5. The long road to democracy in Brazil*

Thirty years ago, in the late 1980s, Francis Fukuyama began to formulate his ideas on the late 20th century triumph of liberal democracy (and free market capitalism) worldwide – ‘the universalization of Western liberal democracy as the final form of human government’. They were presented, first in a series of lectures at the John M. Olin Center for Inquiry into the Theory and Practice of Democracy at the University of Chicago during the academic year 1988–9, and then in an article published in The National Interest (summer 1989). At the time, not only were China, the Soviet Union and much of Eastern and Central Europe still under Communist rule but, in the western hemisphere, besides the notoriously complex case of Mexico, Brazil – the fifth largest country in the world, with the fifth largest population (160 million) – was not an insignificant exception to Fukuyamian triumphalism.

Since its independence from Portugal in 1822 Brazil had had a long history of elections that compared favourably with most countries in the world. Under the empire (1822–89), under the First Republic (1889–1930), in the aftermath of the revolution of 1930, in the period after the Second World War (1945– 64), even under military dictatorship (1964–85), elections had been regularly held in Brazil. There had in fact been only one period in the entire modern history of Brazil without elections: the authoritarian Estado Novo (1937– 45). There is, of course, more to democracy than elections, but elections are fundamental to liberal representative democracy. Brazilian elections, however, were rarely completely honest; although there was always some measure of competition between different parties and candidates, they were usually not entirely freely contested; and most important, although wider sections of the Brazilian population were gradually incorporated into the political process, the level of participation always fell some way short of universal suffrage. The painfully slow process of political liberalisation, initiated towards the end of the military regime, continued after the transition to civilian rule in 1985 and leading eventually to democracy, part of Samuel Huntington’s ‘third wave’ of global democratisation which started in southern Europe in the 1970s and spread to Latin America in the 1980s – was still not yet complete. (And the Brazilian economy remained – and remains – one of the most closed and state-regulated – with one of the largest public sectors – in the capitalist world.)

However, by the time Fukuyama published his book The End of History and the Last Man in 1992, Brazil could unquestionably be counted as a democracy. Under the Constitution in 1988 the presidential elections at the end of 1989 and the congressional elections at the end of 1990 were free, fair, competitive and based on universal suffrage. Brazil had become, after India and the United States, the third largest democracy in the world. Brazil’s democracy has several structural flaws, and political reform has been the subject of debate for more than two decades, but it has so far survived, despite fears at the time of its birth that it might not with little in the past to justify much optimism that it would. Nevertheless, the political/institutional crisis arising from the impeachment of President Dilma Rouseff of the Partido dos Trabalhadores (PT, Workers’ Party) in 2016 was a reminder that liberal representative democracy in Brazil is relatively new and perhaps less than fully consolidated.

I

In contrast to the 13 Colonies in British North America, but like colonial Spanish America, Brazil served no significant apprenticeship in representative self-government under Portuguese colonial rule. For three centuries Brazil was governed by Crown-appointed governors-general (or viceroys), captains-general (or governors), high court judges, magistrates and other lesser bureaucrats. It has been argued that the municipal councils (senados da câmara), like the cabildos in late colonial Spanish America, were rather more than simply self-perpetuating oligarchies: councilmen [vereadores] and some local judges were chosen or indirectly ‘elected’ by homens bons [men of wealth and good standing]. However, the number of ‘voters’ was always small and the powers of the câmaras severely restricted.

The first general elections in Brazil (albeit on an extremely limited suffrage) were held to elect delegates to the Cortes, who, beginning in April 1821, were summoned to meet in Lisbon in the aftermath of the Portuguese liberal revolution of 1820. By that time, as a consequence of the transfer of the Portuguese court from Lisbon to Rio de Janeiro in 1807–8 during the Napoleonic wars Brazil was already no longer strictly speaking a Portuguese colony but an equal partner in a dual monarchy: O Reino Unido de Portugal, Brasil e Algarves. A year later, in June 1822, as Brazil finally moved towards full separation from Portugal, there followed elections to a Constituent Assembly in Rio de Janeiro. They were indirect elections on a strictly limited suffrage after the extreme liberals or radicals of the period (many of them republicans) failed to secure direct popular elections.

The independence of Brazil in 1822 can be regarded as part of the so-called ‘democratic revolution’ of the Atlantic world in the late 18th and early 19th centuries in the sense that liberal democratic ideas were widely proclaimed in the struggle against Portuguese colonialism and absolutism. There was, however, never any intention of establishing in Brazil anything that, even at the time, looked remotely like liberal representative democracy based, however theoretically, on the sovereignty of the people. The main aim of the leaders and supporters of Brazilian independence – fazendeiros [plantation owners], especially in the province of Rio de Janeiro but to a lesser extent also Bahia and Pernambuco, merchants in the principal cities and some bureaucrats – was to achieve political autonomy without sacrificing the stability so crucial for the maintenance of Brazil’s territorial unity and existing socio-economic structures built, above all, on African slavery. (Brazil’s population at the time, in a vast territory of three million square miles, was between four and five million, less than a third white, more than a third slave.)

Once decided upon, independence was secured quickly and peacefully – without either a long and bloody war with the colonial power or a civil war (in sharp contrast to events in Spanish America). The transition from Portuguese colony to independent state was characterised by political, economic and social continuity. The existing Portuguese state apparatus never ceased to function. The economy suffered no major dislocation. Above all, as well as the existing pattern of land ownership, the institution of slavery survived – in all regions of the country and, while heavily concentrated in plantation agriculture, in all sections of the economy and society, rural and urban. Unlike the newly independent Spanish American states, Brazil did not even become a republic. Uniquely, Brazil proclaimed itself an empire, with Dom Pedro, the son of King João VI of Portugal and heir to the Portuguese throne, becoming independent Brazil’s first emperor (succeeded on his abdication in 1831 by his five-year-old son who eventually became Dom Pedro II).

The Constituent Assembly elected in June 1822 was inaugurated in May 1823 but forcibly dissolved in November of the same year. The constitution of the independent Brazilian empire was imposed by Emperor D. Pedro in March 1824. Under the Constitution of 1824 there was an elected Chamber of Deputies (and elected provincial assemblies and municipal câmaras). But governments were only to a limited extent responsible to them. Power was concentrated in the hands of the hereditary emperor himself, his chosen ministers, the counsellors of state he appointed (for life), the provincial presidents he also appointed, and a Senate (with senators appointed, also for life, by the emperor, though from lists of three elected by each province). The purpose of an election was not to form a government but to sustain a government already chosen by the emperor.

The parliamentary elections of the empire were contested by two principal parties, Liberals and Conservatives. In the middle decades of the 19th century, the golden age of the empire, the level of political participation in Brazil was potentially surprisingly high: men (not women, of course) who were 25 years old (21 if married), Catholic, born free, and with a quite low annual income from property, trade or employment had the right to vote in elections for the Chamber of Deputies. Reliable statistics are hard to come by, but it is generally agreed that around one million Brazilians could vote (i.e. half the free adult male population, including many of quite modest means and illiterates, regardless of colour). (This compares favourably with the electorate in England, for example, after the Reform Act of 1832 and even after the Reform Act of 1867.) However, a much smaller proportion of the population registered to vote. And the elections were indirect. The so-called votantes elected eleitores (who were required to have a higher annual income), and only eleitores – some 20,000 of them in 1870, for example – had the right to vote for deputados [deputies]. Moreover, the turn-out was generally extremely low. This was hardly democracia coroada [crowned democracy], as the historian João Camillo de Oliveira Torres entitled a book published in 1957 on the political system of the empire.¹

The number of Brazilians legally enfranchised was severely reduced as a result of the Saraiva law of 1881. During the last quarter of the 19th century, as the coffee economy expanded and the shift from slave to free labour finally gathered momentum, making the final abolition of slavery increasingly inevitable, there was a growing fear amongst the dominant political class – shared by many liberal reformers – that ex-slaves, in the rural areas but more particularly in the rapidly expanding urban areas, would readily acquire the low income sufficient to secure the right to vote. Elections for the Chamber of Deputies were made direct; the property/income qualification for eleitores was removed; and non-Catholics, naturalised citizens (though not resident foreign immigrants) and ex-slaves (freedmen) were eligible to become voters. However, undermining somewhat these apparent liberal/democratic advances, a new requirement for future voter registration was introduced for the first time: namely, education as measured by a literacy test or rather a capacity to sign one’s name. (The individual was responsible for his own voter registration.) This in a country in which 80–85 per cent of the population was illiterate. (In England, John Stuart Mill, the great apostle of liberal democracy, also argued against giving the vote to illiterates, but Mill at least believed in the rapid expansion of public education to reduce the level of illiteracy, not something advocated by many people in Brazil in the late 19th century.) Thus, in the final decade of the empire, while the number of eleitores increased (in 1886 117,000 voted in the elections for the Chamber of Deputies), the vast majority of Brazilians were consciously and deliberately excluded from political participation.²

Liberalism may have been the dominant ideology in 19th-century Brazil but, as in Spanish America, it was liberalism of a predominantly and increasingly conservative variety as it was forced to adjust to the realities of an authoritarian political culture, economic underdevelopment and, most of all, a society deeply stratified (and along racial lines).

Brazil was not only the last independent state in the Americas to abolish slavery, it was also the last to declare itself a republic. It was no accident that the republic was finally proclaimed in 1889, the centenary year of the French revolution. The ideology of republicanism, especially radical republicanism, supported by progressive urban middle-class intellectuals, was profoundly French-inspired. But there was no revolution in Brazil in 1889. The Brazilian republic came out of a military coup born of a conspiracy between a small number of army officers and representatives of the rising coffee-producing landed oligarchy of the state of São Paulo. Like the transition from colony to empire, the transition from empire to republic was marked more by fundamental social and economic continuity than by change.

Under the republican constitution of 1891 the president, state governors/ presidents,³ municipal prefeitos (mayors) as well as both houses of Congress (the Senate and Chamber of Deputies), state assemblies and municipal councils were all elected. Elections under the First Republic (1889–1930), however, were not much less dishonest than elections under the empire, possibly more so, controlled as they were for the most part by state governments and local political bosses known as coroneis (because many had once held the rank of colonel in the National Guard) representing powerful landed oligarchies, often with what amounted to private armies, especially in the more backward states of the north-east and north.

During the First Republic elections were contested by state parties only and in each state the Republican party – the Partido Republicano Paulista (PRP), Partido Republicano Mineiro (PRM), Partido Republicano Riograndense (PRR) and so forth – was dominant. The outcome of elections for president of the republic was pre-determined by prior agreement between state governors [a política dos governadores]. No ‘official’ candidate backed by the governors and Republican political machines of at least one (and it was usually both) of the two states with the largest electorates (Sao Paulo and Minas Gerais) and two or three of the largest second rank states (Rio Grande do Sul, Rio de Janeiro, Bahia, Pernambuco) ever lost, and no ‘opposition’ candidate ever won a presidential election during the First Republic. One of the first decrees of the provisional government in November 1889, four days after the fall of the empire and establishment of a republic, reaffirmed that the only criterion for adult male voter registration was now literacy. The regulations for the election of a Constituent Assembly in 1890 reduced the minimum voting age from 25 to 21. By continuing to deny the vote to women and illiterate men, the republican Constitution of 1891 excluded from politics two thirds of Brazilians of voting age. (In the Constituent Assembly a greater effort was made to extend the suffrage to women than to illiterates: it failed.) Such was the neglect of public primary and secondary education during the First Republic, responsibility for which was devolved to the municípios and the states, that over 75 per cent of the population remained illiterate as late as 1920.

Nevertheless, the presidential and congressional elections of the early republic did represent a substantial advance in direct popular political participation compared with the late empire. There was an electorate of around one million, including sections of the emerging urban middle class and even some urban workers in Rio de Janeiro, São Paulo, Porto Alegre and elsewhere, and half to three quarters of a million voted.⁴ However, even in the city of Rio de Janeiro, the capital of the federal republic, with a population of over half a million in the first decade of the 20th century, only some 120,000 people (60 per cent of the adult male population) were literate and therefore had the right to vote, only 15–20 per cent of these registered to vote in national elections, and only 10 per cent actually voted.⁵ In the country as a whole, in even the most competitive presidential elections with the greatest degree of political mobilisation – for example, the elections of 1910 and 1919 in which Rui Barbosa, the great liberal jurist, stood as an opposition candidate (and lost) – less than 5 per cent of the adult population voted.⁶ It was not until 1930 that more than 10 per cent of the adult population voted in a presidential election. What has been called oligarchical democracy (surely an oxymoron) is, as a description of the political system of the Old Republic, as hard to swallow as is crowned democracy for the empire.

The revolution of 1930 which brought an end to the First Republic and the hegemony of the São Paulo coffee oligarchy was in no real sense a revolution at all. Getúlio Vargas, governor of Rio Grande do Sul and the defeated candidate in the elections in March, came to power in November as a result of an armed rebellion led by dissident members of the political elite, especially in Rio Grande do Sul and Minas Gerais but also São Paulo, and disaffected military officers, which triggered intervention by the federal army to remove President Washington Luis from office. It represented yet another shift in the balance of power between landed regional elites more than the emergence of new social forces and brought the military to the centre of power, where it remained for next 60 years.⁷ The provisional government dissolved Congress, state legislatures and municipal câmaras. State presidents/governors and municipal prefeitos were replaced by interventores nominated by the provisional president and state interventores respectively. For the first time since the implementation of the 1824 Constitution Brazil had no elected politicians in either the executive or legislative branches of government.

However, elections were promised for a new Constituent Assembly. A commission set up by the provisional government in 1932 recommended the introduction of the secret ballot and a system of electoral supervision [justiça eleitoral]. A Tribunal Superior Eleitoral [Supreme Electoral Court] and regional tribunals would be responsible for the registration of voters, parties and candidates, the vote itself and, crucially, the count. (In practice, however, the new electoral legislation was not fully implemented until the establishment of the Liberal Republic after the Second World War.) An electoral law in 1932 lowered the voting age to 18 and, more importantly, for the first time gave women the vote (provided they were literate).⁸ Registration remained the responsibility of the individual, but alistamento ex-offício, automatic voter registration for lists of employees, male and female, in government employment and larger public and private companies was now permitted (and actually enabled some illiterates to be registered). Voting was now obligatory for men up to the age of 70, with sanctions for failure to vote. There was, however, no immediate significant expansion in the number of Brazilians registering to vote. Women in particular were slow to register; only 15 per cent of those eligible to vote in the elections for the Constituent Assembly in May 1933 did so, and only one woman, Carlota Pereira de Queiróz from São Paulo, was elected.

The Constituent Assembly met for first time in November, and promulgated a new Constitution in July 1934. The Assembly elected Getúlio Vargas president for a further four years, and determined that elections for the Chamber of Deputies and for state constituent assemblies (which would eventually become state assemblies and elect state governors and state representatives in a federal Senate) would be held in October. Seven per cent of the population (2.66 million) were registered to vote in the October 1934 elections for the Chamber of Deputies, 7.1 per cent of the population (compared with 3.9 per cent in 1933), ranging from 2.4 per cent in Amazonas to 10.2 per cent in Rio de Janeiro and 11.2 per cent in Rio Grande do Sul.⁹

Under the Constitution of 1934 direct elections for president and state governors were due to be held in January 1938. But the emergence in 1935 of the radical Aliança Nacional Libertadora (ANL) and a failed attempt by the Communist party to seize power in November 1935 (see Essay 7) led to the imposition of a state of siege, which remained in force for two years. When elections threatened to produce a result unacceptable to Vargas and the military – either a victory for Armando Sales and a restoration of the ‘liberal democracy’ (sic) of the First Republic dominated by state oligarchies and especially the coffee interests of São Paulo or a victory for José Américo de Almeida and the establishment of a populist government committed to improving the lot of the poor, they were aborted by military coup in November 1937. Under the authoritarian Estado Novo that followed Congress, state assemblies and municipal câmaras were once again dissolved and the scheduled elections for president and state governors cancelled. Getúlio Vargas was to remain in power for another eight years.¹⁰ No elections were held during Estado Novo (1937–45).

For 120 years after Brazil became an independent state the dominant class had successfully excluded the vast majority of Brazilians from the political system and at the same time prevented any serious challenge to its power and wealth.

II

The Liberal Republic established under the Constitution of 1946 at the end of the Second World War is usually taken to represent Brazil’s first ‘experiment in democracy’.¹¹ During 1944–5, the Vargas dictatorship had come under considerable pressure to democratise or at least liberalise the political system of the Estado Novo. The pressure was both domestic and external. (Brazil was one of the closest allies of the United States in the struggle for democracy against fascism.) In February 1945 Vargas finally promised ‘free’ elections before the end of the year, confident that he had the means (through control of the state apparatus) and support (especially from the ranks of the organised working class and white collar public employees who were the main beneficiaries of the economic and social policies of the Estado Novo) to win them. The process of ‘democratisation’ was initiated and controlled pelo alto, from above. But between May and October 1945 Brazil’s major cities experienced unprecedented mass political mobilisation, orchestrated in part by the Partido Communista Brasileiro (PCB) and, more particularly, by the so-called queremistas (from the slogan ‘Queremos Getulio’, ‘We want Getulio’) in favour of Vargas remaining in power (see Essay 6). There were growing fears among those conservative sectors in Brazil newly committed to ‘democracy’ that popular forces were being dangerously radicalised by both communism and populism. It took a soft intervention by the United States and another military coup (this time to remove Vargas from power) to guarantee the elections as scheduled for December. The presidential election was won by General Dutra, Vargas’s minister of war, the representative of the forces that had sustained the Estado Novo, not least because of the eleventh hour support offered by Vargas.

The presidential and congressional elections of December 1945 at the end of the Estado Novo were the first reasonably honest, competitive, relatively popular elections ever held in Brazil. Approximately 7.5 million Brazilian men and women aged 18 and over registered to vote (31 per cent of the adult population) and 6.2 million, 26 per cent of the adult population, voted. This was three times the number who had registered to vote and voted only 15 years earlier.¹² One explanation is that, although voter registration was generally by individual initiative, alistamento ex-officio through the workplace was more extensively used, not least now by sindicatos [labour unions], especially in the capital cities. In the country as a whole 23 per cent were registered in this way, but in the Federal District (Rio de Janeiro) it was 54 per cent and in São Paulo 31 per cent.¹³ And voting was compulsory.

The Congress elected in December 1945 transformed itself into a constituent assembly. Under the Constitution of September 1946, there would be elections for all three levels of government, executive (president and vice president, elected separately, the governors of the 20 states and municipal prefeitos) and legislative (Senate, Chamber of Deputies, state assemblies and municipal councils). Elections to executive office and to the Senate were to be by direct majority vote, elections to legislative bodies by proportional representation. The vote would be secret, and elections closely supervised by the Tribunal Superior Eleitoral (TSE). Furthermore, elections would be competitive, and contested for the first time by national parties.

Before 1945 the only national political parties or political movements in Brazil were the PCB, founded in 1922 and immediately declared illegal, and the fascist Ação Integralista Brasileira (AIB), founded in 1932 and declared illegal along with all other political parties during the Estado Novo. The three major parties formed in 1945, the Partido Social Democrático (PSD), the União Democrática Nacional (UDN) and the Partido Trabalhista Brasileiro (PTB), were obliged by law to have a national organisation, though the PSD and the UDN at least were essentially confederations of state based organisations. This was inevitable in a country the size of a continent (some of Brazil’s states were the size of the larger European countries), which was still predominantly rural and small town, in which there had only been state-based parties under the First Republic and in which the embryonic national parties which had contested the Constituent Assembly and Congress elections of 1933–4 and the (eventually aborted) presidential election of 1938 had been abolished with the promulgation of the Estado Novo. In the end a dozen political parties, mostly lacking a clearly defined identity based on history, ideology, programme or social base and only notionally national, were formed in the period 1945–8. During the Liberal Republic all presidential and vice-presidential elections were, however, won by candidates belonging to or supported by one of the three major parties. Between them the PSD, UDN and PTB also secured 90 per cent of the seats in the Chamber of Deputies and 95 per cent of the Senate seats in December 1945, 75–80 per cent of the seats in the Chamber between 1950 and 1962, 80–95 per cent of the Senate seats between 1954 and 1962.

As a result of a dramatic growth in the population (from 40 million in 1940 to 70 million in 1960), rapid urbanisation (35 per cent of the population was classified as urban in 1940, 45 per cent in 1960), economic growth, some modest improvement in literacy and a higher level of voter registration, there was a significant expansion of the electorate in the post-war period. It reached 18.2 million in 1962, 49 per cent of the adult population.¹⁴ As the vote was obligatory turnout remained high (79.6 per cent in 1962). However, under the constitution of 1946, like the constitutions of 1889 and 1934, more than half the adult population of Brazil, most particularly in the north and north-east and in rural areas more generally, remained disenfranchised by their illiteracy. The 1940 census recorded 56 per cent of the population illiterate, the 1950 census 48 per cent and the 1960 census 39 per cent. Of the 27.1 million Brazilians of voting age in presidential and congressional elections in 1950, 15.2 million were illiterate. In 1950 Congress restored to the individual sole responsibility for voter registration – on the face of it a liberal measure but in the circumstances of Brazil at the time it was a blow aimed at the political participation of the urban working class.

Brazil’s post war ‘democracy’ was limited in three other respects. Firstly, the PCB, the only significant party of the Left, which had been permitted to contest in both the presidential and congressional elections of December 1945 and gubernatorial, state assembly and municipal elections of January 1947 and had won around 10 per cent of the vote (half a million votes) in both, was in May 1947, at the beginning of the Cold War, once again declared illegal by the Tribunal Superior Eleitoral after only 18 months of de facto legality. Its 14 federal deputies and one senator lost their seats in Congress (see Essay 7). The PCB, which was not for its part fully committed to legal strategies and the electoral road to power, was effectively excluded from democratic politics – and remained so for the next forty years. Secondly, the distribution of seats in Congress under the 1946 Constitution ensured that the less populated, less developed, more politically conservative (that is to say, clientelistic) states, especially in the north and north-east, remained over represented at expense of the states of the south and south-east, especially São Paulo. The problem here was not simply that, as in the United States, all of Brazil’s states regardless of population had an equal number of seats in the Senate (three), but that representation in the Chamber of Deputies was also not proportional to population or electorate (one citizen, one vote, with more or less equal weight). The 1891 Constitution allocated each state one seat in the Chamber of Deputies per 70,000 people, with a minimum ‘floor’ of four seats per state. The 1934 Constitution allocated one seat per 150,000, one per 250,000 above 20 seats, and this was maintained under the 1946 Constitution. That is to say, states with populations over 3 million – three in 1934: Minas Gerais 6.5 million, São Paulo 6.3 million and Bahia 3.7 million; four in 1946: São Paulo 8.1 million, Minas Gerais 7.3 million, Bahia 4.3 million and Rio Grande 3.7 million – were under-represented in the Chamber of Deputies. States with populations of less than 1.5 million, for example, Amazonas, Sergipe and Mato Grosso, were over-represented. And in 1950 the minimum number of seats per state was raised from four to seven. Finally, and most important, the military retained in the post-war period the independent political power it had exercised since 1930. It was an integral part of the political system. It could and did intervene in politics and it remained largely beyond civilian control. Without military support it was impossible for any democratically elected president to survive in power. Nevertheless, along with Chile, Uruguay and Costa Rica, Brazil was one of only four ‘democracies’ in Latin America in the mid 1950s.

Brazil’s limited ‘democracy’ survived several political crises, notably those surrounding the suicide in August 1954 of Getúlio Vargas (who had been elected to the presidency in the second post-war elections in 1950) and the resignation in August 1961 of President Jânio Quadros, after only eight months in office. In the early 1960s, however, a number of factors, principally a sharp economic downturn but also including the impact of the Cuban revolution and Vatican II, combined to radicalise the popular forces to a level not seen before in Brazil. The ‘Right’ (including by now large sections of the urban middle class) was prepared to support (indeed encourage) a military coup if this was the only way of preventing radical economic and social change. An imprudent attempt by President João Goulart (1961–4) to create an opening to the Left led to his overthrow by the military on 31 March 1964, bringing to an end Brazil’s first ‘experiment with democracy’ (see Essay 6).¹⁵

III

The golpe militar [military coup] of 31 March – 1 April 1964 which overthrew the legally constituted government of President João Goulart made use of a good deal of democratic rhetoric: one of the proclaimed aims of the instigators of the ‘Revolution’ was the elimination of the threat, as they saw it, that the Goulart administration posed to Brazilian democracy. In the aftermath of the coup, however, by means of a series of Atos Institucionais [complementary acts] a new constitution in 1967, a revised constitution in 1969, constitutional amendments and various pacotes [packages of arbitrary measures] the military regime established in April 1964, while never entirely destroying them, radically remodelled and severely undermined the democratic institutions, albeit limited and flawed, established in Brazil at the end of the Second World War.

For 21 years, until the transition to civilian rule in March 1985, Brazilians lived under authoritarian military rule. A succession of five presidents, all of them senior (four star) generals, were first selected by the military high command and then indirectly ‘elected’ for a fixed term, the first three by Congress, the following two by an Electoral College consisting of senators, federal deputies and elected delegates from state assemblies, a majority of whose members were guaranteed (until 1984 at least) to support the military’s chosen candidate. From 1966 to 1978 state governors were similarly appointed by the military and then indirectly ‘elected’ by state assemblies or state electoral colleges. Only in 1965 and 1982 were state governors directly elected, half of them (11) in 1965, all 22 in 1982. Mayors of state capitals and cities of importance to ‘national security’ – originally 68, eventually almost a hundred – were also appointed by the military. Both houses of Congress and state legislatures were directly elected on schedule every four years and continued to function under the military regime (apart from one or two brief closures), though with their powers much curtailed. Moreover, electoral rules were frequently manipulated in the most arbitrary and blatant way to guarantee majorities for the pro-military ruling party.

In October 1965, 18 months after the golpe that brought an end to Brazil’s post-war Liberal Republic, all Brazil’s political parties were abolished, as they had been in 1937 at the outset of the Estado Novo. Two parties only, a pro-government Aliança Nacional Renovadora (ARENA) and a minority opposition Movimento Democrático Brasileiro (MDB), formed from the parties in the 1963–6 Congress after significant purges had been carried out, were permitted to contest the congressional elections in November 1966. Only in December 1979 was a multi-party system restored. The ‘reform’ was aimed at splitting the opposition, thus preventing a potential victory for the MDB in the congressional elections in 1982, which would threaten the military regime’s control of the presidential succession in 1985. While on the one hand, the Partido Democrático Social (PDS), the new government party, retained most of the old ARENA and therefore an absolute majority in both houses of Congress, the opposition found itself divided into four: the Partido do Movimento Democrático Brasileiro (PMDB), retaining the bulk of the old MDB; the Partido Popular (PP), dissident arenistas and some moderate members of the MDB; the right of centre Partido Trabalhista Brasileiro (PTB); the left of centre Partido Democrático Trabalhista (PDT). They were joined in 1980 by the Partido dos Trabalhadores (PT, Workers’ Party), the only party born outside Congress without ties to the traditional ‘political class’ (see Essay 7).

As a result of continued population growth (from 70 million in 1960 to 120 million in 1980), rapid urbanisation (45 per cent of the population was classified as urban in 1960, 70 per cent in 1980), and in the 1960s and 1970s for the first time real progress in direction of universal basic literacy, the electorate continued to grow steadily during the 21-year military dictatorship. By 1982 79 per cent of the Brazilian population of voting age was registered to vote (compared with 49 per cent in 1962). Illiterates (some 20 per cent of the adult population) were still excluded. Voting was obligatory, except for Brazilians over 70. In the elections of November 1982 for state governors (the first direct elections since October 1965), the Chamber of Deputies and one-third of the Senate, state assemblies, municipal councils and some prefeitos, 48 million Brazilians went to the polls.

The process that led to a transition to civilian rule in 1985, like the process that led to a limited form of democracy at the end of the Estado Novo in 1945, was initiated and controlled from above. It was not primarily a response by the military to opposition MDB/PMDB victories in the gubernatorial elections of 1982, nor the unexpectedly strong emergence of civil society in the form of new unionism in 1978–9 and the formation of the PT in 1980, nor even the extraordinary mass mobilisation in favour of diretas já (immediate direct presidential elections) in 1983–4¹⁶ – although these all played their part. Democracy was never the intended outcome. When it lost control of the presidential succession process, that is to say, when it was no longer able to count on a majority in the Electoral College, the military threw its weight behind a deal struck between PDS dissidents (who later formed the Partido da Frente Liberal, PFL) and the opposition PMDB under which the 75-year-old liberal-conservative opposition politician Tancredo Neves became the ‘official’ presidential candidate. Tancredo was duly indirectly ‘elected’, but never took office. He was taken ill on the eve of his inauguration and died a few weeks later. The presidency went to the Vice-President-Elect José Sarney, the former president of the PDS, the ruling party under the military regime, who therefore became the first civilian president of Brazil in more than two decades.¹⁷

IV

15 March 1985 witnessed a peaceful transition from military to civilian rule (but not yet to a fully fledged democracy) in Brazil. However, a series of constitutional amendments passed by Congress during the first months of the Sarney administration led to a significant democratisation of the electoral process. Firstly, illiterates (26 per cent of the total population in the 1980 Census, the majority black or pardo) finally gained the right to vote, though for them registration and voting would not be obligatory.¹⁸ Secondly, direct presidential elections were re-established (although without at this stage a date being fixed for the next election). Thirdly, elections for some 200 mayors of state capitals and other cities prevented from holding elections during the military dictatorship were scheduled for November 1985. Fourthly, the Federal District (Brasília) was given representation in Congress (eight deputies and three senators). Finally, less demanding rules for the creation of political parties were introduced and, specifically, parties of the Left, principally Brazil’s two communist parties – the Partido Comunista Brasileiro (PCB) and the Partido Comunista do Brasil (PCdoB), which had split from the PCB in 1962 – could now be legally registered. Throughout 1985, joining the parties created after the party reform of 1979, no fewer than 22 new parties were registered with the Tribunal Superior Eleitoral.

The municipal elections of November 1985 and the elections for Congress (the Chamber of Deputies and two thirds of the Senate), state governors and state assemblies a year later were the first elections in Brazil based on universal suffrage, though only 65,000 analfabetos [illiterates] had had time to register to vote in the first and less than half the total registered to vote in the second.¹⁹

The number of registered voters had increased to 69.3 million, of whom ten per cent were illiterate. Not only were 26 women elected to Congress in 1986, a small number but more than had been elected in the entire period 1932– 86, but also seven black members, including the first black, female deputy, Benedita da Silva (PT, Rio de Janeiro).

The Congress elected in November 1986 transformed itself into a constituent assembly which took 20 months to produce a new constitution (compared with three months to produce the 1891 Constitution, eight months the 1934 Constitution, seven months the 1946 Constitution – and only 45 days the constitution of 1967 imposed by the military). The constitution of 1988 was a long, detailed charter with 245 articles but the political system itself underwent few changes. The main features of the 1946 Constitution were maintained. New features included the election of the president, governors and mayors of cities with more than 200,000 voters in two rounds if necessary to achieve a majority of the valid vote; direct election of the governor of the Federal District; and the lowering of the voting age from 18 to 16, thus extending political rights to 16-and 17-year-olds.

On 15 November 1989, the first direct presidential elections for 30 years were the first ever based upon universal suffrage. They were held symbolically on the centenary of the Republic. The electorate now numbered 82 million (in a population of almost 150 million). With voting mandatory for those over 18 and under 70, the turnout was extremely high (over 80 per cent); 72.3 million voted, of whom 70 per cent were voting for a president for the first time. There were 21 candidates from across the political spectrum, from far Right to far Left, contesting the first round. The election was not won, however, as might have been expected, by the PMDB, the main opposition party for over 20 years and by far the biggest and broadest party in Brazil; nor by the PDT, the party of Leonel Brizola, the heir to Getúlio Vargas and João Goulart; nor by the PT, although its leader, Luiz Inácio ‘Lula’ da Silva did reach the second round (narrowly ahead of Brizola). It was won by Fernando Collor de Mello – a virtually unknown politician from the poor north-east state of Alagoas with only the relatively insignificant, recently created Partido da Renovação Nacional (PRN) behind him. He proved attractive to the dominant class, which, after the 21-year military dictatorship, had no credible candidate of its own; to some sections of the middle class and intelligentsia; and, above all, to the poorest and least educated sectors of Brazilian society who were susceptible to his populist appeal.²⁰ (See Essay 6.)

Brazil’s new democracy showed early signs of fragility against a background of severe economic recession, and from September to December 1992 Brazilians suffered the trauma of the impeachment (on corruption charges) of their first democratically elected president less than halfway through his term of office. In the end, however, the successful impeachment of Collor can perhaps be seen to have demonstrated more the maturity than the fragility of Brazilian democracy.²¹ For the first time in the history of the republic a president was removed from office – and replaced by the elected vice-president, Itamar Franco – by legal, constitutional means. Furthermore, the political crisis surrounding the impeachment of Collor was the first in which the military was not an active participant.

In 1994 and 1998 eleições casadas [presidential, gubernatorial, congressional, and state assembly elections] were held on the same day for the first time since 1950 and 95 per cent of the population of voting age was now registered to vote. In 1994 77.9 million went to the polls and 83.3 million in 1998. Both presidential elections were won outright in the first round by Fernando Henrique Cardoso, a distinguished sociologist with an international reputation and a politician with impeccable democratic credentials and advanced social democratic ideas, the candidate of the small Center-Left Partido da Social Democracia Brasileira (PSDB), which had split from the PMDB in 1988, though on each occasion backed by the parties of the Center-Right, especially the PFL. Cardoso was only the third elected president in 70 years to serve a full term (the first since Juscelino Kubitschek, 1956–61) and, as a result of a highly contentious constitutional amendment in 1997, the first in the history of the republic to be re-elected for a second term. The defeated candidate in both elections, as in 1989, was Luiz Inácio ‘Lula’ da Silva, the candidate of the PT.

The international environment in the 1990s was uniquely favourable to the survival and consolidation of democracy in Latin America. In particular, the United States made support for democracy a central feature of its policy towards the region, as it had done in the past but this time with rather better results. Furthermore, with the end of the Cold War anti-communism was no longer available as the main justification for the overthrow of democratic (or semi-democratic) governments as it had been in Brazil in 1964. The traditional political class (rural and urban), the more powerful economic interest groups and not least the military were, it seemed, now committed to peaceful democratic politics, as they had not always been in the past. Of course, it could be argued that the ‘propertied classes’ (including broad sections of the middle class) were no more than fair weather democrats. When the costs of overthrowing democracy were high and the costs of tolerating democracy low, democracy was likely to survive. But when their interests were threatened by forces favouring a significant distribution of wealth and power, as they were, or were believed to be, in 1964, there was always a possibility that they would look to the military to overthrow democracy. We shall never know whether Brazil’s new democracy would have passed its supreme test – the acceptance of victory by Lula and the PT in the presidential elections of 1989 or 1994. As Adam Przeworski once remarked, only where the Left lost the first elections following a process of democratisation was democracy truly safe.

It was a mark of the growing maturity and stability of Brazil’s democracy that the election of Luiz Inácio Lula da Silva to the presidency at the fourth attempt in October 2002 raised not the slightest doubt that he would assume power in January 2003. Lula could not quite achieve what Cardoso had achieved in 1994 and 1998: outright victory in the first round. In the second round, however, he comfortably defeated José Serra, Cardoso’s Minister of Planning and Minister of Health, the candidate of the PSDB. The election and transfer of power to a candidate of the Left (although the PT had abandoned the label ‘socialist’ before the election) represented an important landmark in the consolidation of democracy in Brazil.

After narrowly avoiding impeachment over the mensalão [big monthly payments to deputies in return for support] corruption scandal (2004–5), Lula recovered and, albeit with electoral support very different from that in 2002, went on to win re-election in October 2006 with a second round victory over Geraldo Alckmin, governor of São Paulo, the candidate of the PSDB. With his popularity at an all-time high (70–75 per cent approval), there was considerable discussion throughout 2009 about whether Lula would introduce an amendment to the 1988 Constitution which would allow him to run for a third term in 2010. In a number of subtle, and not so subtle, ways he indicated that he might indeed run. In the end, however, he declared ‘Eu não brinco com a democracia’ [‘I don’t play around with democracy’]. Instead he actively promoted his personally chosen successor, Dilma Rousseff (see Essay 6). (More than one commentator compared Lula’s imposition of Dilma to the famous dedazo (big finger), the way in which Mexican presidents nominated their successors during the period of PRI domination). Dilma won the 2010 presidential election, defeating José Serra (PSDB), the former mayor of São Paulo and governor of the state of São Paulo (and defeated candidate in 2002) in the second round to become Brazil’s first female president. In 2014 Dilma Rousseff was re-elected, though this time only narrowly defeating the former governor of Minas Gerais, Aécio Neves (PSDB) in the second round. It was the seventh democratic presidential election since the transition from military to civilian rule in 1985 and the sixth time since 1994 that the election had come down in the end to a contest between the candidates of the PSDB and the PT.

V

In 2015, 30 years after the end of the military dictatorship, Brazil could unquestionably be counted a fully fledged, consolidated and apparently stable democracy – with an electorate of 140 million by far the largest democracy in Latin America and the fourth largest in the world after India, the United States and Indonesia. An electoral calendar had been established with regular, free and fair elections for both executive and legislative branches of government, at federal, state and municipal levels. (In the 29 years between the municipal elections of 1985 and the presidential and congressional elections of 2014 Brazilians went to the polls 17 times.) For the first time in the history of the republic Brazil had a political system based on one person, one vote – and a voting age of 16. Elections were highly competitive, contested as they were by a large number of political parties, from Left to Right. The conduct of elections had been improved by the introduction of electronic ballot boxes in all elections in 1996. Brazilian democracy had survived its one institutional crisis – the impeachment of Collor de Mello in 1992 – and passed its ultimate test: the acceptance by the military (and the more reactionary elements in the political and business elite) of the victory of the candidate of the PT in the presidential election of 2002 (repeated in 2006, 2010 and 2014).

There is, of course, more to democracy than elections, however honestly conducted and freely contested and whatever the level and strength of popular participation: the separation of powers, an independent judiciary, the rule of law, a free press, the protection of civil liberties, etc. In 2015 Brazil scored relatively well here and there remained no ‘authoritarian enclaves’, that is to say, remnants of the power apparatus of the military dictatorship not accountable to democratically elected civilian governments. The military itself had steadfastly remained out of politics. Finally, Brazil is a country with remarkably few of the regional, national, racial, ethnic, linguistic, religious divisions, tensions and conflicts that pose a threat to democracies, old and new, throughout most of the world. In this respect it is uniquely fortunate.

And yet Brazilian democracy, like all democracies, is not without its flaws and challenges.²² Some political scientists would go so far as to claim that in Brazil, as in the rest of Latin America, the presidential system itself is a major obstacle to the proper functioning of representative democracy. However poor their performance, however weak their support in Congress, however low their standing in the country, presidents can only be removed in advance of the next scheduled elections by extreme measures: for example, in the case of Brazil, suicide (Vargas, 1954), resignation (Quadros, 1961), military coup (Goulart, 1964) or impeachment by Congress (Collor, 1992). Brazil had two opportunities to change its system of government during the process of democratisation: in March 1988, after prolonged debate on the issue, the Constituent Assembly voted 344 to 212 in favour of a presidential rather than a parliamentary system; and five years later (April 1993), in the plebiscite required under the 1988 Constitution, 55 per cent of the electorate voted for presidentialism, 25 per cent for a parliamentary system of government, with 20 per cent of the vote spoiled or blank. (In the same plebiscite Brazilians were also offered the opportunity to restore the monarchy: 12 per cent voted in favour compared with 66 per cent who supported the republic.)

But it is Brazil’s electoral system that has come in for the greatest criticism. The most undemocratic feature of Brazilian democracy – and the most difficult to reform – is the distribution of seats in Congress under the federal system. Since the beginning of the republic the less populated, less developed, more politically conservative (that is to say, clientelistic) states, especially in the north and north-east, have been over-represented in Congress, and since 1934 the more populated, more developed states, especially São Paulo, under represented. All 26 of Brazil’s states (and the Federal District) have an equal number of seats in the Senate (three). However, representation in the Chamber of Deputies has also not been proportional to population or electorate (one citizen, one vote, with more or less equal weight), as we have seen. The 1988 Constitution fixed a maximum of 60 (raised to 70 in 1994) and a minimum of eight seats per state. This distortion was aggravated by the creation of three new states with small populations: Tocantins, separated from the state of Goiás, Roraima and Acre, upgraded from their status as territories. Moreover, the Federal District (Brasília), also with a small population, was given representation in Congress for first time. No provision was made for periodic revision to account for demographic changes between states.

Thus in 2014, São Paulo (population 43.6 million, electorate 22 million) had 70 seats (the maximum); Minas Gerais (20.5 million), 53 seats; Rio de Janeiro (16.9 million), 46 seats; Bahia (14.5 million), 39 seats, etc. But seven states and the Federal District with populations of less than 3 million, including Roraima with only 500,000, Amapá and Acre each with 800,000, all had eight seats. If seats were distributed proportionally to population or electorate São Paulo would have 111 seats, Roraima only one. São Paulo has one seat per 622,000 people, Roraima one per 66,000. Thus, a vote in Roraima is worth nine times a vote in São Paulo.²³ Brazil’s seven smallest states (by population, not size), which together account for only five per cent of Brazil’s population, elect 25 per cent of the Senate and over 10 per cent of the Chamber.

There can be no democracy without elections, and in a modern mass democracy no elections without parties. Brazilian parties, however, not only lack historical roots – most were formed after 1985 – and, except for the PT, ideological or programmatic identity, but there are an unusually large number of them. It is generally agreed that Brazil has the most ‘underdeveloped’, fragmented party systems in the world.²⁴ There have been no constitutional barriers to the registration of new parties. In the nine elections between 1982 and 1996 76 parties put up candidates, though 39 of them only once. Five parties contested the elections in 1982, the last during the military regime after the two party system had been scrapped, 18 in 1998, 35 in 2014. Admittedly, many were very small: in 2014 13 parties received less than 1 per cent of the vote (together only 6 per cent). But even the number of what political scientists classify as ‘effective parties’ grew from 7.7-8.2 in the years 1995–2002 to 10.5– 11.4 in the years 2007–10 and 13.4 in 2015. Brazil has more parties than any other democracy. In a study of 1,167 elections in 137 countries between 1919 and 2015 by Michael Gallagher, three of the four elections contested by most parties were held in Brazil (2014, 2010, 2006), the other in Poland (1991).²⁵ Moreover, at each election a bewildering number of party coalitions are formed, and different coalitions for federal, state and municipal election. For example, in 2014 the 46 seats in the state of Rio de Janeiro were contested by 865 candidates belonging to 32 different parties, 23 of them in eight coalitions.²⁶

In elections for the Chamber of Deputies (and the Senate) the electoral district, the constituency, in Brazil is the state. Three states have more than 10 million voters (São Paulo 22 million, electing 70 deputies), another eight have over five million. Federal deputies (and senators) thus have minimal identification with, and accountability to, the voters who elected them. Since 1945, and maintained in two Constituent Assemblies (1946 and 1987–8), in elections for the Chamber of Deputies (and state assemblies) Brazil has had a system of proportional representation, together with ‘open’ lists of candidates for election. Since party membership in Brazil is low, and only a minority of voters even have a clear preference for a particular party (half of them for the PT), most Brazilian vote for individuals rather than parties. In the larger states, as we have seen, they have to choose between hundreds of candidates. In 2014 two weeks before the election whereas 93 per cent of voters had decided which candidate they would support for president, only 30 per cent had decided their candidate for federal deputy. Forty-five days after the election less than half could remember the name of the candidate they had voted for!²⁷

Since voting is obligatory for voters over 18 and under 70, in recent elections between 10 and 20 per cent of the electorate have voted either em branco [blank ballot] or nulo [spoiled ballot] – practices common (and understandable) during the period of military rule but disturbing in a democracy and extraordinarily high by the standards of any democracy in the world. In the election for the Chamber in 2014 more than 15 per cent of the electorate voted em branco (10 million) and nulo (7.5 million). In the state of Rio de Janeiro the figure was more than 21 per cent.²⁸ A nulo was always a wasted vote, and a vote em branco also after 1997, because they are not counted as votos validos [valid votes] for candidate or party in the distribution of seats under proportional representation.

Other problems affecting the freedom and fairness of elections in Brazil – the influence of the media, especially television, and the financing of election campaigns – are common to all modern democracies. However, even when compared with elections in richer countries (the United States, for example), Brazilian elections are exorbitantly expensive. And illicit, unregistered funding by private companies and individuals to both political parties and politicians [known as caixa dois] is a generalised phenomenon, as the corruption scandals that have dominated Brazilian politics since March 2014 have demonstrated (see below).

Brazil’s electoral system also creates major problems for democratic governability. The largest party in Congress has never had more than around 20 per cent of the seats in Congress, and the largest party was not always the party of the president: for example, the PMDB after the 1990 election when President Collor’s PRN had only 7 per cent of the seats, the PFL after 1998 when President Cardoso’s PSDB had 15 per cent, the PMDB after 2014 when President Dilma Rousseff’s PT had 13 per cent. The result is what has been called ‘permanent minority presidentialism’ which in practice becomes presidencialismo de coalisão [coalition presidentialism]. The president, elected by national majority vote, in two rounds if necessary, is obliged to conduct intense negotiations with party leaders, with individual politicians and even with state governors, who have a measure of influence over the deputies (of all parties) elected in their states, before and especially after elections, in order to guarantee the Executive a solid, sustainable base in the Legislature. In the 2011–14 Congress, 23 parties had seats in the Chamber of Deputies (15 in the Senate), and in the Chamber elected in 2014 there were 28 parties – more than in any other democracy. Italy, for example, had 15 after the election in 2013, Belgium 13 in 2013, Israel 10 in 2015. Fernando Henrique Cardoso built a coalition of six or eight parties, Lula eight or nine, Dilma 13. This inevitably produces extreme forms of fisiologismo [pork-barrel politics] and corruption. Many of the medium-sized and smaller parties simply become partidos de aluguel [parties for rent]. When the mensalão scandal came to light in 2004–5, individual deputies were discovered to have been bought. Positions in government have to be created to accommodate all parties in a coalition: for example, at the start of Dilma’s second administration Brazil had 39 ministries.

To complicate the situation of the Executive even more, Brazilian parties notoriously lack not only ideological and programmatic consistency (except for the PT) but also cohesion and discipline: party-switching is common. In the seven legislatures elected since the end of military rule (1986–2010) no less than 27 per cent of federal deputies abandoned the party for which they were elected during their mandates.²⁹ And this underestimates the extent of party switching because some deputies switched more than once – in some cases several times! – during the same legislature. Only a small proportion of these changes were the result of the creation of new parties (for example, 27 deputies joined the PSDB in 1986, 48 the PSD in 2011). Party switching was initially considered a question of the natural re-alignment of the political class after 21 years of military dictatorship. It was expected to diminish as democracy was consolidated. But this clearly did not happen and 18 months into the legislature elected in 2014 97 deputies (19 per cent) had switched parties. Among democracies throughout the world, only Italy has a similar level of party infidelity.

Political reform has been an issue of heated debate within the political class and in academic circles in Brazil since the mid 1990s. The issue of presidentialism versus parliamentarism was no longer on the agenda, it seemed. The distribution of seats in the Chamber of Deputies between states was rarely mentioned. An end to re-election for executive office – president, governor and mayor – came up for discussion from time to time, as did an end to obligatory voting. Campaign finance was a constant theme. But the focus was on simplifying for the voter the system for the election of deputies (federal and state), making deputies more accountable to those who elected them (and incidentally improving their quality), and reducing the number of political parties. Reforms proposed have included, for example: dividing states into smaller, single member constituencies with majority voting; adopting ‘closed’ party lists of candidates, if state-wide elections based on proportional representation continue to be preferred; banning party coalitions in elections based on proportional representation; tightening the rules introduced in 2007 for the removal from Congress of deputies who switch parties during the legislature to which they were elected; above all, erecting constitutional barriers to the registration of parties³⁰ and access to the Fundo Partidário and to media time during elections.

There were several congressional commissions on political reform, and many reform proposals put to the Chamber of Deputies – most of them defeated. There was no consensus on political reform in Brazil – between parties or within parties (politicians rarely vote to change the system under which they were elected), and the general public, the electorate, showed little interest.

Beyond the debate about the fairness and effectiveness of Brazil’s electoral system there was the more fundamental concern that Brazil’s relatively new democracy had not been broadly or deeply legitimated. Brazilian public opinion polls and the Latinobarómetro research institute in Santiago de Chile have consistently provided evidence not only of a widespread ignorance of political issues in Brazil³¹ but also, more disturbingly, a lack of trust in politicians, political parties and political institutions – and therefore a lack of confidence in democracy itself.³² Some would argue that democracy is merely ‘formal’, rather than ‘substantive’, if it does not protect the economic and social ‘rights’ of citizens. Despite the economic and social advances made during the previous two decades, especially during the PT administrations (2003–10) (see Essay 7) Brazilians also perceived democracy as having failed to promote a significant reduction of poverty, inequality and social exclusion (which despite Brazil’s claim to be a racial democracy have a clear racial dimension).

Nevertheless, in 2015, despite their many flaws, and the evident low esteem in which they were held by, if not a majority, a substantial minority of the Brazilian electorate, Brazil’s democratic institutions appeared to be fully consolidated and fundamentally stable. Brazilian democracy, however, was about to face its greatest test in the political and institutional crisis arising from the controversial impeachment of a democratically elected and (albeit narrowly) democratically re-elected president, Dilma Rousseff of the PT, and her replacement by vice-president Michel Temer of the PMDB.

Epilogue

Within six months of the start of President Dilma Rousseff’s second mandate in January 2015 there was speculation that she might be impeached and impeachment was the subject of intense debate throughout the rest of the year. The background to this was, first, a petition to the Supreme Electoral Tribunal for the annulment of Dilma’s narrow victory in the 2014 presidential election on the grounds that she had seriously misinformed the country on the state of the Brazilian economy and that her electoral campaign had been financed illegally. Secondly, one of the deepest and most persistent economic recessions in Brazilian history. The rate of economic growth had been in steady decline since 2011. The economy grew only 0.4 per cent in 2014. In 2015 it contracted by 3.8 per cent (and a further 3.6 per cent in 2016). Inflation rose, living standards fell and unemployment increased (see Essay 7). Thirdly, a corruption scandal without precedent in Brazil, indeed in any modern democracy. Corruption, most recently concerning the preparations for the World Cup in 2014 and the Olympic Games in 2016, was already an issue in Brazilian politics. Beginning in March 2014 and gathering strength throughout 2015 the so-called Lava-Jato [Car Wash] investigation revealed a major scandal involving the government-appointed directors of Petrobras (the state-owned oil company), a cartel of private construction companies (of which Odebrecht was by far the largest) and government ministers and politicians belonging mainly to the PT, but also its principal allies in government, the PMDB and the PP.³³ By October Dilma’s approval rate had collapsed to single figures, the lowest for any president since the end of military rule in 1985. An Ibope poll in November showed that 87 per cent of those interviewed found the government ruim or péssimo [bad or very bad], and 67 per cent declared themselves in favour of the impeachment of the president.

President Dilma had begun her second term with the support of 69 per cent of the Chamber of Deputies and 72 per cent of the Senate, although only 16 per cent of deputies and 17 per cent of senators belonged to the PT. Her government was a coalition of 12 parties (eight of which had ministerial posts) in which the PT was in a minority. With no end in sight either to the economic recession (and its social impact) or Operation Lava-Jato – and in view of her own limited experience in dealing with Congress – Dilma found it increasingly difficult to maintain and manage her multi-party coalition government. In October the PMDB, the largest party in Congress, withdrew its support, making effective government virtually impossible.

In December the president of the Chamber of Deputies, Eduardo Cunha (PMDB), a political enemy of the president, and himself under investigation for corruption, accepted a petition for the impeachment of Dilma from three prominent lawyers in São Paulo. The grounds for impeachment were not, as in the case of Collor de Mello in 1992, personal corruption, nor any involvement the president might have had in the various corruption scandals involving Petrobras. She was accused of crimes of fiscal and administrative responsibility in violation of the constitution and the law: the improper manipulation of the public accounts in 2014 by transferring funds from the state banks to cover up a huge fiscal deficit and mounting public debt (what became known as pedeladas fiscais).

After securing the blessing of the Supreme Court (nine of whose 11 judges had been appointed by PT presidents), the petition for impeachment went in March to a commission of the Chamber of Deputies where on 11 April it was approved by majority vote (38 votes to 27), and then to the full Chamber where on 17 April it was approved by the necessary two thirds of the deputies (367 votes to 137). On 12 May the Senate voted (55 votes to 22, with four senators absent) to suspend Dilma from office pending a full trial. Vice-president Michel Temer (PMDB), Dilma’s running mate in both 2010 and 2014, became interim president. On 31 August the Senate voted in favour of the impeachment of Dilma by more than the required two thirds majority (60 votes to 21) and Temer was confirmed as president of Brazil. The impeachment process spread over almost nine months was thus entirely constitutional and legal.

However, the impeachment of President Dilma Rousseff was also, perhaps like most impeachments primarily, political: the removal of a failed president who had lost the support of both the Congress and the people (‘the street’). It was fiercely contested by the PT which, together with its allies, the smaller parties on the Left, had the support of 25–30 per cent of Congress, and which could mobilise not insignificant support from party members, the unions, especially the public sector unions, and important sectors of civil society, not least university teachers and students. For the PT the impeachment of Dilma was a conspiracy by ‘elite’ politicians of the Centre-Right, especially the PMDB, many of them themselves deeply corrupt, big business (domestic and foreign), the judiciary and the media to bring the political hegemony of the PT to an end, and in particular to prevent the election of Luiz Inácio Lula da Silva as president once again in 2018. It was a golpe, a coup, albeit a parliamentary coup, and thus a frontal assault on Brazilian democracy. Militant petistas were (and still are) for the most part in denial about the corruption associated with the PT governments and the economic mismanagement and political incompetence of the Dilma administration which brought Brazil to the sorry state it was in at the end of 2015 and the beginning of 2016.

The Temer government, which the Opposition insisted was illegitimate, was indeed a government of the Centre-Right, full of politicians (including President Temer himself) under investigation for receiving bribes and illegal campaign contributions, out of touch with the majority of Brazilians and equally as unpopular as its predecessor. It survived a turbulent first twelve months, despite the loss of several ministers charged with corruption. And it began to appear that it might at least muddle through to the end of Dilma’s original mandate (December 2018), if only because the alternative would be political chaos. However, in April 2017 the publication of the ‘end of the world’ list of all those accused of corruption by the directors of Odebrecht appeared implicating eight ministers, 24 senators, 39 deputies, the presidents of both houses of Congress and three state governors – and indirectly Lula, who was already being investigated under Operation Lava-Jato, Dilma and Temer. A month later, on 17 May, came the JBS bombshell. In plea bargaining testimony the owners of Brazil’s (and one of the world’s) largest food producing companies, implicated President Temer himself, Aécio Neves, the national president of the PSDB (and defeated candidate in the 2014 presidential election), and hundreds of politicians of all parties in the biggest corruption scandal thus far, which threatened to bring down the Temer government.

Temer refused to resign. In June, in spite of overwhelming evidence of illicit campaign funding, the Supreme Electoral Tribunal (more than two and a half years after the event) finally voted, by four votes to three, not to annul the 2014 presidential election, which would have led to the immediate removal of Temer from office. The decision was made in the interests of political stability and passage of the reforms necessary for the urgently needed fiscal adjustment and the restoration of economic growth. (Technically, Brazil came out of recession in the second quarter of 2017. And with the solid majority in Congress that had voted to impeach president Dilma, the government appeared to be making modest progress on labour reform and on reform of the pension system, which previous governments, both PSDB and PT, had ignored and which was the main reason for Brazil’s fiscal ‘emergency’.) Temer now faced a series of criminal charges for corruption, money laundering and obstruction of justice to be brought before the Federal Supreme Court. However, this required the approval of two-thirds of the Chamber of Deputies. Temer survived a first vote on 2 August, and a second on 25 October, with 263 deputies (of 513) and 251 deputies respectively voting against sending the case to the STF. There was unlikely to be a third, although further charges continued to be made against him. Temer’s continuation in office until 1 January 2019, once no means certain, seemed secure.

Elections – for president, two-thirds of the Senate, the Chamber of Deputies, state governors and state assemblies – were due to be held in October 2018. The PT, the only significant party of the Left, was now discredited with broad sectors of the electorate, as the results of the municipal elections in October 2016 had demonstrated (see Essay 7). And given the levels of poverty and inequality in Brazilian society Brazilian democracy needs a party of the Left. Also discredited was the main party of the Centre-Right, the PMDB, as were to a greater or lesser degree all the parties in the Temer coalition government, including the PSDB.³⁴ None of the so-called pre-candidates for president had a convincing plan for dealing with Brazil’s economic, social, political – and ethical – crisis. More than 60 per cent of Brazilians polled wanted a president from outside the three main parties (PMDB, PT and PSDB), but no credible ‘outsider’ had emerged. Former president Luiz Inácio Lula da Silva, the pre-candidate of the PT, remained by far the most popular politician in Brazil and early opinion polls put him well ahead of the field, with 30–35 per cent of the vote in the first round, despite the fact that he had been sentenced to several years in prison for money laundering and active and passive corruption.³⁵ If in the end, as expected, Lula were to be declared ineligible to run for president,³⁶ his supporters, who argue that his prosecution was entirely political, an extension of the golpe which had brought down Dilma and precisely aimed at preventing him from being re-elected president, would inevitably challenge the democratic legitimacy of whoever was elected. In any event, in the absence of any profound change in the electoral system, and especially the party system, despite the fact that, following Dilma’s impeachment, political reform had continued to be debated as a matter of urgency, both inside and outside Congress,³⁷ the new president in January 2019 would face the same problems of democratic governability as his or her immediate predecessors. There were no grounds for complacency about the future stability of liberal representative democracy in Brazil.

* This is a new essay based in parts on ‘Politics in Brazil under Vargas 1930–1945’, ‘Politics in Brazil under the Liberal Republic 1945–1964’, (with Celso Castro) ‘Politics in Brazil under military rule 1964–1985’ and (with J. Nicolau), ‘Politics in Brazil since 1985’, in Cambridge History of Latin America Vol. IX Brazil since 1930 (Cambridge: Cambridge University Press, 2008) and an inaugural lecture ‘The long road to democracy in Brazil – and present concerns’, delivered in the Faculty of Social Science and Public Policy, King’s College London, 14 March 2017. It includes ideas first formulated in ‘Politics in Brazil: from elections without democracy to democracy without citizenship’, Daedalus, Journal of the American Academy of Arts and Sciences, Special issue, Brazil: burden of the past, promise of the future, 129 (2000); revised version in J. Dunkerley and M. D’Alva Kinzo (eds.), Brazil since 1985: Economy, Polity and Society (London: Institute of Latin American Studies, 2003).

1 R. Graham, Patronage and Politics in Nineteenth Century Brazil (Stanford: Stanford University Press, 1990), p. 109, Table 2, and p. 332, n. 41; J. Nicolau, Eleições no Brasil. Do Império aos dias atuais (Rio de Janeiro: Zahar, 2012), p. 30. On the political system of the empire, see also J.M. de Carvalho, A construção da ordem: a elite política imperial; Teatro de sombras: a política imperial (2nd rev. edn., Rio de Janeiro: Editora UFRJ/Relume Dumará, 1996).

2 Graham, Patronage and Politics, pp. 185–6, 200, 202.

3 In five of Brazil’s 20 states (former provinces) during the First Republic, including São Paulo and Rio Grande do Sul, the chief executive was called president.

4 See B. Lamounier and J. Muszynski, ‘Brasil’, in Dieter Nohlen (ed.), Enciclopedia electoral latinoamericana y del Caribe (San José, 1993), pp. 93–134, especially Table 2.1 ‘Evolucion del electorado 1933–1990 [in fact 1894–1990]’ and Table 2.9 ‘ Elecciones presidenciales 1894–1989’ (pp. 125–30) for statistical information on elections in Brazil after 1889. See also J. Nicolau, Eleições no Brasil, an indispensible guide.

5 J.M. de Carvalho, Os bestializados: o Rio de Janeiro e a republica que não foi (São Paulo: Companhia das Letras, 1987), ch. 3 ‘Cidadãos inativos: a abstenção eleitoral’; Nicolau, Eleições no Brasil, p. 60.

6 Lamounier and Muszynski, ‘Brasil’, in Nohlen (ed.), Enciclopedia electoral, pp. 99, 128.

7 The best book on the revolution of 1930 remains B. Fausto, A revolução de 1930. História e historiografia (1977; 16th, rev. edn, São Paulo, 1997).

8 J. Hahner, Emancipating the Female Sex: the Struggle for Women’s Rights in Brazil, 1850–1940 (Durham, NC: Duke University Press, 1991), pp. 171–3. The suffrage had first been extended to women in New Zealand in 1893, followed by Australia in 1902, some West European countries including Germany at the end of the First World War, though not until 1928 in the United Kingdom. In the western hemisphere women won the right to vote in Canada in 1918 and the United States in 1920, but among Latin American countries only in Ecuador in 1929 before Brazil. It was 1944 before women were given the vote in France, for example, 1946 in Italy, 1947 in Argentina, 1953 in Mexico and 1974 in Portugal.

9 Nicolau, Eleições, Gráfico 2, p. 79

10 On the background to the 1937 coup, see A. Camargo et al., O golpe silencioso. As orígens da república corporativa (Rio de Janeiro: Rio Fundo, 1989).

11 See, for example, T.E. Skidmore, Politics in Brazil, 1930–1964: An Experiment in Democracy (New York: Oxford University Press, 1967) for an early version of this claim.

12 Nohlen (ed.), Enciclopedia electoral, pp. 108, 113, 128.

13 Nicolau, Eleições, p. 95.

14 Nicolau, Eleições, Grafico 3, p. 97. The electorate actually declined between the congressional elections of 1954 and 1958, from 15.1 million to 13.8 million, while the population grew by 11 per cent, as a result of a new voting register in 1956, the first for over a decade, aimed at reducing the number of the deceased on the electoral lists, double registrations, changes of residence and so forth.

15 On the collapse of post war democracy in 1964, see W.G. dos Santos, Sessenta e quatro: Anatomia da crise (São Paulo, 1986); A.M. Cheibub Figueiredo, Democracia ou reformas? Alternativas democráticas a crise política, 1961–1964 (São Paulo, 1993); C. Navarro de Toledo, ‘1964: o golpe contra as reformas e a democracia’, in D. Aarão Reis, M. Ridenti and R. Patto Sá Motta (eds.), O golpe e a ditadura militar quarenta anos depois (1964–2004) (Bauru, SP; EDUSC, 2004); C. Fico, Além do Golpe. Versões e controvérsias sobre 1964 e a ditadura militar (Rio de Janeiro: Editora Record, 2004).

16 See D. Leonelli and D. de Oliveira, Diretas já. 15 meses que abalaram a ditadura (Rio de Janeiro: Editora Record, 2004).

17 See L. Martins, ‘The ‘liberalisation’ of authoritarian rule in Brazil’, in Guillermo O’Donnell, P.C. Schmitter and L. Whitehead (eds.), Transitions from authoritarian rule: Latin America (Baltimore, MD: Johns Hopkins University Press, 1986); T. Skidmore, ‘Brazil’s slow road to democratisation, 1974–85’, in A. Stepan (ed.), Democratising Brazil: Problems of Transition and Consolidation (New York, NY: Oxford University Press, 1989).

18 Brazil was the last country in Latin America to give illiterates the right to vote. Uruguay had done so in 1910, Venezuela in 1946, Bolivia in 1952, Chile in 1970, Ecuador in 1978 and Peru in 1980. In Europe Portugal was the last country to adopt universal suffrage – in 1974, following the overthrow of the dictatorship.

19 For an interesting analysis of the ‘black vote’ in the elections of 1985 and 1986, see E. Berquo and L. Felipe de Alencastro, ‘A emergência do voto negro’, Estudos CEBRAP, 33 (July, 1992).

20 In the first round Collor secured 30.5 per cent of the votos validos (i.e. excluding the blank and spoiled ballots), Lula 17.2 per cent and Brizola 16.5 per cent. In the second round Collor had 53 per cent, Lula 47 per cent. Nohlen (ed.), Enciclopedia electoral, pp. 99, 130.

21 President Fernando Collor de Mello was impeached first in the Chamber of Deputies on 29 September (441 votes to 38) and then, definitively, in the Senate on 29 December 1992 (76 votes to 3 with two abstentions), a few hours after he had in fact resigned.

22 In the Economist Intelligence Unit’s ‘Democracy Index 2016’, 167 countries are divided into full democracies, flawed democracies, hybrid regimes and authoritarian regimes. In Latin America only Uruguay is a full democracy. Brazil (at no. 51), behind Uruguay (19), Chile (34) and Argentina (49), is a flawed democracy with overall score of 6.9. Brazil scores well on electoral process and pluralism (9.55), civil liberties (8.82), less well on the functioning of government (6.79) and political participation (5.56), and poor on political culture (3.75).

23 J. Nicolau, Representantes de quem? Os (des)caminhos do seu voto da urna à Câmara dos Deputados (Rio de Janeiro: Zahar, 2017), pp. 100–4.

24 On the Brazilian party system, see in particular the work of S.P. Mainwaring: ‘Brazilian party underdevelopment’, Political Science Quarterly, 107 (1992); ‘Brazil: weak parties, feckless democracy’, in S.P. Mainwaring and T.R. Scully (eds.), Building Democratic Institutions: Parties and Party Systems in Latin America (Stanford: Stanford University Press, 1993); and Rethinking Party Systems in the Third Wave of Democratisation: the Case of Brazil (Stanford: Stanford University Press, 1999).

25 Cited in Nicolau, Representantes, p. 90.

26 Ibid, p. 37.

27 Ibid, p. 62.

28 Ibid, p. 25.

29 Ibid, p. 80.

30 A threshold of two per cent of the national vote in the previous election for a party to register for the next would potentially reduce the number of Brazilian political parties to 13, a threshold of 3 per cent to 9 parties, a threshold of 5 per cent to 7 parties.

31 The Brazilian electorate is not only overwhelmingly poor but also illiterate, semi-literate or at best poorly educated. ‘We must educate our masters’, famously declared Robert Lowe in the House of Commons on the passage of the Reform Act of 1867. (What he actually said was, ‘I believe it will be necessary that you should prevail on our future masters to learn their letters’.) Almost a century later Anísio Teixeira, one of Brazil’s greatest educators, declared, ‘There will only be democracy in Brazil the day the machine (máquina) that prepares people for democracy – the public school – is assembled in Brazil’. Public education in Brazil, both primary and secondary, remains woefully inadequate.

32 In 2016 the Latinobarometro found that only 32 per cent of Brazilians identified democracy as the ‘preferable’ form of government, a dramatic fall from the already low figure (54 per cent) of the previous year. Brazil was next to bottom in Latin America, above only Guatemala.

33 By the beginning of 2018 more than 100 prominent politicians and businessmen who had previously regarded themselves as above the law and had enjoyed an extraordinary level of impunity had been arrested and condemned to long prison sentences under Operation Lava-Jato. And many more were under investigation. Brazil now had ‘accountability institutions’ – the Federal Audit Tribunal (TCU), the Office of the Federal Public Prosecutor, the federal police and, above all, the federal judiciary – equal to those of any democracy in the world. Of course, this brought with it a potential new danger to democracy: the so-called judicialisation of politics, the increasingly important role played in politics by unelected judges.

34 A DataFolha poll in July 2017 showed only two per cent of those interviewed had complete confidence in Brazil’s political parties, 28 per cent a little and 69 per cent none. The October 2016 municipal elections had produced the greatest percentage of spoiled and blank votes since the introduction of electronic voting in 1996. In Rio de Janeiro, for example, abstentions, nulos and em brancos totalled 42.5 per cent of the electorate. Political scientists were forecasting a tsunami of protest votes in 2018.

35 On 12 July 2017 in Curitiba Judge Sérgio Moro, who was responsible for cases brought to justice under Operation Lava-Jato, sentenced Lula to nine years and four months in prison. On 24 January 2018 the regional court of appeal in Porto Alegre upheld the original conviction and increased the sentence to 12 years and one month. On 7 April, after the failure of a last-ditch attempt to persuade the Federal Supreme Court to grant him habeas corpus and despite ongoing attempts to persuade the Court to reverse its decision, Lula began to serve his prison sentence, initially in the federal police headquarters in Curitiba.

36 The Lei de Ficha Limpa (Clean Slate) of 2010 prohibits candidates with certain criminal records from running for public office for eight years.

37 The only reforms enacted were an end to corporate financing of election campaigns and the imposition of a modest threshold of 1.5 per cent of the national vote for a party to be able to claim financial support from an enhanced state-financed Party Fund and TV/radio time during elections. The latter, however, would only come into effect in 2020. Thirty-five parties were expected to contest the Congressional elections in October 2018.