SciELO - Scientific Electronic Library Online

Home Pagealphabetic serial listing  

Services on Demand




Related links


Revista Brasileira de Ciências Sociais

Print version ISSN 0102-6909On-line version ISSN 1806-9053

Rev. bras. Ci. Soc. vol.33 no.98 São Paulo  2018  Epub Dec 20, 2018 


IMPEACHMENT OR BACKSLIDING? Threats to democracy in the twenty-first century *



Aníbal Pérez-Liñán1 

1 University of Notre Dame (ND), Indiana , USA . E-mail:


The age of military coups has ended, but democracy still confronts major challenges in the early twenty-first century. What are the new threats for democratic survival? Is impeachment a functional equivalent to old-fashioned military coups? Using comparative data for Latin America, this paper shows that the institution of impeachment has been “stretched” for political purposes and that the social conditions that triggered military coups in the pasttrigger impeachments in the contemporary era. However, the paper argues that impeachments should not be confused with coups and presents a surprising result: even though legislators often manipulate the impeachment process to undermine elected presidents, the most common threat to democratic survival does not originate in legislatures, but in the executive branch. Concentration of power by the executive has undermined democracy in Latin Americaand elsewhere.

Key words: Impeachment; coup; Latin America; concentration of power


A era dos golpes militares acabou, mas a democracia ainda enfrenta grandes desafios neste início do século XXI. Quais são as novas ameaças à sobrevivência da democracia? Seria o Impeachment um equivalente funcional dos antiquados golpes militares? Utilizando dados comparativos relativos à América Latina, este artigo expõe a “ampliação” da instituição do impeachment por motivos políticos, bem como explicita que as condições sociais que causaram golpes militares no passado causam, na contemporaneidade, impeachments. No entanto, o artigo argumenta que impeachments não deveriam ser confundidos com golpes e apresenta um resultado surpreendente: mesmo que legisladores frequentemente manipulem o processo de impeachment para debilitar presidentes eleitos, a ameaça mais comum à sobrevivência democrática não se origina do poder legislativo, mas do executivo. A concentração de poder pelo executivo debilitou a democracia na América Latina e em outros lugares.

Palavras-Chave: Impeachment; Golpe; América Latina; Concentração de poder


La reconnaissance de la possibilité de concilier les modes de vie des peuples et des communautés autochtones avec la préservation de la biodiversité a déclenché la création de dispositifs légaux qui permettent à ces individus de demeurer sur leurs terres. Cependant, les aires protégées continuent d’être des espaces disputés, dans lesquels sont projetés les intérêts sociaux pour le territoire, comme un moyen de production matérielle et immatérielle de la vie, les intérêts qui visent à destiner ces espaces exclusivement à la protection environnementale et les intérêts des marchés, qui imprègnent plusieurs des pratiques encouragées dans les Unités de Conservation. Dans cet article, je présente les caractéristiques de ces conflits au sein de la Forêt Nationale du Tapajós (Flona Tapajós), où le “droit de rester” impose des conditionnalités à la population, c’est-à-dire, des restrictions d’utilisation et d’introduction de nouvelles pratiques qui ont modifié les relations communautaires et les dynamiques socio-économiques de ce territoire.

Key words: Impeachment; Coup d’État; Amérique latine; Concentration du pouvoir


Democracy is at a critical juncture worldwide. In advanced industrial democracies, traditional party elites have lost much of their public support and legitimacy, creating an opportunity for the emergence of a new radical right. In Latin America, the process of party renewal inaugurated two decades ago has apparently exhausted its course. Some of the leftist parties that led this transformation have left power, and others have retained power at the expense of their democratic principles. Venezuela is perhaps the most dramatic example of this crisis: once the flagship of a hopeful process of transformation, the Bolivarian experiment has degenerated into authoritarian repression, crony capitalism, mass poverty, and a refugee crisis.

The institution of presidential impeachment has played a visible role in the exhaustion of the most recent political cycle in Latin America. Although presidential constitutions originally included this procedure to address the president’s “high crime and misdemeanors”, the impeachment of presidents Fernando Lugo in Paraguay (2012) and Dilma Rousseff in Brazil (2016) proved that ideological struggles often prevail above legal considerations during the impeachment process.

What is the role of presidential impeachment in contemporary Latin America? Is this institution an instrument of accountability or a functional equivalent of traditionalmilitary coups? Can the destabilization of elected presidents by congressional majorities become the main source of democratic instability in the future? No single case or experience can provide an answer to these questions; a long-term comparative perspective is important to place our answers in context.

This essay will advance three arguments. First, the impeachment procedure has been consistently “stretched” for political reasons. Impeachments were also central to the exhaustion of the previous political cycle dominated by neoliberal presidents, and historical evidence suggests that those impeachments were similarly driven by political goals. Thus, claiming that manipulated impeachments are a new form of coup d’ État would forces us to reassess not only the concept of coup, but also the past three decades of Latin American political history. Second, contemporary impeachments and past military coups are explained by similar factors. The social and political conditions that fostered military coups during the Cold War are also likely to foster presidential impeachments in the present. Taken together, these two arguments help us understand in which ways impeachments represent a functional equivalent of coups, without stretching the concept of coup dangerously. Avoiding such a conceptual stretch is relevant because we may otherwise miss a third, important fact. The last argument of this paper is that Latin American history shows that most threats to democracy originate in the executive, not in congress. Accordingly, the main threat to democracy in the twenty-first century will not result from weak presidents undermined by the legislature, but from hegemonic presidents who undermine the separation of powers. Although commonly seen as a source of institutional dysfunction, the Brazilian system may have the advantage of preventing such an outcome.

Stretching the impeachment procedure

Impeachments were almost unknown in Latin America until the 1990s. During the Cold War era, military coups were the typical mechanism to remove “undesirable” presidents from office. Between 1945 and 1990, only one episode (the removal of Panamanian presidentJosé Ramón Guizado in 1955) can be truly classified as an impeachment process.

It was only in the 1990s that congressional action against the executive became a common fixture of Latin American politics. Between the Collor impeachment in 1992 and the Rousseff impeachment in 2016, there was hardly any year when a president in the region was not challenged and removed from office. During this period, eight presidents were removed from office through an impeachment or by a declaration of incapacity issued by congress: Fernando Collor (1992) and Dilma Rousseff (2016) in Brazil, Carlos Andrés Pérez (1993) in Venezuela, Abdalá Bucaram (1997) and Lucio Gutiérrez (2005) in Ecuador, Raúl Cubas (1999) and Fernando Lugo (2012) in Paraguay, and Otto Pérez Molina (2015) in Guatemala. 1 Two Peruvian presidents – Alberto Fujimori (2000)and Pedro Pablo Kuczynski(2018) – also resigned anticipating an impeachment process.

These episodes share three features. First, the government lost public support among a majority of voters. Almost every episode involved massive mobilizations in the streets calling for the resignation of the president. Such protests were usually triggered by a combination of two factors: major scandals involving corruption or abuse of power – ultimately a key element in the narrative to justify an impeachment – and a looming economic crisis, driven by economic recession and high unemployment or, more commonly in the 1990s, by the government’s attempt to control skyrocketing inflation through unpopular neoliberal policies.

Second, presidents lost their “shield” in congress. Some of them never had strong congressional support, and they were unable to articulate a solid coalition. Fernando Collor’s party only controlled about 8% of the seats in the Chamber of Deputies; Abdalá Bucaram’s controlled about 23% of the seats in the Ecuadorian assembly. Other presidents had stronger parties, but they lost their support in the midst of the crisis. Carlos Andrés Pérez was a historic leader of the largest Venezuelan party at the time, Acción Democrática, with about 48% of the seats in the lower house. Raúl Cubas Grau’s Partido Colorado commanded 56% of seats in Paraguay’s Chamber of Deputies. However, those parties divided into factions, and the largest factions abandoned the president. Sometimes, as in the case of Dilma Rousseff, presidents relied on coalitions that broke down. In the end, all presidents lost their shield in congress, and legislators moved forward with an impeachment or declaration of incapacity.

The third element in common is that legislatures embraced creative interpretations of the constitution to depose the president at a particular moment. Congress members manipulated not only the legal grounds for impeachment, but sometimes the legal procedure as well. In Venezuela, for example, Carlos Andrés Pérez was accused of embezzling 16 million dollars from the Interior Ministry. However, the cause for his ousting was different: Pérez had inherited an economic crisis, adopted highly unpopular neoliberal policies, and massive mobilizations rattled his administration. Politicians – even those in his own party – explored alternative options to remove the president from office, hoping to contain the crisis. They finally found 16 million unaccounted for in the Interior Ministry and used this case as a justification for impeachment. Pérez would later argue that he never disclosed the use of these funds because they had been allocated to support the security detail of Nicaraguan president Violeta Chamorro during the 1990 transition.

Paraguay’s experience illustrates how institutional stretching can affect the legal justification for impeachment as well as its procedure.Raúl Cubas Grau was charged with contempt for the Supreme Court, but the reason behind his removal was the widespread – and perhaps misplaced – suspicion that he had been involved in a conspiracy to kill his vice-president, shot in the streets of Asunción in March 1999. The Chamber of Deputies modified its internal procedures to redefine a two-third majority as 66.5% of the vote, instead of 67%, and the constitutional accusation ultimately passed in the lower house with a difference of just one vote. According to people I interviewed, one of the deputies willing to support the president was locked in a restroom to prevent him from voting. In a more recent episode in 2012, presidentFernando Lugo was impeached in just 48 hours, without any real chance to defend himself.

In Ecuador, massive demonstrations called for the resignation of Abdalá Bucaram in 1997. Opposition leaders realized that this was an excellent opportunity to remove the president from office, but they could not muster the super majority required to activate an impeachment process. In the end, they found a loophole in the constitution that allowed a simple majority to declare the president mentally incapacitated. Congress thus declared president Bucaram mentally insane out of expediency.

These examples suggest that, from a comparative perspective, the two impeachment processes in Brazil have been quite respectful of procedure. To wit, the impeachment of Fernando Lugo took two days, while the process against Dilma Rousseff took nine months. Comparison is always relevant to calibrate our perspective of events, even though the Brazilian experience still raises important and legitimate concerns. The grounds for impeachment were highly contested in the Rousseff case, as the legal boundary of crimes de responsabilidade [crimes of responsibility] was subject to debate and reinterpretation. Even in the Collor case, some legal decisions were controversial. Article 52 of the Brazilian Constitution states that the consequences of presidential impeachment are “limited to the loss of office, with disqualification to hold any public office for a period of eight years”. President Collor resigned right before the Senate’s decision to avoid disqualification, but he was disqualified for eight years anyway, while president Rousseff, although she was impeached, was not disqualified for office (Collor, 2016, pp. 235-252). It is equally clear that Congress did not use consistent standards of proof to assess the offenses of Fernando Collor, Dilma Rousseff, and Michel Temer when deciding whether to initiate an impeachment process.

Although most impeachments have been questionable in terms of motivations or procedure, the legal foundation of the processes that terminated other presidencies during this era has been even weaker. Popular protests forced the resignation of Alberto Fujimori in Peru (2000), Fernando de la Rúa in Argentina (2001), and Gonzalo Sánchez de Lozada in Bolivia (2003). Legislators merely formalized the ousting of Fujimori (who had already submitted his resignation from Japan), appointed a successor to replace de la Rúa, and virtually played no role in Bolivia (Mustapic, 2005). Hochstetler (2006) thus concluded that social movements are the new “moderating power” in Latin America.

The comparative exercise underscores that the impeachment process has been “stretched” for at least 25 years. Recent studies document that the Latin American right has deployed the impeachment process against presidents on the left (Cannon, 2016). This is certainly true in the cases of Lugo and Rousseff, yet only two of the eight cases discussed (twelve, counting the resignations) reflect this pattern. Most crises affected presidents of the neoliberal period, during an era when adverse economic conditions made elected executives extremely fragile. Even in recent years, presidential instability has undermined presidents at the right (Pérez Molina), left (Rousseff), and center (Kuczynski) of the political spectrum.

Impeachments and coups: common causes

Impeachments, declarations of incapacity, and anticipated resignations are purportedly “legal” procedures, which contrast with the overtly unconstitutional military coups employed to overthrow elected presidents in the past. However, the military coups of the twentieth century and the legal procedures of the twenty-first share some important causes. To establish this fact, we need to compare recent impeachments with a broader set of cases, taking into account earlier periods of democracy (Alvarez and Marsteintredet, 2010).

A study published in the journal Democratization analyzed the ousting of democratically elected presidents in 19 Latin American countries between 1945 and 2010 (Pérez-Liñán and Polga-Hecimovich, 2017). The units of analysis are country-years (N = 729) and the sample includes 36 crises leading to the ousting of an elected president. The study employs a duration model with sample selection, which predicts the removal of presidents in two stages. The first stage models the risk that a president will be removed from office, irrespective of the mechanism employed for the ouster in any given year. The sample for this stage includes all country-years (N = 729), and the dependent variable captures whether the president was ousted in any country-year. The second stage predicts whether presidents are removed by a military coup or a legal procedure. The sub-sample for this stage includes only presidential crises (N = 36), and the dependent variable captures whether the president was ousted legally (via impeachment, declaration of incapacity or anticipated resignation).

Although this brief section prevents a detailed discussion of the statistical technique, it is worth summarizing the study’s main results. The first stage of the model indicates that common factors help explain both past coups and contemporary impeachments. We find that three variables underpin both forms of political instability. The first one is (weak) economic growth: economic recessions destabilize presidential administrations in all contexts. The second variable is mass demonstrations against the government: protests weaken elected executives and strengthen their opponents. The third factor is the radicalization of political elites. By radicalization, I refer to a process by which political elites embrace extreme policy positions about which they become intransigent (Mainwaring and Pérez-Liñán, 2013). Their unwillingness to compromise becomes very problematic in a democratic context, as radical governments prefer to ignore institutional constraints and radical oppositions are willing to use any strategy (legal or not) to undermine the government.

Below, Figure 1 simulates the risk of presidential removal using legal procedures – impeachments, declarations of incapacity, and resignations – or military coups, under different historical conditions. The vertical axis in each panel represents the predicted probability of presidential exit via legal removal, in the first row, or via coup, in the second row. The horizontal axis reflects the behavior of three variables with significant statistical effects in the first stage of the model: economic conditions, popular mobilization, and elite norms. The first column (panels 1.1.1 and 1.1.2) simulates the effect of economic growth; the second column (panels 1.2.1 and 1.2.2), the impact of mass demonstrations; and the last column (1.3.1 and 1.3.2), the consequences of radicalization. 2

Source: Estimates based on Pérez-Liñán and Polga-Hecimovich (2017).

Figure 1 Estimated Risk of Legal Removals and Military Coups  

Panels 1.1.1 and 1.1.2 show that an economic recession increases the risk of legal removals in the present, but it also increased the risk of military coups in the past. In periods of fast economic growth, the risk of impeachment or coup approximates zero. But when the economy is bad, people are more willing to depose the president. An acute recession (with negative economic growth of -8%) maps into a risk of impeachment close to 4% and a risk of coup close to 6% per year.

Similarly, mass demonstrations against the government foster presidential instability in the form of impeachments (panel 1.2.1) or military intervention (panel 1.2.2). The variable in the middle column reflects the number of demonstrations against the government reported by The New York Times yearly, coded by Arthur Banks’ Cross-National Time Series (Banks and Wilson, 2012). These protests are large enough to attract coverage of the international press and thus likely to rattle the government. The risk of instability rises consistently with popular mobilization. Without protests, the average risk of impeachment is about 1% per year, and the risk of coup is close to 2%. With five major protests in a year, the risk of impeachment escalates to approximately 7% and the risk of coups to 10%.

Figures in the third column reflect the impact of radicalization. Radical actors have extreme preferences and they remain intransigent in defense of their policy goals. The variable measuring radicalism originates in the study by Mainwaring and Pérez-Liñán (2013). This study identified almost 1500 powerful political actors in 20 Latin American countries between 1944 and 2010. For every country-year, the project typically identified between three and seven actors with major influence in the political regime – in most cases, the list included the president and major political parties, sometimes the military, trade unions, or other organizations. Actors were coded as “radical” when they expressed uncompromising policy goals, showed willingness to subvert the law to achieve these goals, or undertook violent protests against the government to impose policy preferences. All actors were given a score of 1 (radical), 0.5 (“somewhat” radical), or 0 (moderate). The variable presented in Figure 1 measures the percentage of radical actors in every country-year.

The simulations show that the radicalization of political elites consistently expands the risk of impeachments (panel 1.3.1) as well as military coups (panel 1.3.2). When democratically elected presidents pursue policies that radical actors deem unacceptable, a “disloyal” opposition looks for the most effective way to terminate the administration (Linz, 1978).

The statistical analysis therefore demonstrates that traditional military coups and contemporary impeachments share important causes – economic decline, related mass protests, and the radicalization of powerful actors. But if this is the case, it is worth asking why the three destabilizing factors induced military intervention before 1990 and legal ousters afterwards.

The second stage of the model, focusing exclusively on cases of presidential removal, identifies some important factors that help explain why military coups prevailed in the past while impeachments dominate the present. A greater commitment to democracy among elites tends to increase the risk of impeachments instead of military interventions, and a regional context in which coups are rare discourages further military coups in the neighborhood. In the 1970s, influential elites subordinated democratic principles to revolutionary dreams or anti-communist crusades, and the region was populated by military dictatorships. Thus, it was easy for military officers to advance their conspiracies in the context of the Cold War. Today, calls for military intervention are very difficult to justify (Pérez-Liñán and Polga-Hecimovich, 2017).

The coup debate: from political stretching to conceptual stretching

The previous sections showed that the institution of impeachment has been “stretched” by politicians willing to depose elected presidents when the executive is weak – undermined by economic crises, popular protests, and radicalized politics. They also documented that some of the conditions that encouraged military coups in the past are producing impeachments in the present. Thus, it is natural to wonder if impeachments represent an updated version of old-fashioned military coups for the twenty-first century. Indeed, several observers have made this claim, arguing that impeachments are equivalent to coups, or neo-coups, in the contemporary era.

The debate about this issue led to heated political arguments in the aftermath of the impeachments against Fernando Lugo and Dilma Rousseff, but it also generated rich academic exchanges. In an early essay, Santos and Guarnieri (2016) characterized Rousseff’s case as a “parliamentary coup”. The authors noted – in line with the results presented in Figure 1 – that the process leading to Rousseff ’s downfall began with the mass protests of June 2013, crystalized with the adverse economic conditions of 2014 and 2015, and concluded with a “farce” by political elites in 2016. In a response to this paper, Avelar (2017b) challenged the characterization of the episode as a parliamentary coup, observing “a gap between the rhetoric of the coup and Brazilian political science ’s attempts to justify the country’s coalitional presidentialism.” He also warned against reading the 2013 protests as a mere prelude to the impeachment and defended a broader interpretation of the protests’emancipatory legacies (Avelar, 2017a). Nunes and Melo (2017) explicitly declined to call the impeachment process a coup, but noted that the debate remains open because the PSDB refused to accept its defeat in the 2014 presidential election and because the interpretation of “pedaladas fiscais” [fiscal dribbles] as a constitutional crime of responsibility was not sufficiently convincing.

Malamud and Marsteintredet (2017) identify a proliferation of coup types in the political science literature, which increasingly refers to soft, parliamentary, constitutional, neo-liberal, market, electoral, slow-motion, civil-society, and judicial coups. Scholars have used these labels to describe presidential crises in Brazil and Paraguay, but also in Argentina, Ecuador, Peru, and Venezuela. These types add adjectives to the root concept (coup) in order to qualify, or diminish, one of its three defining attributes. In its classic definition, a coup is: (1) perpetrated by a state agent, which (2) targets the chief executive (3) using illegal tactics. The use of diminished subtypes in political science has been widely documented by the literature on concept formation (Collier and Levitsky, 1997; Collier and Mahon, 1993; Goertz, 2006). However, because these labels uncritically expand the extension of the root concept without reducing the number of defining attributes, they implicitly engage in what Sartori (1970) called “conceptual stretching”.

Stretching the concept of coup presents two analytical problems and three unintended political consequences. The analysis presented in the previous section, showing that military coups and impeachments have common historical causes, cannot be performed unless we establish a clear conceptual separation between coups and legal removals. Moreover, lumping together military coups and legal removals in a single category would obscure that some factors, such as the size of the president’s coalition in congress, influence the risk of impeachment but do not alter the probability of a military coup (Pérez-Liñán and Polga-Hecimovich, 2017).

To avoid these analytical problems, I propose that we reserve “legislative coup” to refer exclusively to situations in which congress acts to legitimize a military operation against the president (Pérez-Liñán, 2007). Under this definition, the congressional removal of Brazilian president Carlos Luz in 1955 was a legislative coup, but the impeachment of Panamanian president José Ramón Guizado on the same year was not; the removal of Honduran president Manuel Zelaya in 2009 was a legislative coup, but the impeachment of Paraguayan president Fernando Lugo in 2012 was not. The adoption of this terminology does not require the acceptance of congressional abuses. Guizado was unjustly accused of a conspiracy to kill his predecessor and incarcerated under this false pretense. Lugo was impeached within 48 hours, without any possibility of defense. Both episodes are blatant examples of congressional abuses in the use of impeachment, even if we do not label them as coups.

Besides any analytical problems, the rhetorical strategy linking coups and impeachments is politically problematic. It is understandable that the concept of coup d’état will be contested in reaction to impeachment processes that are seen as illegitimate by vast segments of society (Lopes and Albuquerque, 2018). However, this strategy has three negative implications: it suggests that social movements that mobilized in favor of impeachment processes in the 1990s were antidemocratic; it signals to public opinion that advocating a military coup is morally equivalent to calling for an impeachment; and it suggests that congressional abuse of power constitutes the main threat to democracy in the twenty-first century, even though democratic backsliding usually originates in the executive branch.

First, if we re-label legally dubious impeachments as parliamentary coups, we are forced to conclude that many progressive forces in the 1990s were coup supporters. The first section documented that most impeachment processes have been questionable in terms of motivations or procedure. Moreover, most presidents removed from office since 1992 were on the right, not on the left, of the political spectrum. In virtually every case of presidential impeachment, declaration of incapacity, or anticipated resignation – with the notable exception of Fernando Lugo’s ouster – ample protests demanded the president’s resignation in the streets. Existing evidence suggests that these protests formed “street coalitions” linking heterogeneous groups and different social classes (Pérez-Liñán, 2008). Thus, labeling dubious impeachments as coups would imply that the progressive parties and broad social movements that supported the resignation or impeachment of neoliberal presidents in the 1990s were coup plotters. This characterization, in most cases, would be unfair to the actors involved and distort the historical record.

While contemporary observers tend to link impeachments with coups to delegitimize the use of impeachment, observers of the 1990s, in contrast, tended to link social protests with impeachments to legitimize the role of social movements. Because the constitutional basis for mass protests forcing presidents to resign was controversial, Leon Zamosc (2012) characterized the popular overthrow of neoliberal presidents benevolently as “popular impeachments”.

The second political consequence of equating impeachments with coups is the blurring of moral distinctions between the use of legal strategies to undermine the president and the use of military force to terminate a democratic regime. As a result, this rhetorical strategy can have the unintended consequence of naturalizing military coups. Critics of impeachment intend to convey that a manufactured impeachment is as bad as a military coup. Part of the public may hear the opposite message: that a military coup is as good as a manufactured impeachment.

To illustrate this problem, consider the results of the study A Cara da Democracia [The Face of Democracy], conducted by the Instituto da Democracia e da Democratização da Comunicação [Institute of Democracry and of Democratization of Communication] in March 2018. 3 The project interviewed 2500 Brazilian citizens in 26 states, asking them – among other things – if they believed that the impeachment of Dilma Rousseff had been a coup or had been part of the normal democratic process. The survey also asked citizens whether they would justify a military coup in a situation of “great corruption”. 4 About 52 percent of respondents identified the impeachment as a coup, showing serious concerns about its legitimacy, and the remaining 48 percent saw it as a normal element of the democratic process. More critically, about 51 percent were willing to justify a military coup in a context of corruption, and 49 opposed military intervention. 5

Table 1 presents a simple cross-tabulation of the responses for the two questions. The distribution of attitudes shows four distinct blocs of citizens, although hardline positions on corruption seem to drive support for impeachment as well as for military intervention. Among those who question the legitimacy of the impeachment, a majority (53%) oppose a military coup against corruption; among those who accept the impeachment as normal, a majority (55%) would entertain the possibility of military intervention. This means that 25% of the sample (537 respondents in cell 1) believes that the impeachment of Dilma Rousseff was a coup – and that the coup was in fact justified .

Table 1 Opinions on Rousseff’s Impeachment and Military Coups 

The impeachment A Military Coup would be Total
Was Justified Not justified
A Coup [1]537 [2]603 1140
47% 53% 100%
Normal element of democratic process [3] 572 [4] 467 1039
55% 45% 100%
Total 1109 1070 2179

Source: A Cara da Democracia (2018).

χ 2 = 17.3 (p < 0.01).

As a mental experiment, imagine that we are able to convince 70% of the people who see the 2016 impeachment as “normal” that the process was in fact a coup. This would mean that, in a new survey, 86% of the valid responses (1867 respondents out of 2179) would now question the legitimacy of the impeachment, as opposed to the current 52% (1140). I surmise that this transformation would create an unexpected challenge for Brazilian democracy. Assuming that the new bloc preserved the current 47-53 majority against military intervention, we would still observe some 880 respondents in cell 1. That is, 40% of the sample would now believe that a coup took place in Brazil, and that the coup was fully justified by politicians’corruption.

The point of this mental experiment is not to assert a counterfactual – little can be said without serious experimental research in this behavioralarea – but to advance a hypothesis. Rhetorical strategies linking impeachments and coups may have the unintended consequence of expanding the proportion of citizens who naturalize the idea of coups (broadly defined) as part of an epic narrative against corruption.

This leads to the third unexpected consequence. By focusing too much on impeachment, we may miss the real threat to democracy. In a historical context in which military coups are rare, the link between impeachments and coups conveys that congressional abuses of power have become the main threat to democracy. However, available evidence suggests that this is not the case. Strong presidents, rather than strong legislatures, constitute the main challenge for democratic survival in the contemporary era. This is not a trivial issue when we consider that, in the A Cara da Democracia survey, about 12 percent of respondents indicate that they would accept a presidential autogolpe [self coup] against congress “if the country confronts difficulties”, and 26 percent agree that the president (or congress) could ignore the STF [Supreme Federal Court] if the judiciary “interferes with the government”. 6 I turn to this issue in the next section.

Is impeachment the new threat for democracy?

Reflecting on recent experiences of democratic backsliding in countries such as Hungary, Turkey, and Venezuela, Bermeo (2016) concluded that “executive aggrandizement” is a distinctive feature of those experiences. Traditional military coups, overt electoral fraud, and presidential autogolpes have become quite rare in our times. However, the expansion of executive power is a consistent threat across the globe. By contrast to traditional coups, contemporary episodes of backsliding occur “without executive replacement and at a slower pace […] when elected executives weaken checks on executive power one by one” (Bermeo, 2016, p. 10). Svolik (2015) similarly showed that the end of the Cold War reduced the risk of military coups worldwide, but not the risk represented by “incumbent takeovers”.

Bermeo’s thesis about executive aggrandizement implies that presidents who gain extensive control over the legislature and the judiciary – that is, presidents who are completely shielded from impeachment – represent a contemporary source of democratic erosion. However, the history of Latin America suggests that this is not a new phenomenon. Hegemonic presidents also created democratic instability in the past, for two reasons. Some presidents, like Juan Perón in Argentina (1946- -1955) concentrated institutional power successfully. Unconstrained by other institutions, they were able to exclude the opposition and undermine democracy in the way described by Bermeo. Other presidents, like Rómulo Gallegos in Venezuela (1948), belonged to mass parties with great control over other institutions, but they never consolidated their hegemony. Opposition leaders, fearing an unconstrained president, supported military intervention against the democratically elected government as a preemptive move. Democracies died in both cases, killed by the ambitions of the president or by the fears of the opposition.

To investigate the consequences of presidential hegemony for democratic stability, we developed a research project with Adolfo Garcé and Daniela Vairo at Universidad de la República in Uruguay. 7 The study covers all years of democracy in 18 Latin American countries (Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, El Salvador, Guatemala, Honduras, Mexico, Nicaragua, Panama, Paraguay, Peru, Uruguay and Venezuela) between 1925 and 2010 (n = 830). We measure presidential hegemony as a simple average of four indicators: the percentage of seats controlled by (1) the president’s party and (2) the president’s coalition in the lower House of Congress, and the percentage of judges nominated by (3) the president and (4) any presidents of the same party to the Supreme Court or Constitutional Tribunal. Using this information, we construct an index of presidential hegemony that varies yearly for each case of democracy (Pérez-Liñán, Schmidt and Vairo, 2017).

The dependent variable in this study is democratic breakdown, that is, the establishment of an authoritarian regime, as coded by Mainwaring and Pérez-Liñán (2013). We therefore model the risk of democratic breakdown using presidential hegemony as the main explanatory variable. 8 Although the details of the analysis transcend the scope of this brief section, Figure 2 summarizes the main results of the study.

Source: Estimates based on Pérez-Liñán, Schmidt, and Vairo (2017).

Figure 2 Estimated Risk of Democratic Breakdown 

The horizontal axis in Figure 2 represents the degree of presidential hegemony. Values above 50 indicate that the executive branch has considerable control over the legislature and the judiciary. The vertical axis depicts the probability of democratic breakdown. Panel 2.1 reflects the expected risk of breakdown in a statistical model in which presidential hegemony is the only independent variable, while Panel 2.2 presents the result of a model that includes additional control variables, such as the president’s constitutional powers, the effective number of parties, the proportion of democracies in the region, per capita income, and economic growth. (The grey bands around the predicted values reflect the 95% confidence interval.)

Both panels in Figure 2 lead to the same conclusion: the greater the capacity of the president to control the congress and the judiciary, the greater the risk of democratic instability. The history of Latin America teaches us that the main threat to democracy is not weak presidents undermined by impeachments, but extremely strong presidents who are potentially able to take over the system. Hegemonic presidents control other institutions, suppress political competition, and often trigger undemocratic reactions from the opposition.

Conclusions: on the value of dysfunctional institutions

After a rapid wave of democratization transformed Latin America at the end of the 20 th century, traditional military coups became unlikely. However, presidential impeachments emerged as the most common instrument employed by the opposition to remove unpopular presidents. The previous sections showed that most episodes of impeachment were controversial because of their motivations or their procedures. Moreover, some historical conditions that prompted military coups during the Cold War – economic recessions, mass protests, and political radicalization – are contemporary drivers of presidential impeachment.

Although impeachments can be seen as the functional equivalent of traditional military coups in the twenty-first century, this essay has argued that stretching the concept of coup d’État to label controversial impeachments is problematic for analytic and for political reasons. Analytically, we cannot conduct empirical research to identify similarities and differences unless we have a clear conceptual distinction between the two categories. Politically, the identification of impeachments and coups creates unexpected challenges: it implies – in revisionist fashion – that social movements that advocated for impeachments in the 1990s were antidemocratic, it naturalizes the role of military intervention in anti-corruption narratives, and it obscures the fact that powerful presidents – not powerful legislators – are the main peril for democratic stability.

There is no doubt that congress – in Latin America as well as in the United States – has misused its powers of impeachment (and declaration of incapacity) in several occasions. Yet, a more appropriate way to describe this pattern is to say that political elites are distorting the institution of impeachment to deploy it as the equivalent of a vote of no-confidence in a parliamentary system. Large opposition majorities use impeachments or declarations of incapacity to remove weak presidents, invoking alleged crimes when the executive is unpopular but not when the executive commands solid approval rates. In this way, legislators bend one of the defining principles of presidential constitutions: the fixed terms in office for the president and congress.

Until few years ago, the literature on presidentialism observed this “parlamentarization” of presidential systems with some optimism (Hochstetler and Samuels, 2011; Marsteintredet and Berntzen, 2008). After all, the use of constitutional mechanisms to resolve executive-legislative crises was much better than the termination of executive-legislative politics by a brutal military regime. Yet, recent episodes have led to a more nuanced interpretation of this process, and experts are increasingly warning about distortion and misuse of impeachment (Carey et al., 2018).

From a broader comparative perspective, however, congressional abuses may be the lesser concern. The experience of Latin America in the twentieth century indicates that the main source of democratic instability has not been a strong congress that abuses its powers, or a rogue judicial branch, but a strong president who gains control over congress and the judiciary. Although legislators and judges may occasionally destabilize elected governments, they lack sufficient command of patronage, budgetary resources, and security forces to impose an authoritarian regime. Only the executive branch, with vast resources at its disposal, can sustain a lasting process of democratic backsliding when it acts without constraints. The Venezuelan case is a contemporary reminder of how hegemonic presidents, free from checks and balances, suffocate political competition(Corrales, 2011).

Brazilian readers, shaken by the turbulent waters of coalitional presidentialism, may see the risk posed by hegemonic presidents as a very distant threat.This is why a comparative perspective, in my opinion, becomes particularly useful. The comparative analysis presented in this essay offers an important lesson: dysfunctional institutions may be a blessing in disguise.

Dysfunctional institutions may be a blessing because they prevent the consolidation of executive power. The United States currently illustrates this point well. President Trump would be a much more serious threat for American democracy if it were not the case that congress is often paralyzed. Even though the president formally enjoys a majority in congress, the Republican Party has been unwilling to support some of his key policies. The literature on presidentialism traditionally worriesabout deadlock and immobility, but in the contemporary US case, this kind of paralysis may be the best outcome. We must recognize the value of dysfunctional institutions in a context in which the president’s unilateral control would be a serious problem for democracy.

At the same time, a history of dysfunctional institutions may ultimately inspire popular calls for strong leadership, demands for a reconstitution of executive power that potentially creates conditions for democratic backsliding. Reflecting on the Brazilian case, Santos and Guarnieri (2016, p. 495) wondered “whether the resulting vector of this new stage[…], which will probably not reveal itself until after the general elections of 2018, will incline towards an explicit form of anti-establishment and anti-politics stance […]”. This is the context in which messianic leaders often emerge, driven by popular hopes that a strong president will purge the country from corrupt politicians.

In the short run, the misuse of impeachment may produce presidents that are too weak and fragile administrations. In the long run, reactions against this pattern may encourage popular support for presidents who are too strong. Even if presidential hegemony looks like a very distant threat, we must be aware of its consequences. Presidential hegemony, more than presidential impeachment, will constitute the main threat to democracy in the twenty-first century.


ALVAREZ, Michael E. & Leiv MARSTEINTREDET. (2010), “Presidential and democratic breakdowns in Latin America: similar causes, different outcomes”, in M. Llanos and L. Marsteintredet (eds.), Presidential breakdowns in Latin America , New York, Palgrave Macmillan, pp. 33-52. [ Links ]

AVELAR, Idelber. (2017a), “The June 2013 uprisings and the waning of lulismo in Brazil: of antagonism, contradiction, and oxymoron”. Luso-Brazilian Review , 54 (1): 9-27. [ Links ]

AVELAR, Idelber. (2017b), “A response to Fabiano Santos and Fernando Guarnieri”. Journal of Latin American Cultural Studies , 26 (2): 341-350. [ Links ]

BANKS, Arthur S. & WILSON, Kenneth A. (2012), Cross-National Time-Series Data Archive . Jerusalem, Israel, Databanks International. [ Links ]

BERMEO, Nancy. (2016), “On democratic backsliding”. Journal of Democracy 27 (1): 5-19. [ Links ]

CANNON, Barry. (2016), The right in Latin America: elite power, hegemony and the Struggle for the State . New York, Routledge. [ Links ]

CAREY, John M.; CORRALES, Javier; LLANOS, Mariana; MARSTEINTREDET, Leivand & PÉREZ-LIÑÁN, Aníbal. (2018), “The threat of impeachment can push presidents out the door. But there’s a catch”. The Washington Post . Available in . [ Links ]

COLLIER, David & LEVITSKY, Steven. (1997), “Democracy with adjectives: conceptual innovation in comparative research”. World Politics , 49 (3): 430-451. [ Links ]

COLLIER, David & MAHON, James E. (1993), “Conceptual “stretching” revisited: adapting categories in comparative analysis”. American Political Science Review , 87 (4): 845-855. [ Links ]

COLLOR, Fernando. (2016), Réplica para a história: uma catarse . Brasília, Senado Federal. [ Links ]

CORRALES, Javier. (2011), “Why polarize? Advantages and disadvantages of a rational-choice analysis of government-opposition relations under Hugo Chávez”, in T. Ponniah and J. Eastwood, The revolution in Venezuela: social and political change under Chávez , Cambridge, MA, Harvard University Press, pp. 67-97. [ Links ]

GOERTZ, Gary. (2006), Social science concepts: a user’s guide . Princeton, Princeton University Press. [ Links ]

HANMER, Michael J. & KALKAN, Kerem Ozan. (2013), “Behind the curve: clarifying the best approach to calculating predicted probabilities and marginal effects from limited dependent variable models”. American Journal of Political Science , 57 (1): 263-277. [ Links ]

HOCHSTETLER, Kathryn & SAMUELS, David. (2011), “Crisis and rapid reequilibration: the consequences of presidential challenge and failure in Latin America”. Comparative Politics , 43 (2): 127-145. [ Links ]

HOCHSTETLER, Kathtryn. (2006), “Rethinking presidentialism: challenges and presidential falls in South America”. Comparative Politics , 38 (4): 401-418. [ Links ]

LINZ, Juan J. (1978), Crisis, breakdown, and reequilibration . Baltimore, Johns Hopkins University Press. [ Links ]

LOPES, Monalisa Soares & ALBUQUERQUE, Grazielle. (2018), “O nome das coisas: a disputa simbólica impeachment versus golpe na mídia internacional”. Revista de Ciências Sociais , 49 (1): 279-316. [ Links ]

MAINWARING, Scott & PÉREZ-LIÑÁN, Aníbal. (2013), Democracies and dictatorships in Latin America: emergence, survival, and fall . Cambridge, Cambridge University Press. [ Links ]

MALAMUD, Andrés & MARSTEINTREDET, Leiv. (2017), “Coup with adjectives: conceptual stretching and innovation in comparative research”. Presented at Asociación Latinoamericana de Ciencia Política (Alacip), Montevideo, July 26-28. [ Links ]

MARSTEINTREDET, Leiv & BERNTZEN, Einar. (2008), “Reducing the perils of presidentialism in Latin America through presidential interruptions”. Comparative Politics , 41 (1): 83-101. [ Links ]

MUSTAPIC, Ana María. (2005), “Inestabilidad sin colapso – la renuncia de los presidentes: Argentina en el año 2001”. Desarrollo Económico , 45 (178): 263-280. [ Links ]

NUNES, Felipe & MELO, Carlos Ranulfo. (2017), “Impeachment, political crisis and democracy in Brazil”. Revista de Ciencia Política , 37 (2): 281-304. [ Links ]

PÉREZ-LIÑÁN, Aníbal. (2007), Presidential impeachment and the new political instability in Latin America . Cambridge, Cambridge University Press. [ Links ]

PÉREZ-LIÑÁN, Aníbal. (2008), “Instituciones, coaliciones callejeras e inestabilidad política: perspectivas teóricas sobre las crisis presidenciales”. América Latina Hoy , 49: 105-126. [ Links ]

PÉREZ-LIÑÁN, Aníbal & POLGA-HECIMOVICH, John. (2017), “Explaining military coups and impeachments in Latin America”. Democratization , 24 (5): 839-858. [ Links ]

PÉREZ-LIÑÁN, Aníbal & SCHMIDT, Nicolásand VAIRO, Daniela. (2017), “Power concentration and democratic instability in Latin America, 1925-2010.” Presented at Latin American Political Science Association (Alacip), Montevideo, July 26-28. [ Links ]

SANTOS, Fabiano & GUARNIERI, Fernando. (2016), “From protest to parliamentary coup: an overview of Brazil’s recent history”. Journal of Latin American Cultural Studies , 25 (4): 485-494. [ Links ]

SARTORI, Giovanni. (1970), “Concept misformation in comparative politics”. American Political Science Review , 64 (4): 1033-1053. [ Links ]

SVOLIK, Milan W. (2015), “Which democracies will last? Coups, incumbent takeovers, and the dynamic of democratic consolidation”. British Journal of Political Science , 45 (4): 715-738. [ Links ]

ZAMOSC, Leon. (2012), “Popular impeachments: Ecuador in comparative perspective”, in M. Sznajder, L. Roniger and C. A. Forment (eds.), Shifting frontiers of citizenship: The Latin American experience , Leiden, Brill, pp. 237-266. [ Links ]


1Only in three other episodes (Guatemala in 1993, Ecuador in 2000, Honduras in 2009), presidents were ousted by a military operation. In Venezuela, a military coup in 2002 was only successful for 48 hours, before Hugo Chávez returned to office.

2All six simulations follow an “observed value” approach (Hanmer and Kalkan, 2013). To create each figure, we set the value of the key independent variable (for example, Per capita growth, in panel 1.1.1) at an arbitrary value (say, -8%) for all 729 observations in the sample. Other independent variables are kept at their observed historical values for each case. Using the parameters of the statistical model, we then predict the probability of the outcome (Legal removal, in 1.1.1) for all observations in this hypothetical sample. The average risk for this sample is reported as the point estimate for that situation in the graph. We then modify the sample, adopting a second value for the independent variable (say, -6%), and repeat the exercise.

3See . I am indebted to Leonardo Avritzer for sharing the questionnaire and data.

4The survey items are Q12.4 (Circunstâncias que justificariam um golpe militar? – Diante de muita corrupção [Circunstances that could justify a military coup? – In face of much corruption]) and Q16 (Para algumas pessoas o processo de impeachment foi um golpe; enquanto outras acharam que foi algo normal. Com qual dessas opiniões você concorda mais? [For some people, the process of impeachment was a coup, while other consider it something normal. With which of these opinions do you agree more?)

5Around 13 percent of respondents failed to provide an answer for one of the two questions, and they are excluded from the count.

6Items are Q13 (O/a Sr/a acredita que quando o país está enfrentando dificuldades é justificável que o Presidente da República feche o Congresso e governe sem o Congresso?) [Do you believe that when the country is experiencing dificulties, it is justifiable for the President to close the congress and govern without it?] and Q47 (O/a Sr/a concorda que quando o STF interfere no trabalho do governo, ele pode ser ignorado pelo presidente ou pelo congresso?)[Do you agree that when the Supreme Federal Court interferes with the work of the government, it can be ignored by the president or by the congress?].

7This study is supported by Uruguay’s National Research and Innovation Agency (ANII) under Grant FCE_1_2014_1_103565

8For this purpose, we use logit and probit models with random effects. Simulations in Figure 2 are based on probit models, and were created using the observed value approach, as discussed in the note for Figure 1 .

*This paper is based on a presentation made at the Anpocs annual meeting, Caxambu-MG, October 25, 2017. I am indebted to Fabiano Santos and Gabriela Rosa for inviting me to write this essay and for their patience waiting for its completion.

Creative Commons License This is an Open Access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.