Debating a change to the system Brazil uses to choose its federal deputies has become a post-election ritual. This year was no different, and the push came right after the runoff, with the Speaker of the Chamber of Deputies, Hugo Motta (Republicanos-PB), signaling support from several leaders of the so-called "Centrão" (literally the "big center", the bloc of non-ideological parties that has backed nearly every government since the return to democracy) for a change: leaving the current open-list proportional system for a "mixed-member district" system (distrital misto), with enthusiastic support from leaders such as Gilberto Kassab (PSD-SP) and Aécio Neves (PSDB-MG).
The study below tries to anticipate the effects of such a change, drawing on what we know from other countries' history and on the knowledge political science has accumulated on the subject. Its main conclusions are these:
- Although we do not yet know the details of the proposed system, if we assume that half of the deputies would be elected in single-winner districts and the other half by party lists, there are few scenarios in which the Centrão parties would be better off than under the current proportional system.
- The first reason why approving this system would be a historic mistake for the Centrão lies in the famous Duverger's Law: plurality races (in which only one candidate can win) produce a two-party tendency, in which candidates and voters sort themselves into two blocs with a real chance of winning instead of the proliferation of candidacies seen in proportional systems.
- Although Duverger's Law does not play out the same way in every district (that is, the same parties would not lead the two main blocs everywhere), another phenomenon would put enormous pressure on Centrão candidates: the nationalization of the vote for deputy, which has grown in recent elections, would benefit the party poles with competitive presidential candidacies. While Duverger's Law produces only two sides in each district, the nationalization of races decides which parties those would be.
- Betting on budget earmarks (emendas) to strengthen the dominance of Centrão politicians in their regions (now districts) would help contain these trends, but in general would not be enough to reverse them. In most scenarios the dominant strategy for candidates seeking reelection would be to join a party that represents one of the presidential poles. In other words, the deputies may survive, but the Centrão parties would lose seats.
- Coordination among parties, including representatives of the poles such as PT and PL depending on the district, could soften the Centrão's losses, but for that to happen parties would have to unilaterally give up the dominant strategy of running a candidate in every district to "fatten" the share of the statewide vote that still decides half of the national deputies. PT and PL above all would gain from running candidates in as many districts as possible, to dominate the list vote in their best states and win as many district seats as they can.
- The only advantage left for party leaders would be direct control over access to district candidacies and over the order of the list, now closed. However, this would also heighten the chance of internal conflicts that would lead many lawmakers and would-be candidates to look for other parties, especially those firmly identified with one of the presidential poles, to run in their best districts.
Overall, the study shows that the Centrão's support for the mixed-member system rests on two very low-confidence assumptions: that the change of system would not change politicians' strategies for getting elected to the Chamber, and that there would be no significant rise in the nationalization of elections for federal deputy.
Changing the rules of the game without expecting the relevant players to change their behavior is, to say the least, a very risky bet. Given what we know about Brazilian politics and the trends shown by voters, choosing the mixed-member system would effectively mean the Centrão shooting itself in the foot: in its political strength and in its ability to act as the kingmaker (the fiel da balança) of every government since the return to democracy.
WHAT WOULD CHANGE UNDER A MIXED-MEMBER SYSTEM?
How does the current system differ from the proposed ones? Today Brazil elects its deputies by open-list proportional representation. Each state has a number of deputies set by the Constitution and by its population, with a minimum of eight and a maximum of seventy, which over-represents the smaller states at the expense of the largest ones, São Paulo above all.
All valid votes for deputy (that is, for candidates and for party labels, excluding blank and spoiled ballots) are counted and divided by the number of seats the state is entitled to. The result of this operation is the famous electoral quotient. Next, the combined votes of all of a party's candidates, plus the votes cast for the party label, are counted and divided by the electoral quotient. That number, rounded down, is the number of seats guaranteed to the party. These seats go to the party's candidates with the most votes: that is why the list is "open", since voters set the order. Closed lists would have a preset order of candidates, with voters only choosing the party.
The mixed-member system reshapes this: half of the deputies would still be elected from the party list (in this case probably closed) and the other half in districts, each electing one deputy. In other words, a state with eight seats would have four districts, each electing one person by plurality, with the four remaining seats decided by the parties' share of the vote and filled in the order of their lists. Each voter would thus have two votes: one for the district candidate and another for the party.
There are many ways to organize how these two halves would work in practice, and the various proposals put forward over the past decades differ significantly in how they are calculated. One model, which we call "compensatory" and which is based on the German system, was introduced by then-senator José Serra in 2017 and is technically still pending (PLS 86/2017; PL 9.212/2017): in it, the party vote sets the size of each state delegation and the district vote only fills it. If, in a state with eight deputies, a party wins four districts and 50% of the party vote, it gets four seats, held by its district winners; the other four seats are decided by the other parties' lists. In this case, winning one's district guarantees a place at the top of the final list while proportionality is kept -- except when a party wins more districts than its share of the vote would naturally give it, which can cause distortions.
Another model, which we call "parallel" here, treats the two halves as independent. If, in a state with eight deputies, a party wins four districts and 50% of the party vote, it gets six seats: four for winning the districts and two for its 50% share. This model carries a greater potential for distortions caused by the plurality vote.
We do not yet know which model will be proposed. The constitutional amendment (PEC) announced by Hugo Motta has no text and would need 171 signatures to be introduced. The expected rapporteur is Domingos Neto (PSD-CE), the same rapporteur as for PL 9.212 above. There has been talk of the system being adopted only in 2034, with a transition in 2030 that is still uncertain. Others mention the possibility of reopening the debate on the so-called "distritão", which in practice simply means electing the deputies with the most votes in each state, doing away with party proportionality. We do not deal with the distritão directly here, since the assumptions about its effects are quite different from those of the mixed-member system.
WHY DOES THE CENTRÃO WANT DISTRICTS?
The push to debate and approve a district model reveals part of the Centrão's reading of the election results: it did not expect the reelection rate of sitting deputies to stay similar to past elections1, probably because it expected mandatory budget earmarks to secure many more regional votes than were actually seen. In our estimates2, the earmarks did raise the vote of the candidates who had them, but only enough to defend their regional strongholds better; for candidates at the margin of their lists, the effect turned out too modest to secure reelection.
In this context, it is easy to find an old "villain" that many Chamber leaders have attacked in past debates on electoral reform: the vote-pullers (puxadores de voto). In this election the phenomenon at least doubled in size3, with deputies such as Nikolas (PL-MG), Pavanato (PL-SP) and Erika Hilton (PSOL-SP) "carrying" many candidates into the Chamber on the strength of their personal vote:
Seeing big personal brands as an obstacle to the regional dominance the Centrão parties expected is not unreasonable. Candidates with statewide appeal "steal" votes from local leaders, many of whom still depend on fairly sizable vote totals outside their regions to reach the quotient.
So, for several reasons, districts become tempting. Besides drastically reducing the number of candidacies and the logistical cost of running a campaign -- one can campaign in a single district instead of the whole state -- the district vote would let local leaders ignore the vote-pullers altogether, except in the districts where these chose to run: in theory, Nikolas would only get himself elected, without taking thirteen other deputies along, while the former mayor of a mid-sized city would be well placed to secure a seat in the Chamber.
However, this tempting idea rests on problematic assumptions about electoral patterns carrying over under completely different systems. Changing the rules of the game affects all the players and their strategies, including the deputies themselves but also voters and parties as political organizations. When we explore the potential effects of districting, the picture does not look favorable to the Centrão, mainly because of Duverger's Law.
WHAT IS DUVERGER'S LAW?
In the 1950s the French political scientist Maurice Duverger noticed an interesting regularity. When only one candidate can be elected by plurality (what we call majoritarian elections), the race tends to concentrate on just two parties instead of several. This happens for two reasons: first, would-be candidates with no chance of winning have little incentive to run if second place leads nowhere, and are better off allying with or backing one of the front-runners. Second, voters avoid "wasting" their vote on someone who will not win and tend to move to the candidate closest to their preferences among the two most competitive (what Brazilians call the "useful vote", voto útil). The tendency toward two-party competition in this kind of election became known as Duverger's Law. It does not happen in our current system because seats are allocated proportionally: a party with 10% of the vote elects about 10% of the delegation, so it pays to run as many candidates as possible.
The effects of Duverger's Law can be shown in mayoral elections. In 2011 the economist Thomas Fujiwara published an article exploring the case4. In Brazil, municipalities with up to 200,000 voters elect their mayor in a single round by plurality, while larger municipalities hold a runoff if no candidate exceeds 50% of the valid votes. Fujiwara shows that in single-round cities voters tend to abandon the third-placed candidate and concentrate on the top two. We replicated the test using every municipal election from 2000 to 2024 and the result held: cities just above 200,000 voters had on average about one more candidate than cities just below. They also had about three percentage points more of the vote going to candidates other than the top two. In smaller cities the race is almost always between just two: in 66% of municipalities with fewer than 100,000 voters, the top two candidates add up to at least 95% of the vote.
However, the effect is not as strong when the electorate is larger. In municipalities with 100,000 to 200,000 voters, which also elect their mayor in a single round, the top two reach 95% of the vote in only one election in five. A federal deputy district would have just over 600,000 voters, three times the size of the largest single-round municipality. In Brazilian majoritarian elections of that size, between 15% and 25% of the votes go to candidates outside the top two. Duverger's Law reduces the number of candidates and ensures that only the winner takes the seat, but it does not entirely eliminate the vote for third-placed candidates when the electorate is large. The same holds for the Senate: in years with a single seat, decided by plurality, the race is on average no more concentrated than the first round for governor.
Moreover, Duverger's Law says that the race tends toward just two sides, but not necessarily the same sides across the whole country. In some districts the contest could be between PSD and União Brasil, for example. Regional leaders from the Centrão would tend to be one or two of the strongest candidates in many races. The problem is that there are several signs that Brazilian elections have recently been concentrating along alignment with the presidential poles. This year, the races for governor were the most concentrated on the top two candidates since 2002: 21 of the 27 states were decided in the first round, with only 11% of the vote going to third options. In 2024, municipal elections in the largest cities were also at their most concentrated since 20085, even in races with a runoff, which do not force an immediate binary choice. Everything suggests that national polarization is dominating many state and municipal races.
IS THE VOTE FOR DEPUTY BECOMING NATIONALIZED?
If districting would bring about the effects of Duverger's Law, how likely is it that the two main sides in each district would reflect national polarization? There is no way to be sure about voters' reasoning without opinion polls designed for that, but it is possible to measure recent electoral behavior. We compared votes across electoral zones for three kinds of office: the vote for city councillor in the previous municipal election, the most locally determined; the vote for president, the most national; and the vote in state races, which sit in between (federal deputy, senator, governor). If a party's vote for deputy in a zone looks like its vote for councillors, the vote is more local; if it looks like the vote for president, it is more national, forming a scale from 0 to 16.
Using this measure, we can say that the vote for federal deputy has become increasingly aligned with the presidential vote: it stood between 0.12 and 0.15 from 2006 to 2014, rising to 0.26–0.30 in the last three elections. An even bigger jump happened in the races for governor, which rose from 0.3 in 2018 to 0.54 in 2022 and 0.68 in 2026. In other words, with elections held on the same day, voters ended up "pasting" their presidential vote onto their other votes far more often than in the previous decade. The stronger alignment for governor makes sense, since the proportional vote for deputy tends to scatter voters' options. But since 2018 nationalization has grown for every office with a statewide vote.
And this is the key question for estimating the effects of districting. If half of the deputies are elected in single-winner districts, how will voters behave? Will they keep today's levels of nationalization, or will this alignment grow and look more like the vote for governor or senator? It cannot be predicted with any clarity, but it is possible to estimate what would happen if the alignment grew, with a parameter running from 0 to 100%. If the vote for deputy stays as local as it is, the parameter is 0; if the vote for deputy becomes as nationalized as the vote for governor, the parameter is 100%. From here on, when we speak of "30% nationalization", we mean that the district vote has gone 30% of the way from today's vote for deputy to the vote for governor, not an index of 0.30 on the scale above.
There are several reasons to believe nationalization would grow. First, since elections for deputy are held at the same time as presidential elections, it is natural for them to track the presidential vote more consistently than municipal elections do. Second, the races for governor were already mostly contests between the two presidential poles: 14 of the 27 races7 for state governments in 2026 were contests between the camps represented by Lula and Flávio in the presidential race. Third, the profile of the vote-puller itself is now more ideological and tied to the left and right camps, in contrast to the Tiriricas of the past (the TV clown Tiririca was elected in São Paulo in 2010 with more than a million votes).
This does not mean that districting would automatically produce generalized two-party competition between PT and PL. In 2026, 61% of the gubernatorial finalists who did not belong to those two parties ran in coalitions that included one of them, and party loyalties vary from state to state. For example, PP ran for governor of Paraíba with Lucas Ribeiro in a coalition with PT, while in other states it was allied with PL, as in Mato Grosso do Sul with Eduardo Riedel and in the Federal District with Celina Leão. The PSD of Eduardo Paes, in Rio de Janeiro, and of Omar Aziz, in Amazonas, ran alongside PT; the PSD of Dr. Furlan, in Amapá, alongside PL. In other words, in a district the Centrão candidate could still be one of the two sides, but as a local partner of one of the presidential poles.
HOW MANY SEATS WOULD THE CENTRÃO LOSE?
Estimates of how much nationalization of the Chamber races could cost the Centrão parties depend on many factors beyond what can be seen today, four years before the next election. Alliances, mergers, splits, municipal results, the political moment: all of this will affect the 2030 vote in ways that cannot be anticipated. But to reach a reasonable approximation, we simulated the 2026 elections as if the system were already districted, both under the "compensatory" model of José Serra's bill and under the "parallel" model described above.
For these simulations, we split every state into districts in a hundred different configurations per state, keeping only three rules: districts must be contiguous, keep the current geographic boundaries of the electoral zones and have electorates of similar size8. In each district, the parties field their candidate with the most votes within that district in 2026, with parties that were in coalition for the state governorship fielding only one candidate per coalition, the strongest one. The big vote-pullers are thus confined to running in a single district and, at most, campaigning for the other candidates of their camp in the rest. The parties that count as Centrão are those outside the two presidential poles: PSD, MDB, União Brasil/PP, Republicanos, PSDB/Cidadania, Podemos and other smaller parties, with the exception of Missão. These parties elected 257 deputies in 2026; applying the current rules to the same votes used in the simulations, they get 253, and that is the number we use for comparison9.
From there, the simulations apply different rates of nationalization to the 2026 vote to see how many deputies each of the three camps (PT and allies, PL and Novo, Centrão) would have. The results are telling: the Centrão would gain a few deputies under the mixed-member system if there were no additional nationalization of the vote for deputy, but would lose a lot even with quite modest levels of nationalization. Under the parallel model, 4% nationalization would be enough to wipe out the seat gain. With 30% nationalization, the Centrão would fall to 176 seats under the parallel model and 202 under the compensatory one. Under the parallel model, losses exceed 100 deputies at 50% nationalization.
Who would take these seats that today belong to the Centrão? Under the parallel model, mainly the pole that leads the first-round presidential vote, that is, PL in 2026. With 30% nationalization, the Bolsonarist camp could get 195 seats, against 141 in the no-change scenario, while the left camp would have 140 against 102 without change. The Centrão's losses would be concentrated in the states where the race for governor was far more national than the race for deputy. In São Paulo these parties would get only 11 deputies instead of the 34 they would have without nationalization; in Rio de Janeiro they would fall from 24 to 9, in Pará from 13 to 3. In eight states the number would not change, because there the race for governor was less nationalized than the race for deputy: in Minas Gerais and Rio Grande do Norte, because the vote for deputy was already very national; in Pernambuco, Maranhão and other states, because the race for governor followed local alliances and not the presidential divide. This, however, reflects a limitation of the exercise more than any guaranteed resilience of the Centrão in those states.
The picture is even more complicated than it looks. The Centrão's small gain when the vote for deputy does not nationalize further appears only because, in the simulations, parties run in coalition with the poles of the state races, that is, some degree of inter-party coordination would be taking place. If every party ran candidates in every district, the Centrão would lose deputies even without additional nationalization: 206 seats under the parallel model and 239 under the compensatory one.
WHAT WOULD THE DEPUTIES DO?
The numbers above take the point of view of parties as organizations interested in maximizing their delegations. Each deputy, however, has an immediate interest in keeping their seat in the Chamber. We can use Hugo Motta himself as an example. In the current scenario, Motta is the favorite to win practically any district in Paraíba's Sertão backlands and would remain the favorite with 30% nationalization (he wins in 87% of the maps). But if the vote for deputy became as national as the vote for governor, even the Speaker's seat would be threatened by PT challengers, such as current deputy Luiz Couto, who was not reelected this time.
Domingos Neto, the likely rapporteur of the PEC in question, would already have a hard time depending on how his district is drawn (he wins in 85% of the simulated maps). With 30% nationalization, his districts, which gave Lula 77% of the vote on average, would elect a PT candidate in every map. Both Motta and Domingos could only secure their seats by moving to PT or by making sure that no other candidate represented Lula's camp besides themselves.
To take an example from another region and another ideological camp: the former mayor of Londrina, Marcelo Belinati (PP-PR), would be reelected comfortably in every map as long as nationalization did not happen. With 30% nationalization, Belinati would lose in 98% of the maps to a PL candidate running in direct tandem with the presidential candidacy.
These cases are not exceptions; the Centrão's largest delegations would have dozens of deputies at risk of defeat under nationalization, even a moderate one. Again taking 30% nationalization as the baseline scenario, a Centrão deputy elected in 2026 would win their district in only 22% of the maps when facing a challenger from the presidential pole that dominates their region; by switching parties, the figure rises to 62%. Counting the seats generated by the closed list, 56% of the deputies would keep their seats under the parallel model, against 87% of PL and Novo deputies10. In other words, the best strategy for many Centrão deputies would be to leave their parties for others more clearly aligned with the presidential race, so as not to face challengers from the poles.
But what about budget earmarks: would they be enough to stem the bleeding? We tested the argument by assuming that each Centrão deputy concentrated in their district, for two terms, the median amount of individual earmarks paid per deputy in 2023–2026, about R$ 92 million per term11, thereby giving them the vote bonus that earmark money brings. All that money helps hold on to several seats: in the 30% baseline scenario the Centrão manages to keep between 16 and 26 additional deputies under the parallel model and up to 7 under the compensatory one. Even so, the losses caused by the nationalization of the vote for deputy would far outweigh these gains, which brings us to the Centrão's last chance to keep its size: full coordination among the parties.
COULD THE PARTIES COORDINATE?
Coordination among parties is at once the Centrão's best hope and the biggest blind spot of this study. In the simulations, the parties in the state coalitions field only one candidate per district, with 80% of the other parties' voters following that candidate. This percentage is our choice and is not easy to measure: races for governor and senator in 2022 and 2026 showed widely varying values of this metric (from 30% to over 100%)12, depending on the state.
Presumably, the quality of each candidacy would greatly affect this conversion: a candidate who is weak but convenient for party leaders could lose a district for lack of enough conversion to stand up to a challenger from the presidential poles. In any case, if all the voters of the partner parties followed the coalition's lead, the Centrão would have firmer chances of keeping its delegations. It would still face significant losses, however, at nationalization rates above 30%.
The hard part, as the Brazilian saying goes, is getting the Russians to agree. Or rather, PT and PL. In the district model, the presidential poles have every reason not to cooperate with this effort. Even these poles' defeated district candidates help build up the share of the vote parties need to secure seats through the party lists, which decide the other half of the seats. In other words, running candidates in every district is the best strategy for PT and PL, even where they have no chance of winning -- and many certainly would have one if they managed to nationalize their races.
There is also an effect within the state coalitions themselves. With greater nationalization, the coalition's ideal candidate is no longer the local Centrão politician but the strongest candidate who best represents the presidential pole. One example is Yuri Faro (PT-PA), who failed to get elected in 2026: within the MDB-led coalition in Pará, Faro loses to the MDB names in almost every map if there is no nationalization. However, he is the favorite if nationalization reaches 30%. Faro is not alone: at that level of nationalization, 33 defeated PL candidates and 23 defeated PT candidates in 2026 would win their districts in at least half of the maps. The poles also have their own local leaders. Angioletti Rubens (PL-SC), deputy mayor of Itajaí, got 82% of his votes in the city and was not elected; if he ran in a district containing Itajaí, he would win in practically every map, whatever voters did. During the campaign, he said that Itajaí and its region had not had a federal deputy of their own since 2003.
WHO WINS INSIDE THE PARTIES?
If the change is bad for the Centrão as a whole, why do its leaders defend it so eagerly? The answer lies in noticing that the most active proponents are party presidents: Gilberto Kassab, of PSD, who announced that the party had "closed ranks" in favor of the mixed-member system, and Aécio Neves, of PSDB, who presented the timetable of a transition in 2030 and adoption in 2034. Podemos, chaired by Renata Abreu, was at the meeting with Motta where the timetable was discussed. If the lists are closed, party leaders will decide who runs in each district and in what order candidates appear on the list.
Renata Abreu's case is the clearest. In São Paulo she would not win her district in any of the maps, even in the no-nationalization scenario, because her votes are spread across the state: her best district holds only 9% of her vote. Under the mixed-member system, her seat would depend on the list her own party drew up. The leaders of the party caucuses in the Chamber, all reelected in 2026, would fare differently. Pedro Lucas Fernandes (União Brasil-MA) and Augusto Coutinho (Republicanos-PE) would lose their districts even in the no-nationalization scenario, winning in at most 2% of the maps, almost always to candidates from other Centrão parties: Vinicius Ferro (MDB) in Maranhão and Miguel Coelho (União Brasil) in Pernambuco. Antonio Brito (PSD-BA) and Isnaldo Bulhões Jr. (MDB-AL) would win in most maps (85% and 66%), but only because they are in the right coalitions: running alone, they would almost never win.13
Greater control over candidacies and party lists, especially the order of who gets elected, would be a big advantage for party leaders. On the other hand, party loyalty in Brazil is not exactly rock-solid, and tight control over the delegations can easily become a source of internal conflict. Deputies who found themselves at a disadvantage could decide to switch parties to escape that influence, reducing the chance of coordination. Kassab's threat that any member who disagrees "will be invited to leave" PSD cuts both ways: under the mixed-member system, leaving may be the best option for the deputy.
THE ITALIAN EXAMPLE
Brazil would not be the first country to trade a proportional list for districts. Italy did so in 1993, after a referendum against "partitocracy" amid the Tangentopoli scandal uncovered by the Clean Hands investigation (Mani Pulite), moving to elect 75% of its parliament in districts and 25% by lists, in a system14 known as the Mattarellum. What happened to the Italian political center is an indication of what could happen to the Brazilian Centrão.
In the first election with districts, in 1994, the Christian Democrats split. One part ran outside the two big blocs as the Patto per l'Italia, winning 15.75% of the list vote. However, it won only 4 of the 475 districts, ending up with just 7.3% of the seats. Another part, called CCD, decided to ally with the right-wing bloc led by Silvio Berlusconi. Despite not having its own list, CCD elected 27 deputies, 21 of them in districts ceded by its allies. Together, the heirs of the Christian Democrats elected 73 deputies; they had 206 in 1992. In 2001 the lesson repeated itself: CCD-CDU, inside Berlusconi's bloc, got 3.22% of the list vote and fell below the threshold, but elected 40 deputies in districts, while Democrazia Europea, outside the blocs, got 2.39% and elected no one (Italian Interior Ministry data).
The comparison is not perfect, since much of the Christian Democrats' collapse came from the corruption scandals and not only from the change in electoral law. On the other hand, it is clear that the best survival strategy under districting runs through allying with the major national poles, contesting only the favorable districts that the bloc's larger party finds convenient for its strategy. As soon as it could, the Italian center worked to undo the change: UDC, the political heir of CCD-CDU, was the party that pushed hardest for the return of proportional representation in 2005.
CONCLUSION
Even after exploring the advantages the mixed-member system could offer the Centrão, above all the drastic reduction in the vote-pullers' influence over the makeup of state delegations, the bet looks like a gross misjudgment of how electoral strategies would be affected by the change of rules. The attempt to secure seats for local leaders could backfire if voter behavior becomes more aligned with the polarization based on the presidential vote that we have seen over the past decade.
It is hard to see real advantages for these parties in adopting a mixed-member district system. Success in the venture would depend on the cooperation of the parties that would have the most to gain from playing aggressively under the new system, PT and PL. The competition among the Centrão parties themselves would have to be settled and everyone's actions coordinated, at great complexity, just to keep the delegations as they are. Individual deputies would have strong incentives to join, once and for all, the pole that works best in their regions, undermining the Centrão parties' ability to serve as the kingmaker, the trait that has so often defined their political role.
Because of all these problems, as well as others we have not discussed here (such as the risks of gerrymandering), approval of a constitutional amendment adopting a mixed-member district model does not look viable. PL, one of the potential beneficiaries, has already come out against it through its leader in the Chamber and through the leader of the opposition; we found no public position from PT. Motta's own plan foresees a transition in 2030 through a mixed "distritão", in which half of the seats would go to the candidates with the most votes in each state and the other half to the lists, which would also struggle to pass. Still, should the idea move forward, it would be very interesting to follow the negotiations between the parties and their members as they try to make sure their delegations do not shrink drastically. If the vote for deputy nationalizes significantly and the presidential poles advance on their territory, the Centrão's survivors may find themselves stripped of the title, shrinking from a Centrão into a Centrinho, a "little center".
-
Of the 513 deputies elected in 2022, 405 tried to return to the Chamber in 2026 and 294 succeeded, that is, 73% of those who tried. The other 108 ran for other offices (64) or did not run (44). In previous elections, the success rate of those seeking reelection was 74% in 2014, 64% in 2018 and 67% in 2022. In other words, 2026 had the best result since 2014, but within the range of the past decade. (Our own count from TSE open data, considering the deputies elected four years earlier, identified by voter registration number; substitutes who took office are not counted.) ↩
-
To measure the effect of earmarks, we compared, for each deputy who sought reelection in 2018, 2022 and 2026, how their vote changed in the municipalities that received their individual earmarks and in those that did not (payments recorded in the Portal da Transparência from 2015 to 2026). The comparison nets out the deputy's change across the whole state and the change in each municipality. Where the deputy was already strong, with at least 10% of the local vote, nearly tripling the amount per voter is associated with a gain of about 1.6 percentage points; in other municipalities, 0.9 points. But one caveat is needed: the money the same deputy sends after the election "explains" the earlier vote gain even better than the money sent before. Since money sent afterwards cannot have caused votes that came earlier, this suggests that much of the association comes from deputies sending funds where they already have votes. In other words, these numbers are a ceiling on the effect of earmarks rather than an exact estimate. ↩
-
We call a vote-puller a candidate for federal deputy who alone had enough votes for at least one electoral quotient in their state. Whatever exceeds the first quotient is the surplus that helps elect list colleagues. In 2026, 34 vote-pullers added up to 38 surplus quotients, against 16 in 2022 and between 14 and 19 in every election since 2002; 96% of that surplus went to the two poles, 77% to PL and the PT federation alone. Nikolas Ferreira had votes worth 14.5 quotients in Minas Gerais: one for himself and 13.5 for the list. (Our own count from TSE open data.) ↩
-
Thomas Fujiwara, "A Regression-Discontinuity Test of Strategic Voting and Duverger's Law", Quarterly Journal of Political Science, vol. 6, no. 3-4, 2011, pp. 197-233. DOI: 10.1561/100.00010037. ↩
-
The measure is the sum of the valid votes of all candidates from third place down, in the first round. For governor, the average of the 27 states was 10.9% in 2026, the lowest since 2002; the previous low was 12.5% in 2006 and the high 24.7% in 2018. For mayor, in cities with a runoff, 19.1% in 2024, the lowest since 2008 (18.7%). ↩
-
For each electoral zone and each party, we look for the mix of the party's share of the councillor vote in the previous municipal election and its candidate's share in the first presidential round that best reproduces the party's share for the office being measured. Each presidential candidate's vote counts for their party, and federations count as one party. The numbers in the text are the median of the country's roughly 6,000 zones. The 2002 election falls outside the pattern of the following ones, at 0.24 for deputy and 0.44 for governor. ↩
-
We count a race as between the two poles when the top two candidates were one from each camp: a candidate counts as being in Lula's camp if their party or coalition included the PT federation, and in Flávio's camp if it included PL. Informal endorsements do not count, so the number is a floor. Under the strict rule, in which the candidate's own party must be PT or PL, none of the 2026 races was between the two. In twelve other states only one of the poles was among the top two and, in Maranhão, neither. The number is not a record: there were 20 in 2010 (when the poles were PT and PSDB) and 3 in 2018. The share of the gubernatorial vote going to candidates from one of the two camps, however, has grown since 2018: 45%, 60% and 73%, on average across the states. ↩
-
Each state gets half of its seats as districts, rounded down: São Paulo has 35 districts out of 70 seats, Minas Gerais 26 out of 53 and the states with eight deputies, 4. That makes 253 districts nationwide. Districts are built from the parts of each electoral zone within each municipality, 7,060 in total; in Roraima and the Federal District, where the zones are too large, we use polling places. District electorates within a state differ by at most 5% (7.5% in Amapá and Maranhão, where the zones do not allow more precision). The maps are drawn by an algorithm that does not know the votes (called ReCom), a hundred for each state. Islands such as Ilhabela and Fernando de Noronha are joined to the nearest units. ↩
-
We count as the left camp the PT federation (with PCdoB and PV), PSB, PDT, PSOL-Rede and smaller left parties; as the Bolsonarist camp, PL, Novo and PRTB; Missão separately; and all other parties as Centrão. Federations count as one party. The current rules applied to the votes used in the simulations give the Centrão 253 seats, four fewer than the 257 elected, because the model leaves out the minimum vote each candidate must reach (10% and 20% of the quotient). We use 253 in every comparison so that different rules are applied to exactly the same votes. ↩
-
A deputy elected in 2026 keeps their seat in the simulation if they win their district or enter through their party's list, taking the list in the order of the 2026 vote. The percentages are the average of the 100 maps of each state. ↩
-
The vote bonus comes from the estimate in the note on earmarks, applied to the median amount. The range of 16 to 26 seats corresponds to different readings of the effect: the average estimate across all deputies or the separate estimate for municipalities where the deputy was already strong. At the limit, if the poles stopped contesting the districts a Centrão deputy defends with earmarks, the Centrão would keep 282 seats under the parallel model and 259 under the compensatory one; this is a ceiling, not a forecast. ↩
-
We measured this conversion in the races for governor and Senate, looking for the percentage that best reproduces, zone by zone, the coalition candidate's votes from the councillor votes of the parties that made up the coalition. The best estimates were 30% (governor, 2026), 120% (governor, 2022), 65% (Senate, 2026) and 50% (Senate, 2022). A value above 100% means the partners' base yielded more than the base of the candidate's own party. We use 80% in all simulations. ↩
-
Sitting leaders according to the Chamber's open data on November 3, 2026; Augusto Coutinho has led Republicanos since February 5, 2026, in place of Gilberto Abramo. Without nationalization, Brito would win his district in 85% of the maps and Isnaldo in 66%; with each party running alone, in 3% and in none. ↩
-
Of the 630 deputies, 475 were elected in single-winner districts and 155 by lists, with a mechanism (the scorporo) that subtracted from the list vote part of the votes of those who had already won districts. A list needed 4% of the national vote to receive seats. The vote and seat data are from the Italian Interior Ministry's historical archive. ↩