In 2018, Jair Bolsonaro shocked the world by rising from relative obscurity to become president of Brazil. In these pages, I described a paradox thrown up by his unexpected success. Bolsonaro’s core support clearly lay in Brazil’s predominantly white middle and upper classes. But this is not sufficient to win a majoritarian election in a country where 70 percent earn less than two minimum wages (around US$450 per month) and more than half identify as black or mixed race. If we were to understand Bolsonaro’s victory, I argued, we also needed to understand his appeal to many low-income, black and brown Brazilians.
This seemed particularly paradoxical in the wake of thirteen years of government by the center-left Workers’ Party (PT) (2003–2016), during which it oversaw impressive levels of poverty reduction. While lower-income voters had tended to support the PT during this period, the middle classes fiercely opposed them. Bolsonaro’s victory seemed to defy such established logic. I therefore posed the following questions: “How has Bolsonaro been able to bring together elites wishing to block the social mobility of the popular classes, and a significant proportion of those they seek to block, within the same electoral coalition? And how long can this last?”
After four years of the Bolsonaro government, and in the heat of another presidential election, it is an apt moment to revisit these questions. Certainly, we now have a clear answer to the second: Bolsonaro’s mass appeal is far more durable than many assumed. In last Sunday’s vote, despite having overseen the world’s worst pandemic response, a huge increase in deforestation rates, and launched constant attacks against democratic institutions, Bolsonaro has fallen only slightly short of his previous first-round vote (43 percent compared to 46 percent in 2018). He will now face a tense runoff against the PT candidate, former president Luiz Inácio Lula da Silva, who gained 48 percent of first-round votes.
Even if Bolsonaro loses, it is clear that Bolsonarismo — his ideological agenda, presentational style, and the political forces grouped around him — will continue to shape Brazil in the years to come. His newly adopted party, the Partido Liberal (PL), has become the largest in both houses of congress, definitively displacing traditional center-right parties and presenting a formidable barrier to the reforming ambitions of a prospective Lula government.
But if Bolsonarismo has clearly become a resilient political force, what about the first question? Does Bolsonarismo still rest on a socially heterogeneous electoral coalition, and, if so, how has this coalition been held together since 2018? A comparison of first-round election-eve polling by Datafolha, Brazil’s leading polling company, for 2018 and 2022 offers some clues to how the social composition of Bolsonaro’s vote has evolved.
In 2018, Bolsonaro led by huge margins in the top two income bands: those earning more than ten times the minimum wage, and those earning between five and ten. However, he also had a very significant lead in the third band (two to five times the minimum wage) and only trailed narrowly in the lowest band (under two times the minimum). In the 2022 poll, his vote has fallen by similar margins — around 6-8 percentage points — in each of the three lower bands, and about twice as much for the wealthiest band.
By contrast, the PT vote has increased significantly — around 15-17 percentage points — in the lower three bands, and a huge 25 points among the wealthiest. Taken together, these figures suggest a uniform swing away from Bolsonaro among the lower three groups (albeit from very different baselines) and a much larger one among the highest earners.
On the other hand, thanks to the huge size of his advantage in 2018, Bolsonaro still enjoys narrow leads within the three upper bands and still retains the support of over 20 percent of the poorest voters. Broadly, then, we can say that Bolsonaro continues to exercise significant cross-class appeal today.
However, such analysis only takes us so far. Preelection polling this year appears to have significantly underestimated Bolsonaro’s vote. Meanwhile, these income bands all capture very large populations, each containing significant socioeconomic and political diversity. The very fact that each of these bands is, to some degree, split between the two candidates suggests that other factors must be taken into account if we are to identify the sources of Bolsonaro’s resilient cross-class appeal.
In this regard, the first-round results themselves offer greater insight. While they cannot capture individual-level demographic variables, they do reveal that changing geography of Bolsonaro’s vote. This geography reveals a key fracture that now cuts across the Brazilian electorate: between those who still support and those who now oppose the incomplete project of building a functioning, democratic, and at least moderately redistributive state in Brazil.
Regional Inequality and the State
Looking at Brazil’s five macro-regions — the Northeast, North, Central-West, Southeast, and South — the first round broadly reflected a continuation of recent presidential election results, albeit with some important shifts. Lula’s strongest performance came in the Northeast, Brazil’s poorest region and its second most populous. The Northeastern states have given the PT healthy majorities in all presidential elections since the 2000s, and continued to do so in 2014 and 2018, even as support for the party sharply declined in all other regions. From already high levels in 2018, support for the PT across the region has grown again this year.
Take the state of Piauí, for example, where 63 percent of first-round votes in 2018 went to the PT, and this has increased to a huge 74 percent this year. Even the Northeastern state that gave the PT its lowest vote, Alagoas, still recorded a higher percentage than any state outside of the region, and preferred Lula to Bolsonaro by a margin of more than twenty points (57 percent to 36 percent).
While it may seem intuitive for Brazil’s poorest region to vote for the country’s principal center-left party, this was not always the case. In the 1990s, the Northeast, like the rest of the country, overwhelmingly favored center-right candidates. Its more recent loyalty to the PT represents the consolidation of a strong regionwide identification not only with Lula (himself a Northeasterner) but with the party as a whole. It reflects the historical emergence of a political culture that recognizes the gains made under PT rule and which seeks to defend and renew that legacy.
Of course, the salience of partisan identification in Northeastern politics should not be overstated, and local clientelist dynamics persist, particularly in legislative elections. Nonetheless, the PT’s continued hegemony out of office belies the facile, widely repeated but unsubstantiated claim that the votes of the poor can be bought with small cash transfers paid by incumbent governments — a strategy Bolsonaro has attempted, but with little success to date. Northeastern voters have come to expect a more sincere and sustained commitment to building a redistributive social state.
However, this relationship between regional inequality and voting preferences does not hold up everywhere. The remote and sparsely populated Amazonian states of Acre, Rondônia, and Roraima — three of the poorest outside of the Northeast — are also the most pro-Bolsonaro states in the country, respectively giving him 63 percent, 64 percent, and 70 percent in last week’s vote. The states of the Central-West region, which occupy an intermediate socioeconomic position between the poorer Northeast and North regions and the wealthier Southeast and South, also gave Bolsonaro very large victories.
These states all lie on Brazil’s agricultural frontier, heavily dependent on livestock farming, soya production, and various extractive activities. While state capacity outside a few larger cities is generally precarious, local economies have actually benefited from its further erosion under Bolsonaro’s government. In particular, his dismantling of environmental and Indigenous protection agencies has removed major constraints to land- and resource-grabbing.
This may primarily benefit large landowners, but it also has knock-on benefits for the low-income populations of these areas. With few other economic opportunities available to them, many find paid employment or can operate as small-scale entrepreneurs in sectors with low barriers to entry.
Meanwhile, PT organization has largely failed to penetrate these areas, leaving most state resources that might actually reach the poor largely in the hands of clientelist parties. As a result, in the agro-frontier, the state is largely perceived as a threat to livelihoods, while any benefits the poor receive from the state tend not to be associated with progressive forces. Where the poor of the Northeast were absorbed into a social democratic project seeking to strengthen the democratic state as a redistributive and regulatory entity, the poor of the agro-frontier have been enlisted into a project that seeks to dismantle those very same institutions.
The Institutionalist and Revanchist Middle Classes
The middle classes also straddle Brazil’s political divide, and this division also strongly correlates to geography. The Southeast and South are Brazil’s wealthiest regions, but also contain diverse sub-territories, including rural agricultural zones, small towns and medium-sized cities with varying social conditions, and large metropolises that encompass wealthy districts, middle-class suburbs, and poor urban peripheries and favelas.
In 2018, voters across almost all of these diverse territories voted heavily for Bolsonaro, concealing important differences between them. However, in 2022, these differences have reemerged. The states of São Paulo and Rio de Janeiro — two of the country’s wealthiest and most populous — bear this out. In particular, a clear divergence has emerged between their respective urban and provincial middle classes.
To take two fairly representative examples, between 2018 and 2022, Bolsonaro’s vote in the provincial cities of Petrópolis (Rio de Janeiro) and São José dos Campos (São Paulo) fell from 62 percent to 55 percent and 60 percent to 55 percent, respectively. However, in both cases, he has retained a large lead of twenty-one points over Lula. In other words, while there may have been a downward adjustment from the peaks of 2018, Bolsonarismo has taken root in these places.
Compare that to the professional middle-class territories of metropolitan São Paulo and Rio de Janeiro. In the neighborhood of Pinheiros, in São Paulo’s inner West Zone, the PT vote increased by a huge thirty-six points, from 11 percent in 2018 to 47 percent today, while Bolsonaro’s has fallen from 42 percent to 33 percent. In Rio’s Botafogo, the PT vote grew by an even larger forty-one points — from 13 percent to 54 percent — while Bolsonaro’s has fallen from 44 percent to 33 percent. Such patterns were repeated across Rio’s South Zone and inner North Zone, and São Paulo’s inner North and Southwest Zones. In these areas, it seems, Bolsonarismo is really on the retreat.
Again, this divergence can be linked to these populations’ differing relationships to the democratic state-building process. In both cases, there is little interest in radical government or a strongly redistributive state, which most would associate with populism and corruption. The economic downturn and corruption scandals of the mid-2010s led both the urban and provincial middle classes into an anti-institutional turn — supporting a parliamentary coup against Dilma Rousseff and voting en masse for Bolsonaro in 2018 to “keep the PT out.”
However, the reality of Bolsonarismo has shown that part of the middle class continues to believe in the long-term project of democratic consolidation, including moderate social spending and the protection of minority rights. Many of them are the orphans of the much-weakened center-right Brazilian Social Democracy Party (PSDB), and, thanks to Lula’s moderate positioning, have been absorbed into the PT fold. More importantly, though, horrified by Bolsonaro’s pandemic denialism, dismantling of state institutions, and threats against democracy, they have become emphatically “anti-Bolsonarista.”
What has surprised many, including the polling companies, is how many voters have not accompanied the institutionalist middle class in recoiling from Bolsonarismo. Most prominent among them are those voters that I refer to as the revanchist middle class.
These voters are most commonly found in the small-town interior of the Southeast and South, but also in other regions and in some metropolitan areas. They support, if not outright authoritarianism, at least a radical reform of the state, rejecting the social settlement that emerged from Brazil’s re-democratization process and its progressive 1988 constitution.
For this constituency, investment in public services — most of which they do not use — is a waste of taxpayers’ money. Historically excluded racial groups and sexual minorities are seen to be pampered with state resources and privileges. Gun control and age limits for criminal liability are viewed as preventing police and ordinary citizens from legitimately defending themselves from ubiquitous criminals. In other words, the revanchist middle class seeks to take back a state that they believe was captured by their enemies, in order to neuter its capacity to continue work against their interests.
Seen in this light, what looks to others like Bolsonaro’s incapacity to govern, may look to many of these voters like a deliberate step in the right direction.
Disappointment vs. Indifference in the Urban Peripheries
If the rift between the institutionalist and revanchist middle classes is most clearly visible in comparisons between metropolis and hinterland, a divergence between low-income urban voters is clearest if we compare the county’s two largest cities: São Paulo and Rio de Janeiro.
During the 2000s, the poor urban peripheries of both cities had been electoral strongholds of the PT in presidential elections, much like the Northeastern region. However, they gradually abandoned the party during the 2010s and most voted strongly for Bolsonaro in 2018. This looked like a secular trend, perhaps attributable to a range of social processes common to these territories — for example, labor precaritization, a decline of trade unions and grassroots movements, and a large growth of Evangelical churches.
However, an historic divergence in last week’s election suggests we must look elsewhere for explanations. Again, a crucial factor is the different way the democratic state has been experienced in each of these cities.
In a reversal of recent electoral trends, the peripheries of São Paulo have swung back heavily toward the PT. Take, for example, the eastern periphery of Itaquera, where the PT vote increased from 20 percent in 2018 to 49 percent in 2022, or the southern periphery of Campo Limpo, where it grew from 25 percent to 54 percent. In both zones, Bolsonaro’s vote also fell markedly, leaving Lula with very large first-round majorities. This pattern is repeated across most of São Paulo’s municipal periphery and many of its working-class satellite cities.
The picture in the peripheries of Rio de Janeiro could hardly be more different. Across the distant West Zone neighborhoods of Santa Cruz, Campo Grande, and Bangu, Bolsonaro retained margins of twenty or more points over Lula, and he won convincingly across every part of the Baixada Fluminense. Bolsonarismo seems to be entrenched in Rio’s predominantly low-income metropolitan peripheries just as it has in the small-town hinterland.
There are important differences between the peripheries of the two cities that help to explain this divergence. First, the PT has historically achieved a far greater institutional presence in São Paulo’s peripheries. More broadly, though, institutional conditions in the city have, over time, fostered greater expectations of the state.
Eduardo Marques has identified what he calls a dynamic of “incremental progressivism” in São Paulo, observing that many progressive public policies have gradually been institutionalized, even under center-right municipal administrations. This is partly the product of electoral competition under conditions of minimal consensus, which means that effective policies often survive changes of government.
Just as importantly though, organized civil society has been able to achieve a significant influence over policy-making processes. Such processes are hardly smooth and the public sphere remains highly precarious across much of the periphery. But this dynamic provides evidence that contestation through the democratic state can be worthwhile. It fosters a belief in the possibility of an effective and redistributive state, even if the reality falls far short. Indeed, even disappointment in the state’s current ineffectiveness and unfairness can betray a belief that it could be otherwise.
Conditions in Rio’s peripheries are quite different. The PT never laid down deep roots as a political force and was always forced to form alliances with clientelist parties that maintained a strong hold over local political culture. This meant the PT era was experienced as less of a rupture in the peripheries of Rio than elsewhere.
There has also not been anything like the same institutionalization of progressive policies. In low-income communities, the state is typically articulated via local political-economic power networks, often intertwined with violent militias. Notwithstanding the important presence of social movements in Rio de Janeiro, such conditions heavily constrain their ability to operate freely in the peripheries. The net effect is to reinforce among residents of these areas a certain indifference to the state, at least as a possible vehicle of incrementally institutionalized and improving services.
Rather than being a disappointment, the state becomes an irrelevance, and many of the basic needs of everyday life are sought via other means. It is not difficult to see how Bolsonarista discourse, emphasizing the failures of the democratic state and the need to seek security and opportunity outside of it, might resonate in such a context.
A Clash of Heterogeneous Coalitions
Returning to our original question, the first round of Brazil’s crucial 2022 election reveals that a socially and geographically heterogeneous electoral coalition has durably cohered around Bolsonaro. This includes diverse social groups along Brazil’s agro-frontier, who experience the state largely through environmental regulations that they perceive as a threat to their livelihoods. It includes some of the middle classes of Brazil’s wealthier regions, mainly in its small-town hinterlands, who have abandoned Brazil’s project of democratic consolidation in favor of a revanchist retaking — and reshaping — of the state. And it includes residents of Rio’s (and perhaps other) urban peripheries for whom the state has come to be seen largely as an irrelevance.
In other words, these are groups who, under very different circumstances and for different reasons, have converged on a consensus that Bolsonaro’s “anti-governance” is certainly not a deal breaker and may even be a positive step.
Ranged against them are poor voters, especially the Northeast, who wish to see a resumption of the PT’s project of social inclusion; an institutionalist middle class that, after flirting with Bolsonarismo, has rediscovered its democratic bearings; and a skeptical urban working class, particularly in São Paulo’s urban peripheries, that retains a residual hope in the eventual consolidation of a functioning, redistributive state.
It is these two heterogeneous, internally fraught but now largely consolidated blocs that will clash on October 30, in the next round of voting. Given their similar size, it may be a relatively small number of swing voters and undecideds who determine the outcome, which will have enormous implications for Brazilian democracy in the years to come. But even if Lula wins, Bolsonarismo will endure in opposition, offering a destructive alternative to those who have lost, or perhaps never had, faith in Brazil’s democratic state.