A tight race and bad polls in Argentina: what do we know?
Sign up here to receive our regular Election Note! Read the full analysis and receive the full version of each note as soon as it is published.
Published 8 November 2023
Key takeaways
Sergio Massa is the narrow favourite to win the run-off against outsider Javier Milei. Massa significantly increased his vote share since the primary election using support from the unions, political allies and handouts
Public polling has not been accurate in this electoral cycle and should be treated with caution. Pollsters are struggling to adapt to Argentina’s changing political landscape and failed to accurately predict the primary and first round results
Everything depends on voters who did not vote for Massa or Milei in round one. Using “transition matrices”, our analysis demonstrates how the electoral maths can play out. If Milei manages to win over a significant majority of Bullrich voters, he will win the presidency
What is the state of play?
Amidst a historic economic crisis, Argentina is gearing up for the second round of the presidential election. In a nailbiter of a first round, the current government’s Minister of the Economy Sergio Massa and libertarian outsider Javier Milei secured their places in the run-off by winning 37% and 30% of the vote. Patricia Bullrich, the candidate for the main conservative bloc Juntos por el Cambio (JxC), fell just short with 24% of the vote. With Massa and Milei as the last ones standing, they will face each other in a head-to-head run-off election on the 19th of November to decide the presidency.
There is a lot at stake for Argentina. The country is currently facing triple-digit inflation and a plummeting currency. Moreover, it is the IMF’s largest debtor and has historically failed to meet even the most flexible conditions of its loan programme. What the two candidates have proposed as solutions to the ongoing economic malaise is fundamentally different. Massa, current centre-left Economy Minister, intends to focus on policies to boost the labour market, while increasing fiscal support for households and government aid. Milei’s programme, by contrast, is a much more radical departure from current government policy. He plans to dollarise the Argentinian economy, slash the size of the government and double down on fiscal austerity.
Figure 1: Line chart of vote shares by presidential coalition across the primaries and the first round of the election
Which of these two platforms do voters prefer? Currently, it seems that Massa has picked up some momentum in the race, and he is considered the favourite to win the presidency by most observers. As Figure 1 illustrates, Massa has been able to increase the vote share of his coalition - Union por la Patria - from 27.3% in the primary elections to 36.7% in the first round of the presidential election. These gains have appeared to come at the expense of Bullrich’s camp, which has dropped from 28.3% to 23.8%. Although it is difficult to draw conclusions about where these shifts come from, it appears likely that Massa has successfully attracted more moderate Bullrich voters in an effort to avoid a Milei presidency.
Figure 2: Line chart of weighted polling trends (Sergio Massa v Javier Milei)
There is much speculation around how these gains for Massa have materialised in such a short timeframe. Two factors appear to be behind this. Firstly, Massa’s Unión por la Patria (UP) has full access to the organisational campaign machinery that has facilitated numerous Peronist victories in the past. Over the past months, UP has been able to successfully mobilise the support of affiliated governors, mayors and trade unions, creating a ‘get-out-the-vote’ campaign with a reach Milei can only dream of. Moreover, Massa effectively cashed in on his incumbency advantage through strategically timed government support programmes to boost his personal popularity. The most striking was the expansion of wide ranging income tax exemptions to all workers - a plan that is very popular but also further erodes Argentina’s fiscal position.
Just one and a half weeks before the election, polls currently show Massa leading Milei by approximately 8 percentage points (see Figure 2). It should be noted that this is a relatively recent lead. Ever since the primary elections and throughout the months of August and September, Milei had held a commanding lead in head-to-head polling versus Massa. It was not until the results of the first round of the election were announced that the polls flipped. This is one of the reasons we advocate caution when it comes to Argentinian public polling. Too often, polls are seemingly adjusted or corrected after political events. More on this point in the next section.
How reliable are the polls in Argentina?
Recently, Argentinian pollsters have not had a great track record. For the first round of the presidential elections, pollsters underestimated Massa’s vote share, on average, by 5.2 percentage points. Similarly, Milei’s vote share was generally overestimated. The average absolute error for pollsters on predicting his vote was 4.3 percentage points. Errors of this magnitude constitute a significant polling miss, and this was not the first time. Milei’s victory in the August primary elections was a big surprise to observers, again, because pollsters had severely underestimated him. On average, pollsters underestimated Milei’s vote by 10(!) percentage points.
Figure 3: Average absolute polling error for major candidates per Argentinian election over time
This is poor, even when we compare this to the historical performance of Argentinian pollsters in previous elections. Figure 3 illustrates the absolute mean polling error for Argentinian pollsters when predicting the performance of major candidates per election. On average, pollsters were off by approximately 3.5 percentage points per major candidate between 1999 and 2019. In this year’s primary elections, however, error skyrocketed to 7.4 percentage points per major candidate on average. Pollsters slightly recovered in the first round of this year’s election, but still recorded misses of 3.8 percentage points per major candidate on average. These errors are very large in comparison to polling in other advanced democracies.
Another cause for concern in Argentinian polling is the way in which the industry average appears to “jump” after election results. As Figure 4 illustrates, Milei trailed his two rivals by double digits in polls published in the month before the primary election. Then, after Milei unexpectedly won the primary, polls showed Milei leading the pack by an average of 5-6 percentage points.
What can explain this complete reversal? Some Argentinian pollsters have gone on the record to explain that this phenomenon is related to “shy Milei voters”. These voters would have been reluctant to share their political preference for Milei before the election, but felt emboldened after his victory and started showing their true colours in polls. Given the size of the errors, we find this explanation highly unlikely. Instead, it’s much more likely that either a) polling methodology across the industry failed to capture the core Milei voter demographic or b) the industry got caught herding and failed to publish polling results that deviated from the existing consensus.
Figure 4: Weighted polling trends per coalition over time
All this does not mean that all Argentinian pollsters are equal. Several pollsters significantly outperformed their industry average. Two examples are worth mentioning. Firstly, Atlas Intel - after missing the Milei surge in the primaries - was one of the few pollsters to correctly predict a Massa victory in the first round of the election. Second, CIGP also performed reasonably well in round one, accurately predicting that Massa would finish ahead of Milei. Given the large recent shifts in Argentinian politics, these are two pollsters who seem to have a polling set-up that is well-adjusted with current trends. However, the second round will ultimately show whether this was luck or skill.
Is there anything we can learn from public polling?
Figure 5: Alluvial chart of voter flows between the first and second round according to Opinaia polling
Ten polls have been published since the first round election results were published. Of these polls, four show a Massa victory and six show a Milei victory. Now, as emphasised in the previous section, we cannot necessarily expect these pollsters to get the levels of support of the candidates right. However, it is instructive to take a closer look at how these pollsters got to these different results. For this, we’ll need to dissect the so-called ‘transition matrices’ of these polls - the way in which these polls are predicting the transfer of votes from round one to round two. In this section, we’ll borrow these transition matrices from two public polls and apply these to the real number of votes each candidate got in the first round. Doing so, we illustrate the paths to victory for both Milei and Massa as well as the electoral maths needed to make this work.
First, we will borrow the voter flow matrix from a recent Opinaia survey. The key question that we want to answer is simple: what will the voters of eliminated candidates do in the second round? In total, nearly 6.4 million Argentinians voted for Bullrich, 1.8 million voters for Schiaretti and 700,000 voters for Bregman. Since Massa won just over 9.8 million votes against Milei’s 8 million votes, Milei will need to make up a voter deficit of approximately 1.8 million votes to win the presidency.
Fortunately for him, the polls do give Milei’s campaign reason to believe that he will win the majority of the votes of the eliminated candidates. Opinaia estimates that Milei will capture approximately 47% of Bullrich’s first round centre-right vote, while Massa would only receive 9% of this vote (see Figure 5). 35% would not vote for either candidate, and an additional 9% is currently undecided. Given the size of Bullrich’s electorate, this would see Milei gain 2.4 million votes on Massa, thereby giving him the lead. According to this poll, Milei would also win more of the Schiaretti vote than Massa, by 37% to 26%. However, Bregman’s left-wing electorate would break more positively for Massa, who would take 52% of these voters compared to Milei’s 14%. Overall, if we multiply out these voter flows by the real sizes of their respective electorates, we find that Milei would win the presidency with 51.1% of the overall vote (a margin of approximately 500,000 votes).
Does that mean Milei will win? Not necessarily. There is also no consensus amongst Argentinian pollsters on how the voters of eliminated candidates will break down in the second round. Whilst Opinaia painted a relatively rosy picture for Milei, Proyección Consultores published polling results that inspire more confidence for Massa’s camp. According to this pollster, nearly half of Bullrich’s electorate would not vote for either candidate. Moreover, the ones that do only break for Milei by 60%-40%, a much more favourable breakdown for Massa than Opinaia’s. This poll also shows Massa winning more of the Schiaretti and Bregman votes than Milei. Overall, the electoral arithmetic, given these percentages, would give Massa the victory with 54.8% of the vote and a lead of over 2 million votes.
Figure 6: Alluvial chart of voter flows between the first and second round according to Proyección Consultores polling
These two examples illustrate two entirely different paths to victory. To win the presidency, Milei must make sure Bullrich’s voters turn out in high numbers and win approximately two-thirds of this vote. Then, if the Schiaretti and Bregman votes do not break too unfavourably for him, he would be able to win a majority. In this scenario, Milei could even afford to lose the Schiaretti vote by a small margin. However, Massa’s path to victory is slightly easier. His camp either needs Bullrich voters to turn out in relatively small numbers, or he needs to win a majority of the Schiaretti vote to ensure that he can make up for Milei’s gains from Bullrich.
In short, both candidates have their work cut out for them. Paths to victory exist, but ultimately depend on whether or not Milei can appeal to a big majority of the Bullrich vote. For the outside observer, surveys, exit polls and early results that show a breakdown of the Bullrich vote across these candidates are the key metrics to watch in the run-up to the election.
We are also keeping an eye on:
Dutch parliamentary elections - Without long-serving Prime Minister Mark Rutte, the November elections could cause major turmoil in the otherwise relatively stable politics of the Netherlands. With a united left, two new protest parties and a strong far-right flank, the race is chaotic, unpredictable, and - for the political geeks among us - incredibly fascinating. Voting 22nd November 2023.
European Parliament elections - After a tumultuous year in European politics, citizens of all 27 EU member states will head to the polls to elect their European representatives. Current polling shows a weakening lead for the EPP and increasing strength of the flanks of the Parliament. If that turns out to be true, coalition building will become increasingly complicated in the EP’s next term. Voting 6-9 June 2024.
US Presidential elections - With another presidential election year coming up, primary season is fast approaching. Kicked off by the Iowa caucus on January 15th, the Republican primaries are particularly interesting. Will ongoing legal proceedings affect Trump’s performance? And can any other candidate position himself/herself as a viable alternative? Voting for the presidential election will take place on 5 November 2024.