Nearly 100 million Mexicans are eligible to cast a vote in the country’s upcoming presidential elections on June 2nd which sees Morena’s Claudia Sheinbaum face the opposition alliance candidate, Xóchitl Gálvez, and Jorge Álvarez Máynez of the Movimiento Ciudadano. Ahead of the election, Andreas E. Murr and Michael S. Lewis-Beck present four election forecasting models; the presidential approval model predicts a close win for Sheinbaum, while the economic voting model suggests that she will lose. Models using voting intention and expectation data also suggest that Sheinbaum will win with an absolute majority.
Mexicans will vote in what may well be a historic election on June 2nd, 2024. Who will succeed the incumbent president Andrés Manuel López Obrador from the Movimiento Regeneración Nacional (Morena)? Mexicans face the choice between Claudia Sheinbaum, the candidate of the incumbent party; Xóchitl Gálvez, the candidate of the main opposition alliance between the Partido Acción Nacional (PAN), Partido de la Revolucion Democratica (PRD), and Partido Revolucionario Institucional (PRI); and Jorge Álvarez Máynez, the candidate of the minor opposition party Movimiento Ciudadano (MC). Who will Mexicans elect?
Below we present four election forecasts: two based on standard models of election forecasting, and two based on standard survey questions of pre-electoral polls.
Election forecast models
Presidential approval
The first model bases itself on presidential approval. Voters are assumed to treat an election as a referendum on the incumbent president: if they approve, they vote for the incumbent candidate; if they disapprove, they vote for the opposition candidate.
The incumbent president Andrés Manuel López Obrador has had approval ratings of about 68 percent, among the highest ever. They are similar to the approval ratings of Vincente Fox (PAN) who was elected in 2000. His PAN successor candidate Felipe Calderon won the presidency in 2006 with about 37 percent. Hence, this model would predict López Obrador’s Morena successor candidate, Claudia Sheinbaum, to win a similar vote share, which should be enough to win the presidency, if the minor-opposition candidate Jorge Álvarez Máynez performs well.
We regressed the vote share of the incumbent party’s candidate on presidential approval from nine months before the election, when the candidates were chosen, for the four elections between 2000 and 2018. We then plugged in the relevant approval rating of Andrés Manuel López Obrador. Our presidential approval model then predicts Sheinbaum to win, with 37 percent (though the prediction interval ranges from 13 to 70 percent).
“Claudia Sheinbaum: Inicio de campaña” (CC BY 2.0) by Eneas
Economic voting
The second model is based on economic performance. Again, the idea is that voters treat an election as a referendum on the incumbent president: if the economy did well, they vote for the incumbent candidate; if the economy did badly, they vote for the opposition.
Mexico’s gross domestic product grew by only two percent in the first quarter of this election year. This is among the lowest growth rates Mexico has experienced across the post-2000 period. It is similar to the growth rate for President Enrique Peña Nieto (PRI), who was president from 2012-2018. His PRI successor candidate José Antonio Meade only won 16 percent of the vote. Hence, this model predicts Claudia Sheinbaum to do quite badly.
More precisely, we regressed the vote share of the incumbent party’s candidate on GDP growth in the first quarter of the election year for the four elections between 2000 and 2018. We then plugged in the GDP growth estimate for the first quarter of 2024.
In sum, the economic voting model predicts Sheinbaum to win only 16 percent of the vote (with the prediction interval ranging from 10 to 25 percent).
Polls
Vote intentions
The first poll approach is based on voting intentions. The idea is that asking people how they intend to vote is a good approximation to how they will vote. According to a ‘poll of polls’ by Oraculus, Claudia Sheinbaum, the incumbent party candidate, has had a solid lead over Xóchitl Gálvez, the main opposition candidate. Figure 1 below shows the share of vote intentions for each candidate for each month for 2024. While the proportion of people intending to vote for Claudia Sheinbaum has been decreasing since January, it is still around 60 percent.
Figure 1 – Percent of people intending to vote for each candidate
Source: Oraculus (last updated: 6 May 2024).
More precisely, the vote intention polls forecast Sheinbaum to win about 57 percent (with the confidence interval ranging from 50 to 64 percent).
Vote expectations
The second poll approach is based on vote expectations. The idea is that asking people who they think will win is more informative than asking them how they intend to vote. By asking them about the likely winner, they will consider more information than just their own vote intention.
According to Investigaciones Sociales Aplicadas (ISA), Claudia Sheinbaum has been holding a solid lead over Xóchitl Gálvez in terms of vote expectations. Figure 2 below shows two vote expectations polls for each candidate from March and April. Both give essentially the same result: an absolute majority of citizens expects Claudia Sheinbaum to win.
Figure 2 – Percent of people who think that each candidate will win
Source: ISA. (We thank Felipe de la O Lopez for pointing us to these data.)
With respect to the exact vote share, we cannot be more precise. Suffice it to say that in our experience this share of vote expectations for Sheinbaum indicates a strong win. (For a more on vote expectations in Mexico, see our May 21 2004 post on OxPol.)
Mixed signals about the election result
The models and polls forecast different outcomes. This is unusual. The economic voting model suggests that Sheinbaum will lose the election. The presidential approval model suggests that Sheinbaum will win the presidency, but fail to reach an absolute majority, which her predecessor managed to do. The vote intention polls by contrast strongly suggest that Sheinbaum will have an absolute majority. And the vote expectation polls agree that Sheinbaum is the most likely winner. What explains this discrepancy? Which forecast should we trust more?
We think that polls may be more useful in this election. Polls aggregate all types of information without us forecasters having to explicitly model the information. By contrast, models are only as good as the information that we put into them. We think that the two standard models of election forecasting – economic voting and presidential approval – may be missing an important variable in this election: the level of populism under incumbent president Andrés Manuel López Obrador. Populism may change the electoral calculus. The polls by contrast should incorporate this information. Of course, we will soon know whether this idea is right and which approach was better.
- Please read our comments policy before commenting.
- Note: This article gives the views of the author, and not the position of USAPP – American Politics and Policy, nor of the London School of Economics.
- Shortened URL for this post: https://wp.me/p3I2YF-dTq
