How did the pandemic affect turnout in French elections in March 2020? Elsa Leromain (LSE) and Gonzague Vannoorenberghe (Université Catholique de Louvain) find that participation fell, especially in towns more exposed to COVID, those with a high proportion of people over 60 or where the far right candidate Marine Le Pen had come first in the 2017 presidential election. Why might this be?
The first round of the 2020 French local elections took place on 14 March, when the country was just starting to grasp the severity of the COVID-19 pandemic. All political parties expressed their support for maintaining the election but many voters were worried, or plain scared by the health situation. The nationwide turnout was 44.75%, 18.8 percentage points lower than the previous election in 2014.
The COVID crisis is remarkable in that it made voting potentially life threatening, both for oneself and for others. While common in many countries, this situation is unknown for most voters in advanced democracies. The timing of the French local election also makes it of particular interest. Data on infection at the time were unreliable, disease transmission was poorly understood and protective equipment (sanitiser, face masks) was in short supply. These circumstances provide a unique opportunity to study the factors that made voters respond to an unusual increase in the risk of voting.
Our paper establishes that towns more exposed to risk factors linked to the pandemic saw a larger drop in turnout. We construct a group of towns identified as clusters in the media before the election and of towns connected to these clusters, which we measure by commuting patterns (or alternatively by geographic proximity). Together, these form our group of relatively more exposed towns given the information at the time. We show that the drop in turnout between 2014 and 2020 was significantly higher in exposed towns, in particular if they also had a high proportion of people aged 60 or above – a known risk factor at the time. We also show that, conditional on our covariates, exposed towns had similar changes in turnout in previous local elections as the rest of the country, whether their population is old or not.
Taken together, this suggests that voters did respond to factors influencing the known risk of voting. We confirm the risk interpretation by also looking at data on cinema admissions between early March 2019 and early March 2020. We show that admissions went down more in cinemas close to identified clusters. Cinema visits are independent of political considerations but should be correlated with the perceived risk of going out. The similarity of our results for electoral turnout and cinema admissions points to perceived risk as the main driver of the shift in turnout.
We also look at the role of political affinity in explaining how people perceived the risk of catching COVID. Towns in which Marine Le Pen, the far-right candidate, came first in the 2017 presidential election saw a stronger decrease in turnout, in particular among exposed towns. One possibility is that far-right voters in France have on average a higher degree of risk aversion, and respond more to the same increase in risk than other towns. Another is that they were more prone to lose faith in the political system due to COVID, or were less attached to voting and responded more to a change in risk perception.
To further disentangle the channels, we show that cinema admissions went down more in cinemas close to identified clusters, particularly if the surrounding towns were far-right towns. We take this as evidence that the subjective perception of risk can at least partly explain the differences in turnout there.
Our results speak to the literature linking political preferences to the attitude towards risk or fear. Campante et al. (2020) show that fear of the Ebola outbreak before the 2014 midterm elections in the US did affect voter turnout and that Republican candidates were strategically exploiting this fear. Makridis and Rothwell (2020) find that, in the US, Republicans are less likely to socially distance, to self-isolate, and to wear masks. Adam-Troian et al. (2020) argue that the fear of the virus made voters turn to more conservative parties in the French municipal elections, while Fernandez-Navia et al. (2021) that voters turned more to nationalistic parties in the Basque country in July 2020.
This post represents the views of the authors and not those of the COVID-19 blog, nor LSE. It is an edited extract from Voting under threat: evidence from the 2020 French local elections, a Centre for Economic Performance Discussion Paper.