An Overlooked Problem with the Two-Round System

Coincident Second Round for President and Governor

By Thomaz Alvares de Azevedo e Almeida
January, 2007

Consequences of the Coincident Second Round

In Brazil, since Constitutional Amendment 16/97, the president and governor can run for one subsequent re-election. Since this amendment has been passed, three presidential elections have occurred: 1998, 2002 and 2006. In two cases a presidential candidate was running for re-election, 1998 and 2006. Also, in two presidential elections a delayed runoff occurred, 2002 and 2006. What makes the Brazil case special is the fact that Brazil can have coincident second round runoffs for president and governor. That means, not only presidential and governors' candidates are elected at the same time, but they also campaign at the same time.

I argue that the possibility of coincident second rounds creates a problem whose significance has been underestimated. Put simply, coincident second rounds can bias the voting tendency of the electors in favor of the governor candidates who are associated with their state's favorite presidential candidate. In addition, having a president running for re-election, as can happen in Brazil, also magnifies this tendency, affecting even the first round results of the governors' election. Thus, there is some electoral distortion that demands both our attention and some solution.

What led me to identify the distortion in favor of the governors' candidates related to the state presidential favorite candidate was the comparative analysis of the Brazilian 2002 presidential second round result with the results of the second rounds for governors. In 2002 there was no presidential candidate running for re-election. In 2006, when there was a president candidate running for re-election, we identified the voting distortion again and even stronger.

Re-election Magnifier Effect and Coincident Runoff Effect

The election data shows that amidst the 14 runoffs which occurred in 2002, in 8 the winner was the governor's candidate associated with the state presidential favorite candidate (even though this association was not necessarily formal). In 2006, when a presidential candidate was running for re-election this tendency was magnified. Governors' candidates associated with the presidential candidate running for re-election had a substantial increase in votes: 8 were elected in the first round, especially in the North and Northeast where the current president, running for re-election had massive penetration. That led to a decrease in the number of runoffs from 14 in 2002 to 10 in 2006.

Yet, even amidst the 10 runoffs that took place in 2006, 6 elected governors were associated with the presidential candidate running for re-election. In 5 of these 6 districts, the presidential candidate running for re-election was also the state favorite presidential candidate. In the other 4 runoffs, even though the re-election magnifier effect wasn't significant, the state presidential favorite candidate effect was still present: in 2 of these 4 districts the elected governor was associated with the presidential candidate with the highest support in that state. Thus, looking at the coincident second round elections, only in 3 of the 10 runoffs was the elected governor not associated with the favored presidential candidate.

Cause of Re-election Magnifier Effect and Coincident Runoff Effect

Why do we see such a coincident second round effect? The most parsimonious explanation seems to be that the presidential second round concentrates the media and the electors attention on the presidential candidates, especially on the current president if he/she is running for re-election. In order to test that explanation we looked at the difference between the audience on the presidential and on the governors debate on TV, and found that the governor's debate had far less audience than the presidential debate. For instance, the debate between the Pernambuco government at TV Clube had, according to the Brazilian Institute of Public Opinion and Statistics (Ibope), an average audience of 2.4 points, which means that close to 79,200 people watched the Pernambuco's governor debate. For the presidential debate at Rede Record the audience average was 17 points (again according to Ibope). That means that in the Sao Paulo city and adjacent areas alone, close to 880,000 people watched the presidential debate. That argues that there is a concentration from both the media and the electors attention on the presidential race, putting the governor's race at a minor stage.

The consequence of these considerations is that the presidential election increases the influence of the state's presidential favorite candidate upon the governors’ candidates. Thus the governors' candidates with chances of going to the second round begin to negotiate their support to the state's favorite presidential candidate in the first round. In addition, if the governor's candidate is locked through a party alliance with a presidential candidate who is less favored in his/her state, he/she is naturally encouraged to follow one of the three strategies: try to detach himself from the presidential candidate by ignoring him and not campaigning for him; try to detach himself from the presidential candidate with public attacks; or betray the party or alliance and endorse the other presidential candidate. Whatever strategy he adopts, all of them undermine the party system at the federative level; and yet no one can blame him given the state of the Brazilian party system--mandatory vote combined with an excess of parties, being most of these parties not well-rooted in society and lacking a party loyalty sense in most of their candidates.

Summary and Conclusion

To summarize, the problem with the current two-round system is that it has an effect of concentrating the voters' attention on the presidential race, putting the governor's race in a minor stage. That leads the governor candidates to emphasize their relation with the presidential candidates rather than addressing the states necessities. In that way, a race that should be about local issues turns itself into a subset of the presidential race, hence, undermining the federative ideal. There is a pleura of possible solutions to this problem, all of them based on changes on the electoral procedure. These solutions vary from increasing the president's term from 4 to 5 years--which would make this coincidence happen just once each 20 years--to limiting the president to one term--which would end with the re-election magnifier effect.

The change that I suggest is the replacement of the two-round system (delayed runoff) for president and governor by the instant runoff voting system (IRV), thereby eliminating altogether the need for a second round of voting. The consequence of this is that instead of playing safe and searching the support of the state presidential favorite candidate, which in an eventual second-round would increase his/her chances of winning, with IRV the governor's candidates have the incentive for running a more substantive campaign detached from the state presidential candidate's image. That's because all the media attention that is put upon the presidential runoff would not happen. Thus, the governor election--with all its candidates--would have a smaller chance of being polarized by the presidential election; thereby preserving the federative ideal. In addition, IRV would also provide a gain in legitimacy for the presidential election once instead of only looking at the top preference of the voters the election would be decided according to their general preference; thus, also strengthening the democratic ideal. Above all, IRV would be a valuable incentive for the elector to cast sincere votes instead of the so-called and nowadays popular "pragmatic" votes.