“Don’t worry, be happy (and the vote out the incumbent): economic anxiety and incumbent support”

ABSTRACT

In recent years, there has been a substantial upsurge in the cost of living and income inequality across established democracies. These shifts, intensified by recent public health crises and growing political instability, have subjected voters to an extraordinary mix of social, political, and economic uncertainty. Despite widespread doubt about the future, a consistent finding in the economic voting literature highlights that voters’ evaluation of the past economy strongly influences their decision to either reward or penalize incumbents. Additionally, these studies find that voters’ forward-looking evaluations of personal economic conditions and general economic expectations often yield weak or negligible effects. These patterns motivate an important empirical question – why do future-oriented economic evaluations fall short in yielding a more substantial effect on incumbent support amid considerable personal struggle and uncertainty about economic conditions ahead? Grounded in appraisal theories of emotion, our study suggests that these consistent results may stem from existing measures inadequately capturing the multi-dimensional and affective nature of voters’ future economic concerns. Using data from the Canadian province of Ontario, we explore the extent of voters’ economic anxiety, identify the factors influencing these sentiments, and show that economic anxiety weakens support for incumbents in provincial and municipal elections.

Disclosure statement

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

Notes

1 Vasilopolous and colleagues (2019) acknowledge that the terms “anxiety” and “fear” are often used synonymously in the emotions and voting literature and suggest that individuals do not differentiate between these two emotions practically. The argument is evident in various empirical examples of anxiety and fear leading to very similar behavioural outcomes (e.g., Brader 2005; Parker and Isbell 2010).

2 Several of these studies use language such as economic insecurity, which Rebechi and Rohde (2023) define as “the anxiety produced by a lack of economic safety (e.g., an inability to obtain protection against subjectively significant potential economic losses”) (703). As a result, the term often indicates how a single indicator (or a select few), namely, job insecurity, feeds into voters’ economic anxiety and increases their willingness to support populist leaders and parties.

3 We note that using a general measure of prospective concern in our modelling model choices is somewhat redundant – as will be discussed in the Data and Methods section of the paper.

4 We conceive of this approach as similar to the “bankers” model developed by MacKuen, Erikson, and Stimson (1992). The model suggests voters act like “bankers,” often relying on both past economic performance and information about the future to develop carefully crafted, rational expectations regarding future economic performance.

5 This distinction builds off and adds to previous work by Anderson and Pontusson (2007). The study distinguishes between affective and cognitive economic insecurity in workers’ responses to social protection programs. Among their conclusions is the need to decompose different aspects of economic insecurity into affective and cognitive dimensions. We understand our distinction in this paper between cognitive evaluations of economic conditions and emotional responses to those conditions similarly.

6 We provide more information about these three coping strategies discussed by Folkman (2020). Specifically, people often use problem-focused coping when individuals believe a situation is controllable. Examples of this approach may include using humor, seeking social support, or blaming others. Individuals are likely to use emotion-focused coping when they must come to terms with something (have to accept). These strategies may involve information gathering, negotiating, or drawing on experience. Lastly, people likely deploy meaning-focused coping when facing chronic or largely unsolvable situations. For example, Folkman (2020) suggests that facing a serious illness would encourage meaning-focused coping.

7 The provincial election day was June 2nd, 2022. The pre-election wave field dates for the Ontario survey spanned May 18th to June 1st, 2022. These data contain a sample of 4,007. The post-election survey was fielded from June 3rd to June 13^th, 2022, and had a sample of 3,188.

8 The Ontario municipal pre-election survey field date spans September 29^th to October 23rd, 2022. The post-election survey field dates spanned October 25th, 2022, to November 7th, 2022. These data had a sample of 4,045 in the pre-election wave and 2,982 in the post-election phase.

9 In the results presented in the Online Appendix, we separate the economic stability item from the more egocentric items in the economic anxiety index (e.g., job security, retirement funds, cost of education, and ability to purchase a home). We note that the results of the index of egocentric items are consistent.

10 We provide more information about the wording and reliabilities for the Likert items used to comprise the economic anxiety measure within the Online Appendix.

11 The Consortium on Electoral Democracy’s provincial election studies uniformly tailor the retrospective sociotropic economic evaluation questions to the provincial context and do not ask about the national economy. The municipal election study asks about the local economy to facilitate cross-level comparison within the province of Ontario.

12 We note the long-running concern about endogeneity in economic voting models – notably, prior partisanship strongly shapes the development of economic perceptions used in economic voting analyses (e.g., Evans and Andersen 2006). The literature on this debate is mixed, with some finding that the relationship between partisanship and perception is not very strong (e.g., Lewis-Beck, Martini, and Kiewiet 2013) and others finding that there is a strong relationship (e.g., Evans and Andersen 2006). Finally, some work shows that partisanship has some influence in shaping economic perceptions, but these perceptions respond sensibly to real-world economic conditions (Anderson 2020). Regardless of this debate, all of our economic anxiety models include controls for the traditional measures of economic conditions and partisanship, which we believe alleviates many of these concerns.

13 We measure individual economic anxiety scores using Bayesian factor analysis. This measure is identical to factor scores from a traditional factor analysis (r = 0.99). However, it allows us to avoid listwise deletion of respondents who had partial missing responses for the economic anxiety questions (which would be necessary if we had a traditional factor score). We provide more detail on this measure in the Online Appendix.

14 Table 2 displays these model results and standard errors (in parentheses) in the Online Appendix.

15 NDP stands for New Democratic Party. The NDP is a social democratic party and currently the Official Opposition in Ontario.

16 More specifically, Figure 2 represents the expected change in the predicted probability of voting for the incumbent associated with a shift from the minimum to the maximum level of incumbent support in percentage points. We provide full regression tables for the incumbent support models in the Online Appendix.

17 Our inclusion of controls for respondent ideology and partisanship in our analysis supports our belief that incumbent ideology is not the driving force behind the results observed for the economic anxiety variable. Additionally, while factors such as ideology, interest in politics, partisanship, and a few sociodemographic factors are significant in the vote models, our results suggest that economic anxiety has an independent and negative effect on incumbent support.

Additional information

Funding

This work was supported by the Social Sciences and Humanities Research Council of Canada (grant number 435-2021-0603).