brazil Supreme Court Sets Stage for President’s Impeachment, Jurist (Dec. 18, 2015), https://perma.cc/3LXL-Q5GM; Juliano Zaiden Benvindo, Institutions Matter: The Brazilian Supreme Court’s Decision on Impeachment, Int’l J. Const. L. Blog (Dec. 31, 2015), https://perma.cc/KFB9-RWQZ.
• 77See Benvindo, supra note 76.
• 78See id.
• 79See Theotonio Dos Santos, Brazil’s Controlled Purge: The Impeachment of Fernando Collor, 27 NACLA Rep. on Ams. 17, 20–21 (1993).
• 80See Carlos Pereira, Timothy J. Power & Lucio Rennó, Under What Conditions Do Presidents Resort to Decree Power? Theory and Evidence from the Brazilian Case, 67 J. Pol. 178, 185 (2005).
• 81See Thomas Skidmore, Collor’s Downfall in Historical Perspective, in Corruption and Political Reform in Brazil: The Impact of Collor’s Impeachment 1, 5–6 (Keith S. Rosenn & Richard Downes eds., 1999).
• 82See Rattinger, supra note 74, at 148 (quotation marks omitted) (quoting Thomas Skidmore, The Impeachment Process and the Constitutional Significance of the Collor Affair, in Corruption and Political Reform in Brazil, supra note 81, at 10).
• 83See id. at 149.
• 84See Benvindo, supra note 75. See generally Keith E. Whittington, Is Presidential Impeachment Like a Coup?, 18 Geo. J.L. & Pub. Pol’y (forthcoming 2020).
• 85See Vincent Bevins, Brazil’s President Dilma Rousseff Narrowly Reelected, L.A. Times (Oct. 26, 2014), http://www.latimes.com/world/mexico-americas/la-fg-ff-dilma
  -rousseff-election-brazil-20141026-story.html.
• 86See Emilio Peluso Neder Meyer, Judges and Courts Destabilizing Constitutionalism: The Brazilian Judiciary Branch’s Political and Authoritarian Character, 19 German L.J. 727, 732 (2018).
• 87See Melo, supra note 60, at 52.
• 88See, e.g., van Dijk, supra note 62, at 203 (describing Temer as “the figurehead of what was generally seen as a political coup”).
• 89See Brian Winter, Brazil’s Never-Ending Corruption Crisis: Why Radical Transparency Is the Only Fix, 96 Foreign Affs. 87, 90–91 (2017).
• 90Simone Preissler Iglesias, Martha Viotti Beck & Samy Adghirni, ‘Little Flu’ Can’t Hurt Him: Why Bolsonaro Still Shuns Lockdowns, Bloomberg (Mar. 30, 2020), https://perma.cc/V6Z4-DNSS.
• 91Vasco Cotovio & Isa Soares, Brazil’s Former President Calls for Bolsonaro to Be Impeached, CNN (June 2, 2020), https://perma.cc/7R5R-L7N4.
• 92Lisandra Paraguassu & Gabriela Mello, Brazil’s House Speaker Says Not the Right Time to Handle Impeachment Requests Against Bolsonaro, Reuters (July 24, 2020), https://perma.cc/WYX5-2WVC.
• 93Bryan Pitts, Rosemary Joyce, Russell Sheptak, Kregg Hetherington, Marco Castillo & Rafael Ioris, 21st Century Golpismo: A NACLA Roundtable, 48 NACLA Rep. on Ams. 334, 337 (2016).
• 94See id. at 337–38.
• 95Id. at 339. Serious impeachment discussions were also not new in Paraguayan political culture: President Raúl Cubas Grau resigned in 1999 after impeachment proceedings had been initiated, and Senator Luis González Macchi narrowly survived a Senate removal vote in 2003. See Aníbal S. Pérez-Liñán, Presidential Impeachment and the New Political Instability in Latin America 29–35 (2007).
• 96See Leiv Marsteintredet, Mariana Llanos & Detlef Nolte, Paraguay and the Politics of Impeachment, 24 J. Democracy 110, 112–13 (2013).
• 97See Daniel Jatobá & Bruno Theodoro Luciano, The Deposition of Paraguayan President Fernando Lugo and Its Repercussions in South American Regional Organizations, 12 Braz. Pol. Sci. Rev. 1, 7 (2018).
• 98Id.
• 99Id. at 8.
• 100These other four incidents included: (1) authorizing a demonstration in front of the Armed Forces Engineering Command, with slogans against the “oligarchic sectors,” (2) supporting several land invasions of large estates, (3) growing insecurity and an unwillingness to confront a guerrilla movement, and (4) signing the Ushuaia II Protocol of Mercosur in a way that violated national sovereignty. Id.
• 101See Magdalena López, Democracia en Paraguay: La Interrupción del “Proceso de Cambio” con la Destitución de Fernando Lugo Méndez (2012), 31 Cuadernos del Cendes 95, 112 (2014) (quoting Libelo Acusatorio, Resolución H. Cámara de Diputados No. 1431/2012 (2012)), https://perma.cc/U6UP-2QMX (our translation).
• 102Id. at 112–13 (our translation).
• 103The Constitution of Paraguay allows impeachment of the president and certain other high officials for “for bad performance in office, for crimes committed in exercise of their office or for common crimes.” See Para. Const. 1992, art. 225.
• 104Paraguay’s Impeachment: Lugo Out in the Cold, The Economist (June 30, 2012), https://perma.cc/LVP6-6C33.
• 105See Para. Const. 1992, art. 225.
• 106Marsteintredet et al., supra note 96, at 110.
• 107Id. at 114.
• 108See id. at 114–16.
• 109See Jatoba & Luciano, supra note 97, at 11–13.
• 110Press Release, Inter-American Commission on Human Rights, IACHR Expresses Concern over the Ousting of the Paraguayan President (June 23, 2012), https://perma.cc/5CZH-YA8U.
• 111Id.
• 112See Marsteintredet et al., supra note 96, at 114; Para. Const. 1992, art. 225.
• 113See Daniela Desantis & Didier Cristaldo, Paraguay’s President Faces Impeachment over Clashes, Reuters (June 21, 2012), https://perma.cc/5HVM-RYP3.
• 114See Accion de Inconstitucionalidad en el Juicio: “Fernando Armindo Lugo Mendez c/ Resolución Nro. 878 de Fecha 21 de Junio de 2012 Dictada por la Cámara de Senadores,” Corte Suprema de Justicia (June 25, 2012), https://perma.cc/5Z4M-3BH3 (holding that impeachment is an “exclusive competence” of the Congress (our translation)).
• 115See S. Afr. Const. 1996, art. 102.
• 116See S. Afr. Const. 1996, art. 89.
• 117Roger Southall, The Coming Crisis of Zuma’s ANC: The Party State Confronts Fiscal Crisis, 43 Rev. Afr. Pol. Econ. 73, 80 (2016).
• 118See Sarah Bracking, Corruption & State Capture: What Can Citizens Do?, 147 Daedalus 169, 170 (2018).
• 119See Michaela Elsbeth Martin & Hussein Solomon, Understanding the Phenomenon of “State Capture” in South Africa, 5 S. Afr. Peace and Sec. Studs. 21, 25 (2016).
• 120On corruption and the state-owned enterprises, see the insightful analysis in R.W. Johnson, Cyril Ramaphosa’s Poisoned Chalice, Standpoint (Mar. 28, 2018), https://standpointmag.co.uk/dispatches-april-2018-rw-johnson-cyril-ramaphosa-jacob-zuma-south-africa/.
• 121See Sapa, Nkandla: Nxesi Explains Why Security Upgrade Was Essential, Mail & Guardian (Dec. 19, 2013), https://mg.co.za/article/2013-12-19-nxesi-nkandla-security-upgrade-was-essential; see also Govan Whittles, Zuma Pays Back the Money—but Where Did He Get the R7.8-million?, Mail & Guardian (Sept. 12, 2016), https://mg.co.za/article/2016-09-12-zuma-pays-back-the-money-but-where-did-he-get-the-r78-million/.
• 122See United Democratic Movement v. Speaker of the Nat’l Assembly 2017 (5) SA 300 (CC) at para. 90 (S. Afr.).
• 123See Oriani-Ambrosini v. Speaker of the Nat’l Assembly 2012 (6) SA 588 (CC) at para. 62 (S. Afr.); Mazibuko v. Sisulu 2013 (6) SA 249 (CC) at para. 45 (S. Afr.); Democratic All. v. Speaker of the Nat’l Assembly 2016 (3) SA 487 (CC) at para. 38 (S. Afr.).
• 124See Mazibuko, (6) SA 249 at para. 72.
• 125Id.
• 126See Econ. Freedom Fighters v. Speaker of the Nat’l Assembly 2016 (3) SA 580 (CC) at para. 71 (S. Afr.).
• 127See Aziz Z. Huq, A Tactical Separation of Powers Doctrine, 9 Const. Ct. Rev. 19, 34–35 (2019) [hereinafter Huq, Tactical Separation of Powers].
• 128Econ. Freedom Fighters, (3) SA 580 at para. 2.
• 129See S. Afr. Const. 1996, art. 89.
• 130See Simon Allison, Jacob Zuma Narrowly Survives No-Confidence Vote in South African Parliament, The Guardian (Aug. 9, 2017), https://perma.cc/S7MA-3HS7.
• 131Huq, Tactical Separation of Powers, supra note 127, at 38.
• 132See Tom Ginsburg & Aziz Huq, Democracy’s Near Misses, 29 J. Democracy 16, 22 (2018) (quotation marks omitted).
• 133The first two articles of the U.S. Constitution establish and describe the impeachment process for the president, vice president, and other civil officers, in the following terms:

  Art. I, § 2, cl. 5:

  The House of Representatives . . . shall have the sole Power of Impeachment.

  Art. I, § 3, cl. 6:

  The Senate shall have the sole Power to try all Impeachments. When sitting for that Purpose, they shall be on Oath or Affirmation. When the President of the United States is tried, the Chief Justice shall preside: And no Person shall be convicted without the Concurrence of two thirds of the Members present.

  Art. I, § 3, cl. 7:

  Judgment in Cases of Impeachment shall not extend further than to removal from Office, and disqualification to hold and enjoy any Office of honor, Trust or Profit under the United States: but the Party convicted shall nevertheless be liable and subject to Indictment, Trial, Judgment and Punishment, according to Law.

  Art. II, § 2, cl. 1:

  The President . . . shall have Power to grant Reprieves and Pardons for Offenses against the United States, except in Cases of Impeachment.

  Art. II, § 4:

  The President, Vice President and all civil Officers of the United States, shall be removed from Office on Impeachment for, and Conviction of, Treason, Bribery, or other high Crimes and Misdemeanors.

• 134T.F.T. Plucknett, Presidential Address: Impeachment and Attainder, 3 Transactions Royal Hist. Soc’y 145, 145 (1953).
• 135See Sunstein, supra note 7, at 35; Berger, supra note 7, at 106–07.
• 136See Clayton Roberts, The Law of Impeachment in Stuart England: A Reply to Raoul Berger, 84 Yale L.J. 1419, 1432–34 (1975) (providing evidence for this, and also arguing that parliaments erred about the law).
• 137See Mithi Mukherjee, Justice, War, and the Imperium: India and Britain in Edmund Burke’s Prosecutorial Speeches in the Impeachment Trial of Warren Hastings, 23 Law & Hist. Rev. 589, 594 (2005).
• 138See Sunstein, supra note 7, at 51. This original formulation was subject to many changes. For instance, the Virginia Plan originally envisaged a judicial process for impeachments. John R. Labovitz, Presidential Impeachment 2 (1978).
• 139See Sunstein, supra note 7, at 51–52.
• 140See id. at 47–48.
• 141See id.
• 142See U.S. Const. art. II, § 4.
• 143See Staff of H. Comm. on the Judiciary, 93d Cong., Constitutional Grounds for Presidential Impeachment 5 (Comm. Print 1974).
• 144See generally J.S. Roskell, The Impeachment of Michael de la Pole, Earl of Suffolk in 1386 in the Context of the Reign of Richard II (1984).
• 145See James Fitzjames Stephen, A History of the Criminal Law of England 158–60 (London, MacMillan & Co. 1883) (stating that “with insignificant exceptions, the present law and practice as to parliamentary impeachments was established . . . in the latter part of the reign of Edward III[ ] and the reign of Richard II”).
• 146See S.L. Sutherland, Responsible Government and Ministerial Responsibility: Every Reform Is Its Own Problem, 24 Can. J. Pol. Sci. 91, 94 (1991).
• 147See Jack Simson Caird, Impeachment, UK Parliament (June 6, 2016), https://perma.cc/P6SF-X5HV.
• 148Staff of H. Comm. on the Judiciary, 93d Cong., supra note 143, at 16; see also Jack N. Rakove, Statement on the Background and History of Impeachment, 67 Geo. Wash. L. Rev. 682, 685 (1998).
• 149The Federalist No. 65, at 338 (Alexander Hamilton) (George W. Carey & James McClellan eds., 2001).
• 150Debates and Other Proceedings of the Convention of Virginia 353–54 (David Robertson ed., 2d ed. 1805). During proceedings regarding the potential impeachment of Richard Nixon, the House Judiciary Committee also stated that a finding of criminality was “neither necessary nor sufficient” to constitute an impeachable offense. See Staff of H. Comm. on the Judiciary, 93d Cong., supra note 143, at 24–25.
• 151See, e.g., Michael J. Gerhardt, The Federal Impeachment Process: A Constitutional and Historical Analysis 105 (3d ed. 2019) (“The major disagreement is not over whether impeachable offenses should be strictly limited to indictable crimes, but rather over the range of nonindictable offenses on which an impeachment may be based.”).
• 152Id. at 108; see also Tribe & Matz, supra note 7, at 42 (requiring “corruption, betrayal, or an abuse of power that subverts core tenets of the US governmental system”); Sunstein, supra note 7, at 56 (“distinctly political offenses” that are “abuses or violations of what the public is entitled to expect”).
• 153Griffin, supra note 32, at 419.
• 154Id.
• 155Id. at 427.
• 156See id.
• 157See Richard A. Serrano & Marc Lacey, Clinton Impeached: Split House Votes to Send Case Against President to Senate for Trial; Livingston to Leave Congress, L.A. Times (Dec. 20, 1998), https://perma.cc/4YPP-3EVR.
• 158See Trial Memorandum of President Donald J. Trump, supra note 33, at 1–2.
• 159See U.S. Const. art. II, § 4.
• 160See Jared P. Cole & Todd Garvey, Cong. Rsch. Serv., R44260, Impeachment and Removal 1 (2015).
• 161See Lawrence J. Trautman, Presidential Impeachment: A Contemporary Analysis, 44 U. Dayton L. Rev. 529, 536, 543 (2019).
• 162See Sunstein, supra note 7, at 103–06.
• 163At the same time, the constitutional text states that the chief justice must preside over the impeachment trial of the president of the United States. U.S. Const. art. I, § 3, cl. 6. The presence of the chief justice at the most important impeachments (those of the president) suggests perhaps some judicial role, but there is great uncertainty as to what the role entails. For a useful discussion of the ambiguous circumstances in which the chief justice’s role was created at the Philadelphia Convention (by the Committee of the Eleven) and the absence of floor debate, see Michael F. Williams, Rehnquist’s Renunciation? The Chief Justice’s Constitutional Duty to “Preside” over Impeachment Trials, 104 W. Va. L. Rev. 457, 468 (2002).
• 164See Nixon v. United States, 506 U.S. 224, 226 (1993).
• 165See id. at 230–31.
• 166See Matthew Soberg Shugart & John M. Carey, Presidents and Assemblies: Constitutional Design and Electoral Dynamics 19–20 (1992).
• 167See, e.g., Robert Elgie, Semi-Presidentialism: Sub-Types and Democratic Performance 10 (2011) (arguing as well that there are different subtypes of semi-presidentialism).
• 168See Comparative Constitutions Project, https://perma.cc/8QDF-2D6C. For details on the conceptualization and measurement of constitutions and constitutional systems, see Conceptualizing Constitutions, Comparative Constitutions Project, https://perma.cc/B3C9-ESQQ.
• 169Data on file with authors.
• 170See supra text accompanying notes 64–65.
• 171See Young Hun Kim, Impeachment and Presidential Politics in New Democracies, 21 Democratization 519, 527 (2014).
• 172This row includes some cases in which an impeachment vote was held, but the president was either removed beforehand or resigned, and so does not count as being formally removed by impeachment. For example, President Viktor Yanukovych was deposed in Ukraine’s Revolution of 2014, fleeing to Russia. Parliament voted to remove him from office for being unable to fulfill his duties but did not pass formal articles of impeachment. See Maria Popova, Was Yanukovych’s Removal Constitutional?, PONARS Eurasia (Mar. 20, 2014), https://perma.cc/SN8N-R2LX.
• 173Writing earlier, Professors David Samuels and Matthew Shugart report that out of 223 individuals elected in presidential democracies from 1946 to 2007, only 6 were ultimately impeached. See David J. Samuels & Matthew S. Shugart, Presidents, Parties and Prime Ministers: How the Separation of Powers Affects Party Organization and Behavior 111 (2010).
• 174Kim’s analysis also finds that impeachment attempts are more common when the president is involved in political scandal, and in systems with strong presidential powers. See Kim, supra note 171, at 521–23.
• 175See Samuels & Shugart, supra note 173, at 111.
• 176See id. at 108.
• 177Id. at 112.
• 178See supra text accompanying notes 42–48.
• 179See Clifford Krauss, Paraguay Glides from Desperation to Euphoria, N.Y. Times (Apr. 1, 1999), https://perma.cc/D479-YQV4.
• 180See Samuels & Shugart, supra note 173, at 117.
• 181In 2015 in Guatemala, the country’s attorney general made a motion for impeachment that was unanimously approved by the Supreme Court. The President was facing allegations of corruption. After the vote by the Supreme Court, the President submitted a resignation that was unanimously accepted by Congress. Congress also unanimously voted to strip him of his immunity from prosecution. This vote by Congress can thus be seen as akin to impeachment. See Guatemala’s President Otto Perez Molina Resigns, BBC News (Sept. 3, 2015), https://perma.cc/ZP84-ZFD7.
• 182See A Coup by Congress and the Street, The Economist (Apr. 25, 2005), https://perma.cc/7N7A-VCXD.
• 183Archigos dataset supplemented by authors. Note that the result in the final column holds if we extend the period to five years before and after impeachment, but data is then incomplete for the final two cases. Indonesia, where the impeachment of President Abdurrahman Wahid occurred just two years after the country became a democracy, shifted from a score of –7 to 8 by this broader temporal metric.
• 184President Carlos Andres Pérez’s removal may have been irregular—Congress declared the post “permanent[ly]” vacant after Pérez fled the country and the Supreme Court issued a preliminary ruling declaring the complaint to be well-founded, without waiting for the Court’s impeachment trial to finish. See Pérez-Liñán, supra note 95, at 21.
• 185President Alberto Fujimori had already fled the country in response to allegations of corruption and attempted to resign, but Congress insisted on completing the impeachment proceeding. See id. at 184–85.
• 186President Rolandas Paksas was impeached for violating the Lithuanian Constitution and his oath of office. His impeachment followed news that he had granted citizenship to a Russian businessman who was the main contributor to his campaign. After being found guilty by the Seimas (National Parliament), he was removed from office on the same day. See Terry D. Clark & Eglė Verseckaitė, PaksasGate: Lithuania Impeaches a President, 52 Probs. of Post-Communism 16, 20–21 (2005).
• 187The score for 2018 is used for this Article. No change is anticipated for 2019.
• 188See Edward Morgan-Jones & Petra Schleiter, Governmental Change in a President-Parliamentary Regime: The Case of Russia 1994–2003, 20 Post-Soviet Affs. 123, 156–57 (2004).
• 189Catherine M. Conaghan, Fujimori’s Peru: Deception in the Public Sphere 31–32 (2006).
• 190See, e.g., Robert Muggah, Opinion, Can Brazil’s Democracy be Saved?, N.Y. Times (Oct. 8, 2018), https://perma.cc/4Y2X-SJT8.
• 191Juliano Zaiden Benvindo, The Party Fragmentation Paradox in Brazil: A Shield Against Authoritarianism?, I-CONnect (Oct. 24, 2019), https://perma.cc/3GL3-QCW4; Tom Ginsburg & Mila Versteeg, The Bound Executive: Emergency Powers During the Pandemic 29–30 (July 26, 2020) (unpublished manuscript) (on file with authors).
• 192The case of Abdullá Bucaram, discussed below, is one example. See infra text accompanying notes 247–54. Viktor Yanukovych of Ukraine is another. See supra note 172. In addition, presidents have sometimes resigned under threat of impeachment, as occurred with Richard Nixon in the United States. For example, Raúl Cubas Grau resigned in Paraguay in 1999 after impeachment proceedings had been initiated, and during a deep political crisis. See Pérez-Liñán, supra note 95, at 32.
• 193See Adam Tooze, Crashed: How a Decade of Financial Crisis Changed the World 20 (2018). For a more general analysis of the relation of economic crisis and democracy, see Wolfgang Streeck, The Crises of Democratic Capitalism, 71 New Left Rev. 5 (2011).
• 194Data on file with authors.
• 195Adapted from Freedom House data 2016.
• 196Writing in 2001, Lilia Shevtsova noted that the “fundamental problems of democratic development . . . have still not been resolved.” Lilia Shevtsova, Ten Years After the Soviet Breakup: Russia’s Hybrid Regime, 12 J. Democracy 65, 65 (2001).
• 197Presidential and semi-presidential democracies are coded by the Democracy and Dictatorship Dataset, as supplemented by Professors Christian Bjørnskov and Martin Rode. See José Antonio Cheibub, Jennifer Gandhi & James Raymond Vreeland, Democracy and Dictatorship Revisited, 143 Pub. Choice 67, 68 (2010); Christian Bjørnskov & Martin Rode, Regime Types and Regime Change: A New Dataset 3–4 (Aug. 2018) (unp­ublished manuscript) (on file with authors).
• 198The head of state in the Marshall Islands is called a president, but can be removed on a vote of no confidence. Marsh. Is. Const. art. V, § 7.
• 199Phil. Const. 1987,, art. XI, § 2 (emphasis added).
• 200See Gam. Const. 1997, art. 67(1)(a); Uganda Const. 1995 art., 107(1)(a); Zim. Const. 2013, art. 97(1)(c); Ghana Const. 1992, art. 69(1)(a).
• 201See Griffin, supra note 32, at 419.
• 202See, e.g., Katie Zezima, Obama Action on Immigration Should Spark Impeachment Talk, GOP Lawmaker Says, Wash. Post (Aug. 3, 2014), https://www.washingtonpost.com/news/post-politics/wp/2014/08/03/obama-action-on-immigration-should-spark-impeachment-talk-gop-lawmaker-says/.
• 203Ghana Const. 1992, art. 69(1).
• 204Tanz. Const. 1977, art. 46A(2)(c).
• 205Uganda Const. 1995, art. 107(1)(b)(i).
• 206Hond. Const. 1982, art. 234.
• 207Calculated by summing Comparative Constitutions Project variables HOSPDM1, HOSPDM2, and HOSPDM3, corresponding to lower, upper, and both houses.
• 208Calculated by summing Comparative Constitutions Project variables HOSADM1, HOSADM2, and HOSADM3, corresponding to lower, upper, and both houses.
• 209Ir. Const. 1937, art. 12.
• 210See Fiji Const. 2013, art. 89.
• 211See Fiji Const. 2013, art. 89 (“In deciding whether to remove the President from office, Parliament must act in accordance with the advice given by the tribunal . . . .”).
• 212See generally Adrian Vermeule, Submajority Rules: Forcing Accountability upon Majorities, 13 J. Pol. Phil. 74 (2005) (defining and exploring the use of submajority rules).
• 213See Tanz. Const. 1977, art. 46A(3)(a).
• 214See Tanz. Const. 1977, art. 46A(3)(b), (4).
• 215See Tanz. Const. 1977, art. 46A(5), (9).
• 216See Nixon v. United States, 506 U.S. 224, 236 (1993).
• 217See supra Part I.A.
• 218This is also a fairly common design in Latin America, at least for some kinds of impeachments (such as those involving common crimes). See, e.g., El Sal. Const. 1983, arts. 236–237; Venez. Const. 1999, arts. 265–266.
• 219Colom. Const. 1991, art. 175.
• 220See Colom. Const. 1991, art. 175. As noted above, the Brazilian Constitution contains a similar provision, with roughly the same bifurcation of trial procedures between the Supreme Court and the Senate. See supra text accompanying note 70.
• 221Hond. Const. 1982, art. 313(2); see also generally Norma C. Gutiérrez, Honduras: Constitutional Law Issues, Law Libr. of Cong. (2009).
• 222See Noah Feldman, David Landau, Brian Sheppard & Leonidas Rosa-Suazo, Report to the Commission on Truth and Reconciliation of Honduras: Constitutional Issues 74–75 (Mar. 26, 2011), https://perma.cc/7YBT-B5DH (discussing this unusual feature).
• 223See, e.g., Tom Ginsburg, James Melton & Zachary Elkins, On the Evasion of Executive Term Limits, 52 Wm. & Mary L. Rev. 1807, 1810 (2011); David E. Landau, Rosalind Dixon & Yaniv Roznai, From an Unconstitutional Constitutional Amendment to an Unconstitutional Constitution? Lessons from Honduras, 8 Glob. Const. 40, 50 (2019).
• 224See Landau et al., supra note 223, at 50.
• 225See J. Mark Ruhl, Trouble in Central America: Honduras Unravels, 21 J. Democracy 93, 98–100 (2010) (discussing Zelaya’s relationship with the Liberal Party).
• 226See Scott Mainwaring & Aníbal Pérez-Liñán, Cross-Currents in Latin America, 26 J. Democracy 114, 118 (2015).
• 227See Feldman et al., supra note 222, at 5–6, 46 (explaining the content of the warrants and the difficulty with verifying when they were issued).
• 228Organization of American States [OAS], CP/Res. 953 (1700/09), Current Situation in Honduras, ¶ 3 (June 28, 2009).
• 229See Ruhl, supra note 225, at 102.
• 230See Matthew Pomy, Honduran Lawmakers Approve Impeachment Process, Jurist (Jan. 24, 2013), https://perma.cc/NV2M-9CC7.
• 231See Gam. Const. 1997, art. 63.
• 232See Austria Const. 1920, art. 60(6).
• 233See Colom. Const. 1991, art. 178(4).
• 234See, e.g., Susan Spronk, Pink Tide? Neoliberalism and Its Alternatives in Latin America, 33 Can. J. Latin Am. & Caribbean Stud. 173, 182 (2008).
• 235See Bol. Const. 2009, art. 240; Ecuador Const. 2008, art. 105.
• 236See Bol. Const. 2009, art. 240; Ecuador Const. 2008, art. 105.
• 237See Venez. Const. 1999, art. 72.
• 238See Venez. Const. 1999, art. 72.
• 239See Venez. Const. 1999, art. 72.
• 240Amy Zacks, Recalling Governors: An Overview, Rutgers Ctr. on the Am. Governor (Apr. 25, 2012), https://perma.cc/D6XL-QGK9.
• 241See id.
• 242Jane A. Hudiburg & Christopher M. Davis, Cong. Rsch. Serv., R45087, Resolutions to Censure the President: Procedure and History 5–8 (2018) (describing history).
• 243U.S. Const. amend. XXV. See generally Joel K. Goldstein, Talking Trump and the Twenty-Fifth Amendment: Correcting the Record on Section 4, 21 U. Pa. J. Const. L. 73 (2018) (canvassing debates on the meaning of the Twenty-Fifth Amendment).
• 244U.S. Const. amend. XXV.
• 245See Roy E. Brownell II, What to Do If Simultaneous Presidential and Vice Presidential Inability Struck Today, 86 Fordham L. Rev. 1027, 1033 (2017); Sunstein, supra note 7, at 148.
• 246See, e.g., Richard Cohen, How to Remove Trump from Office, Wash. Post (Jan. 9, 2017), https://perma.cc/MB5T-RZGZ; Lawrence M. Friedman & David M. Siegel, The Most Important Qualification for a Post in President Trump’s Cabinet, New Eng. L. Rev. F. (Feb. 15, 2017), https://perma.cc/P6YQ-HGHW (discussing the importance of Cabinet members to be willing to fulfill responsibilities under the Twenty-Fifth Amendment and the failure of senators to question Cabinet nominees on the subject).
• 247See Aníbal Pérez-Liñán, Impeachment or Backsliding: Threats to Democracy in the Twenty-First Century, 33 Revista Brasileira de Ciências Sociais 1, 2 (2018) (noting that Bucaram’s party had only 23% of legislative seats).
• 248See Susan C. Stokes, Mandates and Democracy: Neoliberalism by Surprise in Latin America 1–2 (2001).
• 249Leon Zamosc, The Indian Movement and Political Democracy in Ecuador, 49 Latin Am. Pol. & Soc’y 1, 11–12 (2007).
• 250Ecuador Const. 2008, art. 145.
• 251Ecuador Const. 2008, art. 145.
• 252See Gabriel Escobar, Ecuadorian Lawmaker Renamed President, Wash. Post (Feb. 12, 1997), https://perma.cc/CUC7-JSQJ.
• 253See Carlos de la Torre, Populist Seduction in Latin America 92 (2d ed. 2010).
• 254See Laurence Whitehead, High Anxiety in the Andes: Bolivia and the Viability of Democracy, 12 J. Democracy 6, 7 (2001).
• 255See Tanz. Const. 1977, art. 46A(2)(c) (“[N]o such motion shall be moved within twenty months from the time when a similar motion was previously moved and rejected by the National Assembly.”).
• 256See, e.g., Angl. Const. 2010, art. 127(2).
• 257The European Court of Human Rights struck down this ban in 2011, holding that it was disproportionate. See Paksas v. Lithuania, 2011-I Eur. Ct. H.R. 1.
• 258See Braz. Const. 1988, art. 52, sole paragraph.
• 259See supra Part I.B.
• 260Colom. Const. 1991, art. 175:

  If the charge refers to crimes committed in the exercise of his/her functions or that he/she becomes unworthy to serve because of a misdemeanor, the Senate may only impose the sanction of discharge from office or the temporary or absolute suspension of political rights. But the accused shall be brought to trial bef­ore the Supreme Court of Justice if the evidence demonstrates that the individual to be responsible for an infraction deserves other penalties.

• 261See Rattinger, supra note 74, at 149.
• 262For all constitutional systems, we count forty-eight in which another official serves out the remainder of the term, and eighty-three in which there are new elections, with seven that we are unable to determine.
• 263Data on file with authors.
• 264For the canonical formulation of this point, see William J. Stuntz, The Uneasy Relationship Between Criminal Procedure and Criminal Justice, 107 Yale L.J. 1, 55–59 (1997). See also Donald A. Dripps, The Substance-Procedure Relationship in Criminal Law, in Philosophical Foundations of Criminal Law 409, 423–25 (R. A. Duff & Stuart P. Green eds., 2011).
• 265George Tsebelis, Veto Players: How Political Institutions Work 17 (2002).
• 266For our purposes, votes by a joint session of two houses are considered as a single actor. See, e.g., Burundi Const. 2018, art. 117.
• 267See generally Tom Ginsburg, Constitutional Knowledge, 2 Know 15 (2018); Aziz Z. Huq, Hippocratic Constitutional Design, in Assessing Constitutional Performance 39 (Tom Ginsburg & Aziz Z. Huq eds., 2016).
• 268See Huq, supra note 267, at 41–42.
• 269See supra Part I.E.
• 270See Pérez-Liñán, supra note 95, at 36 (noting the importance of a president’s “le­gislative shield” in determining whether they would survive impeachment proceedings).
• 271See supra Part I.A.
• 272See supra Part I.B.
• 273See supra Part I.D.
• 274See supra Part I.A.
• 275See supra Part I.D.
• 276See supra Part I.B.
• 277See supra Part I.C.
• 278See generally, e.g., Samuels & Shugart, supra note 173; Matthew Soberg Shugart & Stephan Haggard, Institutions and Public Policy in Presidential Systems, in Presidents, Parliaments, and Policy 64 (Stephan Haggard & Mathew D. McCubbins eds., 2001).
• 279John Gerring, Strom C. Thacker & Carola Moreno, Centripetal Democratic Governance: A Theory and Global Inquiry, 99 Am. Pol. Sci. Rev. 567, 571 (2005).
• 280See Alfred Stepan & Cindy Skach, Constitutional Frameworks and Democratic Consolidation: Parliamentarianism Versus Presidentialism, 46 World Pol. 1, 3 (1993).
• 281See id. at 3–4.
• 282See, generally e.g., Cheibub et al., supra note 26.
• 283For an elegant demonstration of the difficulty of drawing simple comparisons, see José Antonio Cheibub, Presidentialism, Parliamentarism, and Democracy 140 (2007) (suggesting that military legacies, rather than the choice between presidentialism and parliamentarianism, lead to democratic breakdowns).
• 284See, e.g., Linz, supra note 31, at 51–52.
• 285See Arturo Valenzuela, Latin America: Presidentialism in Crisis, 4 J. Democracy 3, 6 (1993). For an insightful challenge to this view, see Josh Chafetz, The Phenomenology of Gridlock, 88 Notre Dame L. Rev. 2065, 2073 (2013).
• 286See Arturo Valenzuela, The Breakdown of Democratic Regimes: Chile 15 (1978).
• 287The 2011 Fixed Parliament Act in the United Kingdom may have inadvertently created friction on this dynamic by making it harder for resets to occur. See Petra Schleiter & Sukriti Issar, Constitutional Rules and Patterns of Government Termination: The Case of the UK Fixed-Term Parliaments Act, 51 Gov’t & Opposition 605, 608–09 (2016).
• 288See Huber, supra note 27, at 270.
• 289See Miguel Carreras, Outsiders and Executive-Legislative Conflict in Latin America, 56 Latin Am. Pol. & Soc’y 70, 83–84 (2014) (finding that risks of interbranch conflict and attempted dissolution of Congress increase significantly when the president is an outsider).
• 290See Steven Levitsky & Daniel Ziblatt, How Democracies Die 13–15 (2018) (explaining the risks to democracy posed by the “charismatic outsider”); David Landau, Populist Constitution, 85 U. Chi. L. Rev. 521, 524–25 (2018).
• 291See generally, e.g., Meyer, supra note 86; see also Fabiano Santos & Fernando Guarnieri, From Protest to Parliamentary Coup: An Overview of Brazil’s Recent History, 25 J. Latin Am. Cultural Stud. 485, 488 (2016).
• 292See, e.g., Alayna Treene & Margaret Talev, The Normalization of Impeachment, Axios (Dec. 13, 2019), https://perma.cc/JGL3-PHDE (collecting quotes by congressional leaders); David Marcus, Impeachment Has Become the Dangerous New Normal, The Federalist (Jan. 21, 2020), https://perma.cc/2XAS-R6A6 (arguing that the U.S. is on a “wildly dangerous trajectory” because of more frequent use of impeachment).
• 293As an example, consider Ecuador, which had seven presidents between 1997 and 2007, and in which no president completed his or her term in that time period, despite none of them being removed by impeachment. See Pérez-Liñán, supra note 95, at 29 (calling irregular removal in Ecuador “a chronic disease”).
• 294See supra text accompanying notes 221–30.
• 295Many U.S. scholars are in fact critical of the modern practice, although they maintain a focus on finding criminal or noncriminal bad acts. See, e.g., Sunstein, supra note 7, at 118 (rejecting requirement of common crime but requiring “egregious abuse of official power”); Tribe & Matz, supra note 7, at 45; Black, supra note 7, at 38–40 (arguing that impeachment is not limited to common crimes, but criminality “helps”); Gerhardt, supra note 17, at 105. Many also emphasize that the standard cannot be so broad as to reach acts such as “maladministration,” given its explicit rejection at the Convention. See, e.g., Sunstein, supra note 7, at 76; Black, supra note 7, at 27.
• 296See generally Griffin, supra note 32.
• 297See Huq, Legal or Political Checks, supra note 9, at 1522–23.
• 298See generally Raymond Carver, What We Talk About When We Talk About Love, in What We Talk About When We Talk About Love 170 (1981).
• 299See Para. Const. 1992, art. 225.
• 300See Braz. Const. 1988, art. 85.
• 301See supra Part I.E and note 294.
• 302See Gerhardt, supra note 17, at 108.
• 303See Sunstein, supra note 7, at 56 (quoting The Federalist No. 65 (Alexander Hamilton)).
• 304See supra Table 4.
• 305We say “normally” because federal judges may benefit from an additional textual protection: they cannot be fired unless they fail to show “good Behaviour.” U.S. Const. art. III, § 1, cl. 2.
• 306U.S. Const. art. II, § 4.
• 307See Braz. Const. 1988, arts. 51, 52, 85, 86, 93, 96, 102, 105, 107, 108 (setting out different procedures and standards for different actors, including the president, vice president, cabinet members, and different types of judges).
• 308Professor Cass Sunstein argues that although the constitutional standard for impeachment for judges and presidents “is exactly the same,” judicial impeachment should have a “mildly different and somewhat lower bar” because of pragmatic factors, especially the “uniquely destabilizing” consequences of presidential removal. See Sunstein, supra note 7, at 115. This argument may be reconcilable with ours: the predicate grounds of judicial impeachment should be narrower than presidential impeachment, but deliberations on whether a president eligible for impeachment should be removed ought to take greater account of pragmatic considerations.
• 309See, e.g., Braz. Const. 1988, art. 86; Colom. Const. 1991, art. 175.
• 310See supra Parts I.A, I.D.
• 311See supra text accompanying notes 126–29.
• 312See, e.g., Matt Ford, Can the Senate Save Robert Mueller?, The New Republic (Apr. 12, 2018), https://perma.cc/5MCT-MSJ7 (discussing legislative proposals to provide more protection to Mueller’s office and tenure).
• 313See, e.g., Daphna Renan, Presidential Norms and Article II, 131 Harv. L. Rev. 2187, 2209 (2018) (discussing how the firing of the special prosecutor investigating Nixon, as well as several attorneys general, sparked normative changes in the executive branch).
• 314See, e.g., Michael D. Shear & Lola Fadulu, Trump Says Mueller Was ‘Horrible’ and Republicans ‘Had a Good Day’, N.Y. Times (July 24, 2019), https://perma.cc/SG9A-248R.
• 315The Ethics in Government Act under which Starr was appointed was examined and upheld in Morrison v. Olson, 487 U.S. 654, 696–97 (1988). The law provided for an independent counsel who was appointed, under certain conditions, by a panel of judges, and who could be removed by the Attorney General only for good cause. See id. at 660–64. The independent counsel was also required to report to the House of Representatives any “substantial and credible information . . . that may constitute grounds for impeachment.” Id. at 664–65. The law was permitted to lapse in 1999, in the wake of the failed Clinton impeachment.
• 316See, e.g., Tribe & Matz, supra note 7, at 21; Sunstein, supra note 7, at 100–01.
• 317See supra Part I.E.
• 318Scholarly treatments of the United States generally also find this noninvolvement to be a good thing. See, e.g., Black, supra note 7, at 55.
• 319See supra note 114.
• 320See supra Part I.A.
• 321See supra Part I.B.
• 322See supra text accompanying notes 221–30.
• 323See, e.g., Aaron Ansell, Impeaching Dilma Rousseff: The Double Life of Corruption Allegations on Brazil’s Political Right, 59 Culture, Theory & Critique 312, 318 (2018).
• 324See generally Huq, Tactical Separation of Powers, supra note 127.
• 325See supra text accompanying notes 108–11.
• 326506 U.S. 224 (1993).
• 327See S. Kor. Const. art. 68(2) (requiring a new election within sixty days of a vacancy in the presidency, including those caused by disqualification via judicial decision).
• 328See Linz, supra note 31, at 52–53.
• 329See Brazil’s Dilma Rousseff Accuses Deputy of Coup Plot, BBC News (Apr. 13, 2016), https://perma.cc/8TNA-3JZC; Paraguay’s Impeachment: Lugo Out in the Cold, supra note 104.
• 330See, e.g., van Dijk, supra note 62, at 203.
• 331See The Arrest of Michel Temer, The Economist (Mar. 21, 2019), https://perma.cc/C4FH-953D.
• 332See Attorney-General Opens Investigation into Former President, The Economist Intel. Unit (Oct. 26, 2015), https://perma.cc/CBA5-PJVQ.
• 333We focus here on impeachment, but the logic of our argument also applies to other forms of political control of chief executives such as recall. In Venezuela, for example, the consequence of a successful referendum to recall the president is peculiarly sensitive to time. A president can only be recalled during the second half of his or her six-year term. See Venez. Const. 1999, art. 72. However, if recall happens during the last two years of the term, then the vice president takes over for the remainder of the original term, instead of a new election being held within thirty days. See Venez. Const. 1999, art. 233. The combination of these provisions provides only a narrow window of one year (the fourth year of a presidential term) in which recall can be planned and carried out in a way that triggers a reset.
• 334A related question is whether there should be a de minimis exception to the rule requiring new elections in cases where the former presidents had very little time left in their term. If such an exception exists, we would suggest it should likely be fairly short (say, no more than six months or a year) in order to allow impeachment to play the reset function that we lay out here.
• 335See, e.g., Ecuador Const. 2008, art. 148.
• 336Alternatively, one could include a rule that failed attempts at impeachment mean that no new motion can be brought for a set period, as described above for Tanzania. See Tanz. Const. 1977, art. 46(A)(2)(c).
• 337See supra Part I.B.