A new form of anti-government resentment? Making sense of mass support for the Yellow-Vest Movement in France

ABSTRACT

Anti-elite and anti-political resentment have become a permanent feature of political life in many if not most contemporary democracies, leading to support for populist parties, systematic anti-incumbent voting, and new types of movements, such as the Yellow Vests protests that shook France in 2018–2020. The aim of this paper is to explain the unusual popular support they mobilized. Going beyond the somewhat tautological “populist” label attached to the movement this paper proposes a class-based explanation. Using original data from a survey run after the European Elections of 2019, it shows that social precarity, combined with a lower /working class position, is the main driver of affinity with the YV. The movement is disproportionately supported by the most insecure segments of production workers (mostly men) and service workers (mostly women), giving an identity to those who feel excluded and not represented by mainstream parties and unions. This disaffected “precariat” can be seen as a magnifying glass of the crisis of political representation affecting most Western democracies. A reservoir of discontent that is here to stay and that the economic and political impact of the Covid-19 pandemic could revive.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Elabe survey for BFM TV, conducted 12–13 November 2019. Respondents were asked if they support the movement, have sympathy for it, oppose to it, feel hostile, or indifferent. 55% expressed support or sympathy in November 2019, compared to 73% on average in November 2018 (https://elabe.fr/gilets-jaunes-1an-apres/).

2 See for instance Nuit Debout (“Up all Night”) against labor law reform or the ZAD (“Zone to Defend”, against the construction of a new airport) near Nantes.

3 The left-hand panel is based on 10-points scales. The bars indicate the share of respondents having indicated 7 or more out of 10. The right-hand panel is based on yes/no questions: the bars indicate the share of those respondents that answered “yes”.

4 INSEE, PCS: Socio Professional Occupations and Categories Classification, 2003. See Appendix.

5 We thank Daniel Oesch for his useful comments on the specificity of the French categorization.

6 With pensioners and unemployed recoded in their previous occupation.

7 Originally developed for the French Social Security medical centres to detect health problems in socially vulnerable populations, it consists in eleven simple yes/no questions covering financial difficulties, housing conditions, type of health insurance, support from family and friends, cultural activities, etc. The indicator stems from a large-scale survey conducted in 1998, with a host of socioeconomic indicators. A correspondence analysis showed all items loaded on a first factor of precariousness. The 11 items at the time most correlated to the factor, and explaining 91% of the variance, were selected for the EPICES score, and weighted accordingly. The questions (Appendix 2) take little time, are easily understood, the non-response rate is negligible.

8 On the importance of this subjective dimension, or perceived poverty, as indicator of economic and social insecurity see Duvoux, 2010; Papuchon & Duvoux, 2019.

9 In our panel, the mean score Epices is 23.1, the median value 19.2. See Appendix 5 for more information.

10 As for “affinity with the YVM”, we rescale and center the resulting indicator to contain it between 0 and 1.

11 We have rerun all models with a binary “outsiderness” indicator instead of precariousness (see Appendix 9). As expected this indicator, while poiting into the same direction, does less well in terms of explanatory power, as it misses the social-isolation question, at the heart of our precariousness index.

12 Appendix 8 presents alternative specifications of political self-placement, testing in particular for “radicals”. These alternative models do not fundamentally challenge the conclusion that supporters are more left-wing than rightwing. As a matter of fact, right-wing “radicals”, placing themselves between 9 and 11 (out of 11) have lower probability than the rest of the population to show affinity with the YVM (cf. model 3). When controlling for left-right, while testing for radicality, it seems that radicals of both sides are more likely to support the YVM, though (cf. models 5 and 6).