The French presidential election of 2022 formed a domestic and international pivot in the world of established democracies. It is not coincidental that, for the first time, PS: Political Science & Politics offered a symposium on forecasting these contests, bringing together the work of nine researchers (or research teams). The French presidential election was key because of its economic and political weight, especially within the European Council, in which France exercises a leading governance role—and more so in the context of the COVID-19 crisis and its long-term internal and external consequences. This presidential race also raised serious concerns about the structuring of a nation’s political life around the Left–Right divide in France as well as in several other contemporary democracies. Would the Centrist “shock” observed in the 2017 French presidential and legislative elections (although not in the local elections held thereafter) reassert itself in 2022, ushering in a durable change in the traditional Left–Right French political equilibrium?

More specifically, would the moderate forces gathered around the incumbent, President Emmanuel Macron, either from a divided Left or the Classical Right by Valérie Pécresse, be able to recover from their 2017 debacle and reach the second round of the contest? Would Marine Le Pen create history by lifting the Extreme Right to the second round for a second consecutive election? Would Eric Zemmour and his new party Reconquête! upset the old game of the installed rival parties? Or would Macron again take advantage of his enfeebled moderate opponents, winning by default against a candidate that most voters considered too radical? These were among the many questions that this symposium, dedicated to the forecasting of the French presidential election, could help to answer.

Academically, an extensive literature has emerged around election forecasting in leading democracies during the past three decades. Perhaps oddly, however, the French case remains understudied. The use of forecasting models to predict the results of French presidential elections especially dates back to Lewis-Beck’s Reference Lewis-Beck1995 article. A few years later, Jérôme and Jérôme-Speziari (Reference Jérôme and Jérôme-Speziari2001) described another model that uses regionalized data to address the small-N problem (see also Foucault and Nadeau Reference Foucault and Nadeau2012). Fauvelle-Aymar and Lewis-Beck (Reference Fauvelle-Aymar and Lewis-Beck2002) presented the Iowa model to predict the French presidential elections. Appropriately, in this symposium, Stegmaier and Adou (Reference Stegmaier and Adou2022) review 27 years of French presidential election forecasting, with modelers adapting to the complexities of a two-round balloting system impacted by new issues and changes in the party system. Looking toward the future, they herald the new challenges that forecasters had to face in the Spring 2022 contests.

To meet the 2022 challenge directly, this symposium presents six different forecasting methodologies as applied to the French presidential election. Revisiting the classic Iowa model, Bélanger, Feitosa, and Turgeon (Reference Bélanger, Feitosa and Turgeon2022) propose a modified specification and a new definition of the dependent variable. The article by Jérôme-Speziari and Bélanger (Reference Jérôme-Speziari and Bélanger2022) follows the tradition of disaggregated-voting analysis on a regional basis but also offers a seemingly unrelated regressions (SUR) model. Innovating on two levels, the efforts of Jérôme, Mongrain, and Nadeau (Reference Jérôme, Mongrain and Nadeau2022) develop a “synthetic” model to predict the outcomes of French presidential elections (combining it with the SUR approach). In a pioneering turn, Dufresne, Jérôme, Lewis-Beck, Murr, and Savoie (Reference Dufresne, Jérôme, Lewis-Beck, Murr and Savoie2022) test the utility of citizen forecasting whereby respondents provide voter expectations rather than voter intentions. Graefe’s (Reference Graefe2022) contribution represents the first attempt to use the PollyVote approach to forecast national elections in France, including the combination of forecasts and the addition of a panel of French experts. Facchini (Reference Facchini2022) tests an innovative approach that involves the use of political-party popularity scores to predict the results of the next French presidential election.

In addition to forecasting models per se, the symposium invited two articles that focused more on electoral process and institutions. Since the reform of the five-year term, the legislative elections immediately follow the presidential elections, which means that the majority that will govern France actually arises from a “four-round election,” as characterized by Parodi (Reference Parodi2007). Dolez and Laurent (Reference Dolez and Laurent2022) used forecasting results of the French legislative contests to foretell whether the new president-elect could rely on a majority in the French National Assembly. Berg, Gruca, and Rietz (Reference Berg, Gruca and Rietz2022) discuss something never before tried in the French presidential-election context: How and under what conditions can the Iowa Electronic Market approach be applied to forecasting 2022?

LOOKING AT THE RESULTS: A SHIFT TOWARD THE RIGHT?

Table 1 is a systematic review of the projections provided by the different forecasting models (as of the time of writing in February 2022) to highlight points of convergence and divergence. The first important finding is that the models agreed, concluding that incumbent President Emmanuel Macron should finish first in the first round with a significant lead over his closest opponent. Indeed, the average forecast for a sitting president in the first round is 25.5% of the votes, which gave him a significant lead (according to the models) of 5.9 percentage points over his most serious competitor. All of the models also predicted a clear victory for the incumbent in the second round, regardless of who his opponent was. Given the forecasts presented in table 1, it therefore would seem surprising if the Centrist candidate Macron were not only excluded from the second round of the 2022 presidential election but also suffered defeat in it.

All of the models also predicted a clear victory for the incumbent in the second round, regardless of who his opponent was.

Table 1

Summary of the 2022 French Presidential Vote Forecasts

Notes: Results for Unweighted Average 1 are calculated from models predicting a Macron–Pécresse contest in the second round; results for Unweighted Average 2 are calculated from models forecasting a Macron–Le Pen duel. Results from the first model (Bélanger, Feitosa, and Turgeon Reference Bélanger, Feitosa and Turgeon2022) are not included in the calculations because its methodology is somewhat different from the other models.

The model forecasts also converged on several other points. All of the models predicted that the Left would underperform in the presidential election. None of the models predicted a duel including a left-wing candidate in the second round. The Left therefore would be excluded from the decisive round of the French presidential election for a second consecutive election—another first in the history of the Fifth French Republic. Thus, it appears that the French Left had not recovered from its 2012 collapse, following the presidency of François Hollande. The “desert crossing” for the French Left was not yet over. The political dynamic in this country would continue to be dominated in 2022 by the opposition between candidates of the Center or of the Right on the political spectrum.

Another point of convergence in the forecasts also was notable. None of the models predicted that Zemmour would reach the second round. Thus, the support that this candidate garnered seemed insufficient to allow him to further shake the French party system, which apparently has evolved from a traditional Left–Right divide to a quadripolarization of the political game: Left, Center, Classical Right, and Extreme Right. That said, the presence of this polemical candidate could have weighed heavily on the identity of Macron’s opponent in the second round, depending on whether he drew more support from the Classical Right candidate, Pécresse, or the Far Right candidate, Le Pen. This element of uncertainty is reflected in the forecasts presented in this symposium: three models predicted that Le Pen would make it to the second round, whereas two models concluded that the Classical Right candidate, Pécresse, would advance to the second round.

In any case, the forecasting models discussed in this symposium predicted a significant shift of the French electorate to the Right. This was reflected in two ways: (1) as noted previously, by the exclusion of a left-wing candidate from the second round for a second consecutive election; and (2) this shift could manifest in a possibly “historic” performance by Le Pen, if she reached the second round. Admittedly, the models predicted a clearer victory for Macron in the second round against Le Pen (58.9%) than against Pécresse (53.5%). However, Macron’s victory over Le Pen might not be as decisive as in 2017 insofar as the models allowed for the possibility that the Far Right candidate could cross the 40% threshold of support in the second round.

…the forecasting models in this symposium predicted a significant shift of the French electorate to the Right.

Only the results of the presidential election could confirm the validity of the conclusions that emerged from the forecasts presented. However, it seemed that their richness demonstrated that electoral forecasting in France has matured and, after being inspired largely by the work of other democracies, the French case now could enrich the electoral forecasting work in those other democracies. It is from this perspective that this unique symposium was conceived.