This is the first post in a series we will be publishing over the coming weeks: Perspectives on the Brazilian Elections
As Brazil’s election approaches, all eyes are glued to the polls. Since the start of the campaign, the two frontrunners – former President Luiz Inácio Lula da Silva of the Worker’s Party (PT) and President Bolsonaro of the Liberal Party (PL) – have remained more or less stable in their relative positions. In “stimulated” polls on first-round voting intentions (where candidates’ names are read out to respondents), support for Lula has hovered in the mid-40s, while Bolsonaro has bounced around in the low-30s. This seems to signal that Lula is in a comfortable lead, but is still short of the 50% threshold he would need to win in the first round and avoid a runoff.
Beyond these headline figures, a more challenging task is finding out where these votes are coming from. Following an election, it is easy to visualise the geography of a given candidate’s vote. Votes are counted individually and compiled at the scale of electoral zones (usually the size of a small town, or a district within a larger city), giving a fairly precise picture of their spatial distribution. By contrast, pre-election polling is carried out via small samples usually covering large areas, most commonly the national scale. Most polling companies also identify which of Brazil’s five regions respondents come from, giving some data at this scale. However, these are very large and diverse territories, making it hard to draw more specific inferences about the geography of voter behaviour. Fortunately, some polls are also conducted at the smaller scale of the federal state, which do give somewhat of a clearer picture. Since August, between them, Brazil’s major polling companies have conducted polls in every state of the federation (see Table 1). These polls have used different methodologies and sample sizes, which will likely produce distortions in the data. Nonetheless, together they provide the clearest picture we currently have of the popularity of the main candidates in each state.
Table 1. Comparison of the 2018 election result (first round) and latest polling for 2022 (first round)[1] by state
They show, for example, that the most pro-Bolsonaro states are in the far North and the South of the country. In a recent poll, the Amazonian state of Roraima favoured Bolsonaro over Lula by a huge 66% to 16% margin (with the remainder made up of other candidates and undecideds). The margin for Bolsonaro was only slightly smaller in polls in two other small Amazonian states – Rondônia (54% to 27%) and Acre (53% to 30%). The only other state where Bolsonaro has a similarly large margin – and would meet the 50% threshold for a first-round victory – is the southern state of Santa Catarina (50% to 25%). By contrast, Lula passed that threshold in polls conducted in eight states, all of them located in the Northeast region. His largest lead came in Maranhão (66% to 18%), followed by Bahia (62% to 20%), Pernambuco (62% to 22%) and Piauí (61% to 20%).
However, these heavily Bolsonarista and Lulista states, respectively, will have a limited impact on the overall result. With a total of around 2.2 million voters between them in 2022, Rondônia, Roraima and Acre account for just 1.5% of Brazil’s 157 million-strong electorate. The nine states of Northeast account for a far higher proportion – about 28%. However, the PT already performed very strongly in the region in 2018, so has relatively little room to improve its performance this time around. For example, in Maranhão, 61% already voted for Haddad in the first round in 2018, and just 24% for Bolsonaro. If we compare the 2018 result to the latest polling, it signals just a 5% increase in the PT vote in Maranhão and a 6% decline in Bolsonaro’s. Instead, the areas that will be most crucial are major population centres that voted heavily for Bolsonaro in 2018, but which seem to be swinging towards Lula this time around. In both respects, it is the Southeast region of the country – accounting for a huge 43% of the electorate – that looks like being decisive.[2]
Brazil’s three most populous states are all located in this region. By far the largest is São Paulo, accounting for 22% of the electorate, followed by Minas Gerais with 10% and Rio de Janeiro with 8%. If we compare the results of the first round in 2018 and the current polling, we get some idea of the movement of travel in each of them. In the first round in 2018, 48% of Mineiro voters voted for Bolsonaro and just 24% for Haddad. Current polling shows a significant reversal, with 43% planning to vote for Lula and 33% for Bolsonaro, ie. a swing of +20% in the PT vote, and of -12% in the Bolsonaro vote. In São Paulo, just 16% of the electorate voted for Haddad in the first round in 2018, whereas polls suggest 43% for Lula – a swing of +27% for the PT. By contrast, 53% of Paulistas voted for Bolsonaro in 2018, whereas just 33% plan to vote for him this time – a swing of -20%. But the largest swing of all looks set to occur in Bolsonaro’s home state of Rio de Janeiro. The polls suggest 44% will vote for Lula on 2nd October, compared to just 15% who voted for Haddad in 2018 (a swing of +29% for the PT), while they signal a huge fall in the Bolsonaro vote from 60% to 36% (a swing of -24%).
While these comparisons rest on very different types of data, which were collected using different methodologies, they do allow some informed inferences. They suggest that Bolsonaro has maintained a lead – although in most cases somewhat a diminished one – in sparsely populated agricultural frontier states of the Amazon and the Central West region, as well as in some southern states. They also show that the PT has maintained – indeed slightly increased – its large advantage in Brazil’s poorest region, the Northeast. More broadly, polling indicates a moderate to large swing towards the PT in most states. This means that many states that voted heavily for Bolsonaro in 2018 will likely give him smaller margins this time, and states that gave him a narrow victory will this time lean towards the PT.
However, there looks to be a very large swing in the most populated economic centres in the Southeast of country, especially São Paulo and Rio de Janeiro. These are of course very large and internally diverse states. They contain different kinds of territory, from small towns and rural areas, to wealthy urban neighbourhoods, low-income peripheries and favelas in the country’s two largest metropolises. As Liz McKenna and I have shown previously, in the urban peripheries of São Paulo and Rio de Janeiro, which constitute a sizeable electorate, there has been a long-term drift away from the PT. It will be interesting to see whether this trend is fully reversed this year. An alternative possibility is these areas will swing back towards the PT more gently than wealthier areas. This would signal a significant change in the social and spatial composition of the PT vote, with implications for the Party’s future electoral prospects and political positioning. Given the limitations of polling data for such analysis, we will have to wait until after election day to find out.
[1] Data aggregated by Poder360, using original polling data from IPEC, Datafolha, Quaest, Paraná Pesquisas, Ranking Pesquisa e Percent Brasil. [2] In addition to the more heavily populated Southeast and Northeast regions, the South region accounts for about 14%, the North region for 8% and the Centre-West region for 7%.