It’s the Moral Economy, Stupid! Anger Toward Economic Inequality and Populist Voting

Abstract

The present study sets out to investigate the mechanism by which ordinary citizens are moved to anger toward economic inequality, and its political consequences. Since anger toward economic inequality is triggered by the perception that core distributive norms have been violated, it is argued that this emotional response mainly benefits political parties who defend redistributive values and establish clear responsibility for their violation. Results from twenty-six ISSP countries provide strong evidence that angry citizens are more inclined to vote for economically progressive populist parties and for economically progressive pluralist parties than for parties taking conservative stances on economic issues. This study thus lends empirical support to the assumption that populist parties attract angry voters by speaking the language of moral economy.

Keywords:

• Anger
• fairness perceptions
• populist voting
• wealth inequality
• welfare image

Notes

1 One specific feature of anger is that it is a negatively-valenced emotion, felt when one is frustrated or irritated. Emotions with a negative valence lead to negative evaluations of the persons, groups or entities that elicited them.

2 Admittedly, anger is both an emotion and a personality trait. While personality traits are internally based psychological characteristics that define who we are as individuals and tend to remain stable over time (Mondak 2010), emotions are best conceptualized as momentary feelings. However, emotions attached to politics are both trait-based and state-based (Marcus et al. 2000).

3 The latest data set version (14.10.2022): ZA7600 (v3.0.0), doi:10.4232/1.14009.

4 To be sure, I checked for multicollinearity with variance inflation factors (VIF) statistics. No multicollinearity issues were found with variables measuring anger toward inequality, fairness perceptions and anti-elite attitudes in models 2, 3 and 4 (i.e. VIF values smaller than 2). Regarding model 1, also note that anger toward inequality and fairness perceptions display a moderate correlation (0.2615; p < 0.05).

5 26 countries were included in the analyses: Australia, Austria, Bulgaria, Chile, Croatia, Czech Republic, Denmark, Finland, France, Germany, Iceland, Israel, Italy, Japan, Lithuania, New Zealand, Norway, Russia, Slovenia, Suriname, Sweden, Switzerland, Taiwan, Thailand, United States and Venezuela. Great Britain, the Philippines and South Africa were not included because voting information is lacking. The ISSP data for the 26 countries contain 36274 individuals. Voting preferences are available for 56.7 percent of the sample. Using the GPS classifications, effective voting was coded for 20011 individuals, out of the 20552 who expressed their voting preferences in the ISSP. Individuals who did not vote in the last general election were coded as ‘no vote’ (n = 6629, representing 25 percent of the sample). Respondents who did not report their voting behavior (n = 682) or refused to answer (n = 396) were excluded.

6 Models were estimated with RStudio (RStudio Team 2021), using the function lmer for model 1 and mblogit for models 2 to 4. Results are displayed in Supplementary Appendix 2. Note that independent variables have low nonresponse rates (don’t know and no answer categories below 5 percent), with the except of household income (19 percent of missing values). For this reason, and since the disenfranchised are known to have specific voting behaviors, a category was computed for respondents who did not report their household income.

7 It has been demonstrated that anger leads to demobilization among individuals with low political efficacy (Magni 2017). Lower education, lower income, and lower social position are closely related to low efficacy (Wu 2003). Therefore, although the ISSP 2019 Social Inequality module did not include an item for political efficacy, it is reasonable to assume that the sociodemographic variables included in the models are appropriate proxies to control for political efficacy.

8 The issue of influential cases at the country level is addressed in Supplementary Appendix 3.

9 Predictive margins were generated with the function predmarg from the mpred package.

10 Additional analyses with ‘no vote’ as the reference category indicate that anger significantly increases the propensity to vote for economically-left pluralist parties (1.08; p < 0.01) and for economically-left populist parties (1.06; p < 0.01), while decreasing the propensity to vote for economically-right pluralist parties (0.90; p < 0.01) and for economically-right populist parties (0.96; p < 0.05).

11 The odds ratios of voting for economically-left populist parties significantly decrease with HDI; meaning that left-wing populist parties are more successful in countries with low HDI. For an analysis of how macrolevel variables influence votes for radical left and radical right parties in Europe, see, e.g. Rooduijn and Burgoon 2018.

Additional information

Notes on contributors

Frédéric Gonthier

Frédéric Gonthier is Professor of Political Science at Sciences Po Grenoble, University of Grenoble Alpes. His work focuses on welfare attitudes and the linkage between value systems and party systems in a comparative perspective. It has appeared in journals such as Acta Politica, European Journal of Political Research, French Politics, Party Politics or Political Psychology.