Are “bad” jobs bad for democracy? Precarious work and electoral participation in Europe

Azzollini, Leo; Macmillan, Ross

doi:10.3389/fpos.2023.1176686

ORIGINAL RESEARCH article

Front. Polit. Sci., 01 June 2023

Sec. Political Participation

Volume 5 - 2023 | https://doi.org/10.3389/fpos.2023.1176686

This article is part of the Research Topic Agents of Political Socialization in the 21st Century View all 7 articles

Are “bad” jobs bad for democracy? Precarious work and electoral participation in Europe

$\r\nLeo Azzollini,,$ Leo Azzollini^1,2,3^*

Ross Macmillan⁴

¹Social Sciences Division, Department of Sociology, University of Oxford, Oxford, United Kingdom
²Social Sciences Division, Department of Social Policy and Intervention, University of Oxford, Oxford, United Kingdom
³Nuffield College, Oxford University, Oxford, United Kingdom
⁴Department of Sociology, Faculty of Arts, Humanities and Social Sciences, University of Limerick, Limerick, Ireland

Introduction: While socioeconomic inequality in voting has been central for research on electoral participation, recent years have seen radical changes in labor relations. The key issue is increasing prevalence of precarious work, involving dimensions as non-indefinite tenure and limited control over work activities. While occupations traditionally were the locus of political socialization, there is scarce research connecting occupational uncertainty to electoral participation. To fill this void, we develop a framework that connects the multiple dimensions of precarious work to electoral participation.

Methods: We test these ideas using data from 32 countries from the European Social Survey (2008–2018).

Results and discussion: Results indicate that work precarity is both strongly connected to traditional indicators of SES and has large, independent effects on probability of voting. We corroborate these results with heterogeneity analyses across countries. Findings show how precarious work heightens socio-economic stratification in electoral participation, undermining the universality of the right to vote and the health of democracies.

1. Introduction

Electoral participation is the cornerstone of democracy with the question of who does and does not participate in formal politics intimately connected to issues of representation, equity, and justice (Verba, 1995; Piven and Cloward, 2000; Dahl, 2008; Schröder and Neumayr, 2021). Not surprisingly, the issue of socioeconomic variation in electoral participation has been central (Lijphart, 1997). In research spanning several decades, the relationship between socioeconomic status (SES), be it social class, education, employment status, or income, and electoral participation has organized both theory and research (Wolfinger and Rosenstone, 1980; Powell, 1986; Jackman, 1987; Blais et al., 2004; Franklin, 2004; Blais, 2006; Leighley and Nagler, 2013; Smets and Van Ham, 2013; Jeannet, 2022).

At the same time, some have stated a need to move “beyond SES” (Brady et al., 1995), with an increased focus on cognitive and socio-psychological resources. A recent stream of research has intersected these two bodies of literature: after the Great Recession, several studies have focused on the political consequences of labor market disadvantage (Brand, 2015; Häusermann et al., 2020), focusing mainly on unemployment experiences through political efficacy (Emmenegger et al., 2015, 2017; Azzollini, 2021, 2023; Österman and Lindgren, 2021; Wiertz and Rodon, 2021; Österman and Brännlund, 2023), but also on further dimensions of labor market outsiderness (Rueda, 2005; Lindvall and Rueda, 2014; Rovny and Rovny, 2017), such as occupational risk (Häusermann and Schwander, 2012). However, this stream of research has scarcely addressed the role of a broader labor concept: that of precarious work (Polanyi, 1944).

While far from a new concept, precarious work has re-emerged after the radical reconfigurations of Bretton Woods economic system in the 1980s (Kalleberg, 2009), and encompasses a mixture of non-permanent contracts, low job control and organizational influence, cyclical unemployment, and financial vulnerability (Burgess and Campbell, 1998; Kalleberg, 2011; Kalleberg and Vallas, 2018), with dire effects on several socio-demographic outcomes beyond the labor market (Schneider and Harknett, 2019; Macmillan and Shanahan, 2021, 2022). By highlighting the “uncertain, unpredictable, and risky” character of precarious work (Kalleberg, 2009, p. 1), we argue that precarious work has important consequences for electoral participation, beyond already established predictors of the latter such as current unemployment (Marx and Nguyen, 2016) and education (Smets and Van Ham, 2013). Building on the classic literature envisaging occupations as the key locus for political socialization (Alford, 1967; Lipset and Rokkan, 1967; Evans et al., 1991), our core argument is that uncertainty and disadvantage in the occupational realm hamper political socialization and the formation of a clear identity, therefore leading to political marginality. In essence, if in the past “industrial citizenship” was crucial for “political citizenship” (Standing, 2011), precarious work links together industrial and political “denizenship”, i.e., inhabiting the socio-political space without exercising membership fully. Therefore, assessing the role for precarious work is important to understand the new pathways through socio-economic inequalities translate into inequalities in electoral participation.

2. Materials and methods

2.1. The origins of precarious work

The last decades of the 20th century saw the emergence of new logics of labor that emphasized flexibility above all else in the relations of production (Rodgers and Rodgers, 1989; Kalleberg, 2009; Standing, 2011). Emphasis on flexible arrangements was largely unconstrained and flexible labor contracts produced temporary jobs and “casual” employment. Employers are increasingly “brokers” of labor that staff firms depending upon short-term needs. Even within firms, workers were expected to be flexible and do different tasks depending upon ever changing need. Modes of compensation changed from salaried positions or predictable wages to piece-meal compensation based on short-term contracts or tasks completed. Indeed, the new relations of labor were increasingly ephemeral, unpredictable, and often insufficient for the management of everyday life (Kalleberg, 2009; Standing, 2011).

Traditional notions of SES and social class that frame much of the prior work on electoral participation assumed, quite rightly in most cases, that work was reasonably stable in form and largely permanent at least within generations. People had a job and that job, or a very similar one, was something that they would do over the majority of their working lives. Given this, type of occupation or broader configurations of work were used to operationalize variation in SES (e.g., Hauser and Warren, 1997) or class position (e.g., Erikson and Goldthorpe, 1992).¹ Political scientists could then ask questions about SES or class differences in electoral participation and develop theories about the political consequences of variation in SES or class position as a way of understanding the nature of political inequality and its implications.

In contrast, precarious work is more “informal” with few of the social contract relationships typical of the traditional working class (Kalleberg, 2009, 2011). Some suggest that a lack of control over labor is the key feature (Standing, 2011). While limited control stretches back to the dawn of the industrial revolution, contemporary analysts suggest much greater scope. Precarious jobs are characterized by insecure employment, in subordination to labor brokers rather than producers (Kalleberg, 2009, 2011). Alternatively, it involves work for some multifaceted and multinational entity of which workers have little knowledge. Precarious work also involves increased job insecurity, a lack of stable, continued employment (Rodgers and Rodgers, 1989). For some, duration of work is known; for others, employment may cease with little warning. In either situation, stringing together jobs will involve a complex combination of efficacy, planning, and luck. As such those in precarious work are likely to have significant gaps in employment and precarious workers are typically subject to cyclical unemployment (Standing, 2011). Precarious workers often try to mitigate the situation by having multiple “part-time” jobs or by having “side gigs” or “side hustles” (Worth, 2018).

Limited control also stretches to tasks. While the traditional proletariat lacked control over what they did, they typically did the same types of tasks day after day and this may have been fundamental to occupational identification and class affiliation (Kohn and Schooler, 1983; Weeden and Grusky, 2005). In contrast, precarious workers tend toward being “jacks-of-all-trades” where a multiplicity of skills or basic soft skills has greater value in the modern marketplace. Lack of control also applies to prospects for social mobility. In traditional work, hierarchies of positions were well established and procedures for movement between ranks are known. For modern precarious workers, positions are known to be temporary and disconnected from formal organizational structures. As such, social mobility involves not work but attainment of stable, long-term employment (Wright, 2016).

Finally, all of the above culminates in financial vulnerability whereby those in precarious work struggle to manage expenses of everyday living because of low compensation, gaps in employment, or a combination of the two. In several countries, the vast majority of people have very little in the form of liquid savings (UK Office of National Statistics, 2022) and the average unemployed person can last a couple of months at most without either income support or incurring debt (US Survey of Consumer Finances, 2016). For those in precarious work, financial vulnerability is endemic, cyclical, and a feature of everyday life. Indeed, considerable bodies of work highlight how higher levels of socio-economic vulnerability are strongly associated with lower levels of electoral participation (see Smets and Van Ham, 2013 for a review), be it in the form of lower levels of income decile (Solt, 2008), or unemployment and broader labor market hardships (Marx and Nguyen, 2016). We articulate in detail how each of these dimensions of vulnerability is associated with turnout in the subsection below.

2.2. Precarious work and electoral participation

While considerable work addresses the negative socio-economic consequences of precarious work, especially on health (Kalleberg, 2000; Schneider and Harknett, 2019; Macmillan and Shanahan, 2021, 2022), there is limited empirical work on the consequences in terms of electoral participation. This is particularly evident if compared to the considerable streams of research on social class and electoral participation (Evans and Tilley, 2017; Lahtinen et al., 2017; Jeannet, 2022), with the key pattern being the relationship between working class identity and lower propensity to vote. Indeed, the few articles addressing the issue have focused on the relationship between precarious workers and the traditional working class (Standing, 2016; Wright, 2016) and the broad implications for political orientation (Johnson, 2016), without focusing on specific features of precarious work.

Here, we address this gap by following previous research on precarious work and other socio-demographic outcomes (Macmillan and Shanahan, 2021, 2022): we focus on five key characteristics of precarious work, and articulate why they should influence electoral participation. We do so for each characteristic and then for the cumulative experience of precarious work.

Consider first the issue of non-indefinite work. Research on political socialization has long viewed work as a locus of political identity (Alford, 1967; Lipset and Rokkan, 1967; Evans et al., 1991), with workplaces constituting also the place where industrial and political coordination occurred (Radcliff and Davis, 2000). Here, prior work has emphasized type of work as producing particular political alignments with a particularly salient example being links between manual labor, working class identity, and socialist/social democratic politics (Evans, 2000). Yet, recent research has highlighted that the traditional working class may lead to different outcomes in different countries (Evans et al., 2022), with British and American working classes being less likely to vote (Evans and Tilley, 2017; Jeannet, 2022), while in Denmark it is still associated with pro-redistributive preferences (Evans et al., 2022).

Yet, the working class continues to have important ramifications in terms of socialization, with origin class shaping concurrently with destination class both participation (Jeannet, 2022) and attitudes (Paskov and Weisstanner, 2022). Theoretically, we argue that this does not hold for precarious work, precisely due to its temporary character: holding temporary contracts, usually across different sectors, hampers the formation of a precise occupational identity (Standing, 2011), with potential ramifications for political identity. Ephemeral work prevents one from seeing oneself as a particular type or class of worker, as well as undermines the perception, by co-workers, by friends, by family, that one is a type of worker or member of a well-defined social class, which would (a) anchor oneself to a position in society and (b) allow continued exposure to those agents which would traditionally socialize politically those in a specific position, as floating between jobs would hamper the time and opportunity to forge social ties that would foster a particular identity. Without a firm occupational identity, political preferences are unlikely to gel, and electoral participation should diminish.

A second connection to electoral participation comes from control over work and organizational influence. Classic work in social psychology shows that the type of work that people do translates into variation in “occupational self-direction,” a sense of agency and efficacy for occupational tasks (Kohn and Schooler, 1983). Occupational self-direction further generalizes into self-directed orientations to self and society (Kohn and Schooler, 1983), with lack of control over work and organizational influence potentially translating into political abstention. More specifically, Acevedo and Krueger (2004) argue that belief in personal relevance for electoral outcomes is a key determinant of voting. Importantly, belief in personal relevance has conceptual affinity with generalized perceptions of agency (e.g., Emirbayer and Mische, 1998) that sit at the core of ideas around self-directedness (e.g., Kohn and Schooler, 1983). Other aspects of agentic personality are also associated with greater likelihood of voting, including higher cognitive ability (Denny and Doyle, 2008), diligence (Smets and Van Ham, 2013), lower ambivalence/alienation, and higher political efficacy (Pattie and Johnston, 1998). In general, powerlessness in the workplace translates into perceived powerlessness in other realms of social and political life. If individuals lack basic control on the everyday tasks they perform or on the conditions of their organization, they may consider it pointless to participate in elections and to try to influence the much larger-scale political system.

A further dimension of precarious work is unstable employment or the presence of employment gaps. For many (e.g., Kalleberg, 2009), the defining feature of precarious work is its non-indefinite character. This, virtually by definition, increases the likelihood of gaps in employment. A recent stream of research highlights how unemployment experiences decrease political engagement and electoral participation (Emmenegger et al., 2017; Rovny and Rovny, 2017; Azzollini, 2021; Giustozzi and Gangl, 2021; Österman and Lindgren, 2021; Österman and Brännlund, 2023), with the key mechanisms being higher social stigma (Laurence, 2015), lower social trust (Mewes et al., 2021; Azzollini, 2023), and lower political efficacy (Emmenegger et al., 2015). Considering however unemployment together with precarious work, we get additional insights in terms of political socialization (or lack thereof). Frequent entry and exit from the labor market prevents the opportunity to form the long-term stable relationships necessary for socialization (Rosenstone, 1982; Gray and Caul, 2000), but may also undermine coordination by fostering competition between precarious workers, competing for a coveted long-term stable job (Standing, 2011), decreasing further job loss risks. Such competition again undermines the formation of a collectivity and a collective identity that fosters political socialization, and in turn boosts participation (Radcliff and Davis, 2000).

Finally, financial insecurity should also undermine electoral participation. Financial insecurity is endemic to precarious work. Even when employed, precarious work is typically associated with lower pay, as well as poorer benefits including support for health care and pensions (Kalleberg et al., 2000). At the same time, financial insecurity also reflects vulnerabilities that accrue from gaps in employment. Evidence from the US Survey of Consumer Finances (2016) indicates that the average person has savings for approximately 1–2 months of support. Likewise, the UK Office of National Statistics reports that forty percent of adults have < £1,000 in savings which would last them about 1 month. Although largely at a conceptual level, research on social stratification and voting emphasizes financial stress as the link between low SES and decreased electoral participation. Rosenstone (1982, p. 26) for example argues that inequality in electoral participation reflects the focus of those of lower status being focused on “keeping their body and soul together” and caring for their families. Particularly in “post-material” societies (Inglehart, 1997), political discourse increasingly reflects what Maslow (1943) would deem to be “higher order” values. Yet those experiencing financial insecurity are likely to be focused on “lower order” needs, such as securing shelter and subsistence, and hence may have less interest in politics and political issues that emphasize other priorities. As a corollary, financial vulnerability may also undermine trust in political institutions by fostering perceptions that political institutions are unresponsive and ambivalent to the needs of precarious workers (Lijphart, 1997). Financial vulnerability in the end should uniquely undermine electoral participation.

Four further issues follow. First, the impact of the different dimensions of precarious work is likely cumulative, even if they operate independently in depressing electoral participation. As argued by Standing (2011), the multiple trappings of precarious work deprive individuals of an occupational identity within a clear hierarchy. Therefore, individuals experiencing multifaceted precarious work lack “industrial citizenship” (Standing, 2011), or membership in a defined social class, making very difficult to coordinate competing individuals into a voting bloc, as unions historically did for the proletariat (Lipset and Rokkan, 1967; Evans, 2000). Second, the consequences of precarious work should be independent of traditional indicators of low SES. These include family and respondent's social class, educational attainment, employment status, and income. As such, they represent both a new vector of socioeconomic inequality in electoral participation that is not easily subsumed under traditional conceptualizations of class and status (Standing, 2016; Wright, 2016). Third, markers of precarious work should add to the effects of traditional indicators. There is no reason to think that traditional markers of SES have declined in importance and hence the addition of indicators of precarious work should increase the overall scope of socioeconomic differentiation.

Fourth and finally, the relationship between the different dimensions of precarious work and electoral participation is likely to be heterogeneous across countries. Barbieri (2009) argues that variation in welfare state regime shapes the meaning of precarious work for individual workers. It does so in two ways. First, it shapes overall exposure to precarious work. Countries with more expansive social welfare are more likely to have labor regulations that limit both temporary work and increase control over work and employment. Countries that are more market oriented are less likely to intervene in labor relations. Second, welfare state regimes provide variable amount of extra-market supports on issues of low income, unemployment, health care, and pensions. Stratification research has speculated that the consequences of precarious work should be weakest in the strong welfare state contexts of Scandinavian countries and strongest in settings where market forces and market logics are more universal such as Anglo-Saxon and Eastern European regimes (Kim and Zurlo, 2009). Expectations are less precise for the hybrid market-welfare orientations of Continental countries or the familial-oriented Mediterranean countries. As prior work is ultimately quite vague, our goal is to establish how robust the association between precarious work and turnout is across countries with an eye toward whether welfare state regimes moderate effects.

With this background, the analyses that follow address three questions. First, how does precarious work relate to traditional indicators of SES? In answering this question, we flesh out the broader dimensionality of SES for the contemporary era. Second, how does precarious work influence the likelihood of voting? Here we focus on both individual indicators, as well as a cumulative index. Our broader hypothesis is that all the five dimensions of precarious work depress electoral participation, and we further posit a similar effect for the cumulative index of precarious work, ultimately determining empirically their relative strengths in terms of effect size. Finally, we assess generalizability of results across a range of welfare state regimes that are expected to alter aggregate exposure to precarious work and perhaps moderate its consequences. In sum, these analyses highlight precarious work as a critical aspect of SES and its implications for electoral participation and political inequality.

3. Data and methods

3.1. Data

The data we use comes from the European Social Survey (ESS), a biennial cross-national survey of attitudes and behavior established in 2001. The ESS uses cross-sectional, probability samples which are representative of all persons aged 15 and over resident within private households in each country. Given the focus of our research, our analytic sample is restricted in two ways. First, we only include those respondents who are eligible to vote in their resident country. Second, we only include respondents who reported some labor market activity over their life course and hence could answer questions about conditions of employment. The final sample consists of just over 132 thousand respondents from 32 countries spanning the years 2008–2018. Countries include Austria, Belgium, Bulgaria, Croatia, Cyprus, the Czech Republic, Denmark, Estonia, Finland, France, Germany, Great Britain, Greece, Hungary, Iceland, Ireland, Israel, Italy, Lithuania, the Netherlands, Norway, Poland, Portugal, Russia, Serbia, Slovakia, Slovenia, Spain, Sweden, Switzerland, Turkey, and Ukraine. The European context is a particularly promising arena for research given its importance as a sector of the global economy and its diversity with respect to labor market institutions and political systems. To ensure representativeness, we incorporate both design weights and probability weights, following best practices.

3.2. Analytical strategy

Our focal outcome is whether the respondent voted in the last national election in their country of residence (coded 0=did not vote; 1 = did vote). Our predictor variables fall into one of four categories. First, standard control variables to address unobserved heterogeneity include age (in years), gender (coded 1 = male; 0 = female), marital status (coded 0 = never married; 1 = currently married; 2 = separated or divorced), nativity (coded 0=foreign born; 1 = native born), and religiosity based on how frequently one attends church (coded 1 = never to 7 = everyday). Second, we include family SES given the long history in research on electoral participation (e.g., Wolfinger and Rosenstone, 1980; Powell, 1986; Jackman, 1987; Franklin, 2004; Blais, 2006; Smets and Van Ham, 2013). We capture this through a measure of the family social class based on the Erikson-Goldthorpe-Portocarero (EGP) schema that differentiates respondents based on whether their father worked in “professional/technical jobs”, “high administration”, “clerical work”, “sales work”, “service work”, “skilled labor”, “semi-skilled labor”, “unskilled labor”, and “farm labor” when there were 14 years old. We model effects based on a set of dummy variables indexing group membership (reference = professional/technical jobs). We also include a measure of parent's educational attainment that is the highest level of educational attainment of the respondent's mother or father ranging from “less than lower secondary” (coded 1) to “tertiary education” (coded 5). The combination of parental occupation and education has a long and valuable history in research on SES and its implications (Hauser and Warren, 1997).

We also control for traditional indicators of the respondent's SES. Social class of the respondent is measured through the ESEC schema developed by Rose and Harrison (2007), which is essentially a slightly more simplified version of the EGP schema (Ganzeboom, 2010). As the ESEC schema is not directly available in the ESS, we convert the ISCO-88 (iscoco) and ISCO-08 (isco08) measures of occupation to ESEC relying on the iscogen STATA package developed by Jann (2019).

Educational attainment is measured with the same metric as parent's educational attainment. Labor market position is the respondent's “usual” state of activity during the survey year that includes those “employed (coded 1)”, “in school (coded 2)”, “unemployed (coded 3),” “disability (coded 4)”, “retired (coded 5)”, and “other [not in labor force] (coded 6)”. Finally, poverty status is measured as a flag capturing those in the bottom decile on household income.

The final set of measures capture various dimensions of a precarious work. Employment contract differentiates those with contracts of indefinite duration, those with contracts of limited duration, and those with no formal contract (coded 1). Two further aspects capture the nature of control within one's job. Job control is indexed by how much they can “decide how your own daily work is/was organized? (ranging from 0 = “I have/had no influence” to 10 = “I have/had complete control)”. Accompanying this, organizational influence captures the degree to which respondents can/could “influence policy decisions about the activities of the organization” (ranging from 0 = “I have/had no influence” to 10 = “I have/had complete control)? For both measures, limited job control and limited organizational influence are indexed as respondents who fall in the bottom terciles on both measures.² Another dimension is the presence of gaps in employment. This is captured by an ESS question that asks respondents whether they had “ever been unemployed and seeking work for a period of more than 3 months” (yes coded 1; no coded 0). The final aspect measures financial vulnerability based on a question asking respondents whether they find it “difficult” or “very difficult” (coded 1; other coded 0) to live on one's income. To capture the overall extent of precarious work, we sum up all indicators to create a cumulative index score (ranging from 0 to 5). Our focus on a continuous or graded measure of precarious work is consistent with arguments that the precariat, at least in its current form, is not a fully developed social class but instead is a “class-in-the-making” that involves multiple dimensions (Standing, 2011, p. vii).

We further include controls for both general and specific features of country context. In the former respect, our initial models include country fixed effects that effectively control for all time-stable attributes of countries, including also year fixed effects to control for any common temporal trends that could bias estimates. This should effectively eliminate any bias associated with long-run political culture, system level effects, or cultural proclivities for participation. Accordingly, we cluster standard errors by country. In the latter respect, we explicitly examine differences across welfare state regimes (Esping-Andersen, 1990) given expectations that variation in social welfare should both limit exposure to precarious work and buffer its social and psychological consequences through market and extra-market supports (Barbieri, 2009). At minimum, the latter models assess generalizability of the relationship between precarious work and electoral participation across a range of political-economic contexts. Descriptive statistics are shown in Table 1.

TABLE 1

Table 1. Descriptive statistics (n = 132,443).

4. Results

The first stage of our work models extent of precariat status particularly in relation to traditional indicators of SES. Poisson regression coefficients are shown in Table 2 (generated using poisson in Stata 17). We estimate three models with the first model including basic socio-demographics, the second adding SES in family of origin, and the third adding indicators of respondent's SES. To start (see model 1), risk of precarious work declines by 0.3 percent per year (e^−0.003 = 0.997), is 11 percent higher among women (e^0.102 = 1.107) and 18 percent lower for those born in the country of residence (e^−0.193= 0.824). It is also 20 percent lower for those that are married (e^−0.229 = 0.795), and decreases by 2 percent for each additional day per week of attending church (e^−0.022 = 0.978). With model 2, there is further evidence of socioeconomic differentiation with respect to family of origin. There is a clear class gradient where risk of precarious work is higher among those from service (e^0.030 = 1.030), skilled labor (e^0.092 = 1.096), semi-skilled labor (e^0.188 = 1.207), unskilled labor (e^0.309 = 1.362), and farm labor (e^0.186 = 1.204). Incremental increases in parent's educational attainment further reduce risk of precarious work by 6 percent (e^−0.058 = 0.944).

TABLE 2

Table 2. Poisson coefficients: precarious work and socio-demographic variables.

Direct indicators of respondent's SES are also associated with risk of precarious work. Greater educational attainment for example reduces risk by approximately 6 percent (e^−0.057 = 0.945). Precarious work is also related to labor market position with increased risk among those in school (e^0.268 = 1.307), those currently unemployed (e^0.554 = 1.740), those with disabilities (e^0.396 = 1.486), and those otherwise outside the labor force (e^0.236 = 1.266). Low income is associated with 34 percent increased risk of precarious work (e^0.293 = 1.340). There is also a clear gradient in the association between social class of the respondent and risk of precarious work. In comparison to higher managerial/professional workers, risk is marginally higher among the lower managers (e^0.089 = 1.093), higher among those in lower supervision and technician work (e^0.226 = 1.254) and intermediate occupations (e^0.395=1.484), and substantially higher among those in lower sales and service (e^0.585 = 1.795), lower technical (e^0.663 = 1.940), and routine workers (e^0.736 = 2.088). Concomitantly, self-employed workers, both agricultural and non-agricultural, have substantially lower risk. Poverty status is also associated with somewhat higher risk of precarious work (e^0.293 = 1.340). An important caveat here is that we cannot know causal order for any of the contemporaneous SES variables, except for educational attainment, but it is still instructive that lower SES is strongly associated greater risk of precarious work.

Figure 1 provides further insight into the relationship between indicators of precarious work and traditional markers of socioeconomic position by plotting estimated precariat scores (from 0 to 5) by extreme values on the Erickson-Goldthorpe-Portocarero family class measure, respondent's educational attainment, respondent's labor market position, poverty status, and respondent's social class. The most striking finding is the difference in number of precarious work indicators for the most advantaged vs. the least advantaged members of the sample. For the latter group, those not poor, from a professional class background and destination, and with high educational attainment, the estimated number of precarious work indicators is slightly under 1 (0.93). In contrast, the most disadvantaged respondents, those scoring low on all five traditional indicators of SES, have estimated precarious work scores of almost five (4.4) which is the maximum on the index. In other words, maximal variation in socioeconomic position increases risk of precarious work five-fold. Again, we do not make causal claims, particularly with respect to the contemporaneous measures of employment, poverty status, and respondent's social class, but emphasize the fact that precarious work adds to and expands socioeconomic differences among individuals in ways not captured in prior work.

FIGURE 1

Figure 1. Marginal effects at the means with 95% confidence intervals, after model 3 in previous table. Data source: European Social Survey Multilevel (2008–2018).

The heart of our analyses is shown in Table 3 with further discussion of Figure 1. For purposes of space, the base model with select controls is shown in Appendix A. Here, being older, being native-born, being married, and being more religious is associated with greater likelihood of voting, while being separated or divorced is associated with lower likelihood. With respect to SES, there is a clear gradient for family social class with reduced likelihood of voting among those from “labor” backgrounds. At the same time, likelihood of voting increases with educational attainment, is lower among those unemployed, with a disability or living in poverty. There is also a strong social class gradient based on current occupation with linear decreases in electoral participation over the entirety of the index. Aggregately, these results echo the large body of work showing that various dimensions of SES undermine electoral participation (Wolfinger and Rosenstone, 1980; Powell, 1986; Jackman, 1987; Franklin, 2004; Blais, 2006; Smets and Van Ham, 2013).

TABLE 3

Table 3. Coefficients: probability of voting and precarious work, with controls.

Table 3 show effects for the various elements of precarious work, first with each indicator by itself (models 1 through 5) and then for the cumulative score (model 6). To start, those working with non-indefinite contracts have odds of voting that are 8 percent lower (e^−0.086 = 0.918) than those with indefinite contracts. Having limited control over one's work and having limited influence in one's workplace also undermine the likelihood of voting with odds around 8 percent lower (e^−0.088 = 0.916, e^−0.084 = 0.919, respectively). Those with gaps in their employment have odds of voting 14 percent (e^−0.137 = 0.872) lower than those without gaps and likelihood of voting is also lower for those who are financially vulnerable (e^−0.236 = 0.790). Aggregating the indicators of precarious work into a cumulative index decreases the odds of voting by 9 percent (e^−0.095 = 0.909) for each additional indicator. These effects are substantial given that the shift from no indicators of precarious work to all five indicators reduces the odds of voting by 44 percent (0.909⁶ = 0.566). To determine the relative strengths of each indicator, we rely on effect size calculations: we take the ratio of the logit coefficients to the SD of the dependent variable, and divide that by the ratio between the shift in the value of the covariates (which is 1 given their binary status) and the SD for each covariate. A 1 SD shift in non-indefinite contract, low job control, and low organizational influence is associated with modest effect sizes, between −8% and −9% of a SD in electoral participation, whereas employment gaps and in particular financial vulnerability have stronger effects (−16%, −25% SD respectively). On the other hand, a 1 SD increase in the cumulative index is associated with a −29% SD change in electoral participation, corroborating the relevance of such an index for electoral participation.

The significance of precarious work for SES differences in voting can be demonstrated in two ways. First, Figure 1 shows expected probabilities of voting for all the different SES indicators. In the case of precarious work (panel A), the overall shift in likelihood of voting is likelihood of voting is 7 and a half percent (0.848 vs. 0.772). This compares with 12 percent for educational attainment (0.862 for those completing tertiary education vs. 0.741 for those with lower secondary attainment). The education effect is actually more complicated given changes over time in mandated education that has raised the floor for attainment to lower secondary achievement. Taking this into account in determining the range of educational attainment changes the overall effect to 8.7 percent (0.866 vs. 0.779. Panel C shows the effect for attained social class and shows a 7 percent swing from higher managerial/professional (0.864) to those in routine work (0.793). Estimated effects for the other indicators are somewhat smaller. For family class (see panel D), the variation in voting is just over 4 percent (0.851 for those from clerical work backgrounds to 0.810 for those whose fathers were unskilled labor). Effects for poverty are just over two percent (0.809 vs. 0.830), while the largest differences for any income contrast is 4.2 percent (not shown). Finally, the current employment—unemployment gap is 2.4 percent (0.833 vs. 0.809). In the end, precarious work has large effects on inequalities in voting, dwarfing the effects of many traditional indicators of SES. Only educational attainment has larger effects and, given the attention that educational attainment has garnered in studies of electoral participation (Verba et al., 1978; Gallego, 2010), precarious work seems a vital dimension of socioeconomic differences in voting. Moreover, the detrimental effects of precarious work on likelihood of voting are independent of both the sociodemographic correlates and the traditional sociodemographic factors that have been the spine of traditional research on the topic. This remarks their independent relevance in the broader relationship between socioeconomic inequalities and electoral participation.

Figure 2 provides a second lens on the compounding nature of precarious work in relation to traditional indicators of SES. Panel A is the condition of maximal disadvantage based on the traditional indicators of SES. Estimates for the expected number of precarious work indicators are calculated when family class is unskilled labor (8), parental educational attainment is low (1), respondent's educational attainment is low (2), the respondent is contemporaneously unemployed, and the respondent is in routine work and poor. The alternative, Panel B, is maximal advantage with estimates for respondents whose parents were professionals (i.e., family class = 1), where parental and respondent's educational attainment is high (5) and the respondent is employed in higher managerial/professional work (1) and not poor. Bracketing the effects of precarious work for the moment, socioeconomic inequality in voting based on cumulative advantage/disadvantage is large. The difference in likelihood at average levels of precarious work is almost 26 percent (0.9095 vs. 0.652). The implications of this in terms of “class” representation are striking in that almost 9 in 10 of those advantaged voted in the most recent election compared with just over 6 in 10 of those disadvantaged.

FIGURE 2

Figure 2. Marginal effects at the means with 95% confidence intervals, after model 6 in previous table. Data source: European Social Survey Multilevel (2008-2018).

When precarious work is added to the discussion, the non-linear estimation strategy produces a small difference in the magnitude of the differences in voting with increasing precarity. When SES is low, the difference in the probability of voting when precarious work moves from 0 to 5 is 0.135 (0.679 vs. 0.545). The same contrast when socioeconomic advantage is high is 5.4 percent (0.919 vs. 0.865). Combining the two aspects of SES shows the significance of precarious work for overall socioeconomic inequality. Differences at the extremes of SES when precarious work is incorporated shows a gap of 37 percent (0.919 vs. 0.545), just under half the sample mean (0.795) for voting. On average, the contribution of precarious work to cumulative socioeconomic inequality exceeds 30 percent ([0.374–0.26]/0.374 = 0.305).

4.1. Assessing generalizability

We conclude by assessing heterogeneity in the relationship between precarious work and electoral participation. Our interest here is largely in the robustness of associations across different political economic contexts. We assess this variation by including statistical interactions between extent of precarious work individual countries. We report the full model with individual countries in Table 4.

TABLE 4

Table 4. Logit coefficients: probability of voting and precarious work, interactions with country.

Starting from the country perspective, 26 of the 31 product terms are statistically significant. Given this, we have very strong grounds for examining country-level heterogeneity in the political consequences of precarious work. Although there is variation in the size of effects, a negative relationship between precarious work and voting is rather robust. Here, estimated effects are negative in 29 of 32 countries of which 21 are statistically significant. No country shows a statistically significant, positive effect. Countries showing non-significant effects are also rather idiosyncratic and do not graft on to any particular welfare state regime. In fact, the latter pool contains countries representative of four of the five regimes and does not include the majority of countries in any specific regime. In the end, the negative effect of precarious work on voting is rather widespread across countries.

5. Discussion

Large-scale transformation of the global economy has radically changed the nature of paid employment for growing numbers of people. A key aspect of this is the expansion of precarious work and by extension “bad jobs” in the latter 20th century, jobs that provide marginal and ephemeral compensation and ultimately fail in providing a foundation for social life (Kalleberg et al., 2000; Kalleberg, 2011; Osterman and Shulman, 2011).

While there is some speculation that precarious work has important implications for political activities and engagement, ours is the first work to theoretically elaborate how precarious work translates into political apathy and then tests its effects with a large, cross-national sample. In doing so, we contribute to and extend the vast literature on socioeconomic marginality and its negative effect on political engagement (Wolfinger and Rosenstone, 1980; Powell, 1986; Jackman, 1987; Blais et al., 2004; Franklin, 2004; Blais, 2006; Leighley and Nagler, 2013; Smets and Van Ham, 2013; Jeannet, 2022), integrating their different mechanisms to assess the impact of the broader concept of precarious work (Kalleberg, 2009) on electoral participation.

Our research operationalizes precarious work in a multidimensional manner and examines its effects on likelihood of voting. With rigorous statistical models, results suggest large detrimental effects on electoral participation. Both independently and cumulatively, precarious work lowers the likelihood of voting, and most importantly does so independently from traditional indicators of low SES, whose impact on electoral participation is well-established in the literature (Smets and Van Ham, 2013). These patterns support our broad hypothesis on the multifaceted negative impact of precarious work on electoral participation.

While all the five dimensions have negative and statistically significant effects on electoral participation, their impact differs in magnitude: while there are lower effect sizes for non-indefinite contract and the two forms of control and influence in the workplace (around −9% SD), impact is more profound for employment gaps (−16% SD) and especially for financial vulnerability (-25% SD), with the cumulative index exerting the most powerful effect (−29% SD). How can we explain theoretically this variation in impact? Theoretically, a possible explanation lies in the social consequences of each dimension: the strongest impact is driven by severe experiences of socio-economic hardship which may spill-over outside the workplace and affect family and the broader social network which is instrumental for political participation (Rosenstone, 1982; Laurence, 2015; Mewes et al., 2021; Azzollini, 2023), while the forms of limited control/influence may have smaller consequences outside the workplace, for instance not directly fostering social stigma among family and friends. The lower magnitude of the non-indefinite contract may be associated to different social norms attached to not having a permanent contract, and indeed related to the broader cross-national variation which we address below. In sum, dimensions of precarious work with a likely stronger impact on societal standing affect electoral participation more than those dimensions that are largely limited to the workplace. Further research may test formally this mechanism, potentially by examining the impact of different dimensions of precariousness on social engagement.

The cross-national nature of the dataset allows us indeed to assess that the focal relationship applies across 21 countries, and in ways that are not reflective of variation in welfare state regime. On average, high work precarity reduces voting by 8 percent with estimates exceeding 10 percent for 11 countries and exceeding 5 percent for another 10 countries, with only educational attainment having similar or stronger effects. Then, how can we explain the seeming pervasiveness of the precarious work-electoral participation relationship, and the lack of predictable welfare regime patterns? There are several potential explanations, on both demand and supply sides. On one hand, recent research shows that regional dynamics have stronger effects than national ones on life satisfaction (Clark, 2003), electoral participation (Azzollini, 2021, 2023) and socio-political trust (Fairbrother and Martin 2013; Lipps and Schraff, 2021), as more proximate social contexts may influence more powerfully social norms and perceptions (Eichhorn, 2014). On the other hand, the extent to which precarious workers participate or not may be affected by the electoral and political systems, as different parties may employ different strategies on whether to represent labor market insiders and outsiders (Emmenegger et al., 2015; Häusermann et al., 2020), potentially linking or excluding precarious workers from the representation of the traditional working class (Standing, 2011; Wright, 2016). Building on the individual relationship examined in this paper, and the assessment of its generalizability across countries, future research may formally test these explanations for this puzzling finding in a separate paper, testing both demand and supply side explanations at different geographical levels.

Given that many elections across the United States and Europe are won or lost based on a few percentage point differences (e.g., the U.S. Presidential Elections of 2000, 2016, and 2020, the UK General Elections of 2017, the Italian National Elections of 2006 and 2013), variation in precarious work may be crucial in affecting political outcomes, at regional, national and even global scales.

Collectively, the combination of precarious work and other socioeconomic differentials highlight a worrying dimension of political inequality. While those of low socioeconomic standing have lower likelihoods of voting, their unique vulnerability to precarious work makes differences in electoral participation much, much larger than traditionally measured. Precarious work both compounds the effects of low family social class, poor educational attainment, and chronic unemployment and exacerbates similar processes among young people and immigrants. Theoretically, this phenomenon may be explained with the contrast between citizens and denizens. As Standing (2012) argues, citizens typically enjoy the full spectrum of social and political rights, while denizens have only a limited range of those rights. Given that those in precarious work are “industrial denizens”, due to their lack of security over a plethora of labor-related aspects, industrial denizenship translates into political denizenship. Those that suffer from socio-economic marginality largely refrain from exercising their right to vote, despite being formally enfranchised, and work precarity compounds existing political disengagement. Given this, socio-economic marginality continues to drive political inequality in elections, with the most socially struggling groups refrain in large numbers from formal electoral participation.

While our research is silent on the issue of who precarious workers might vote for if they were to vote, there are good reasons to anticipate implications for political outcomes. The clear link between socio-economic marginality and political inequality determines not only which political forces win or lose, but also is likely to influence the policy-making process in favor of the socio-economic affluent and electorally engaged. As argued by Lijphart (1997; see also Verba, 1995), there is a concrete risk of a vicious circle between socio-economic marginality driving down the electoral participation of those in the lower strata of society, increasing inequalities in the political representation, and responsiveness of elected officials toward the latter. And while we do not address the issue of which party do they vote for when they do, Standing (2011) speculates that different categories of the precariat might be particularly susceptible to populist, right-wing propaganda that increasingly present in contemporary political discourse. And while low levels of electoral participation among the socio-economic disadvantaged clearly favor the Republican party over the Democratic party in the United States (Uggen and Manza, 2002), the situation is less clear in Europe. Even if those in precarious work and the traditional working class may in principle have similar interests (Wright, 2016), their insiders-outsiders divide in the labor market may translate into the political arena by posing a representation dilemma to center-left parties (Lindvall and Rueda, 2014; Häusermann et al., 2020), for instance on issues such as the strictness of Employment Protection Legislation (Biegert, 2017) or on the universal basic income (Vlandas, 2021). Still, the implications of precarious work for party orientations are ultimately empirical and should be the subject of future research, as well as other political engagement forms including civic participation and political interest. Another direction for future research comes from a limitation of our study: we rely on a repeated cross-section, which allows us to assess generalizability across countries, but does not allow us to observe the same individuals before and after the transition to precarious work, nor other within-respondent changes in the precarious work indicators. This would only be possible with panel data, which however is available for a limited amount of countries, typically those with most established research infrastructures. Therefore, having established the general pattern, future research may examine the impact of precarious work on electoral participation over time.

In the end, the expansion of democracy as a global project was one of the most successful social dynamics of the 20th century. In the early 1900s, less than a dozen countries were democratic. By 2000, almost 100 countries were fully democratic and another 50 were a mix of democratic and autocratic principles and practices, with notable democratic backsliding (Waldner and Lust, 2018). And the right to vote is the cornerstone of democratic governance. It is the right that “makes all other political rights significant” (Piven and Cloward, 2000: 2). Yet the universality of the right to vote means very little when there is large scale and socially structured abstentionism (Lijphart, 1997). Our research on the political consequences of precarious work illuminates how changing incarnations of socioeconomic inequality continue to undermine electoral participation, which poses a direct challenge to the health of contemporary democracies.

Data availability statement

Publicly available datasets were analyzed in this study. This data can be found here: European Social Survey Data Portal.

Author contributions

All authors listed have made a substantial, direct, and intellectual contribution to the work and approved it for publication.

Funding

This work was supported by funding from the European Research Council (ERC) Synergy Project DINA (Toward a System of Distributional National Accounts, Grant no. 856455) and from a Leverhulme Trust Grant for the Leverhulme Center for Demographic Science.

Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpos.2023.1176686/full#supplementary-material

Footnotes

1. ^Social mobility was clearly an important aspect of both class and socioeconomic analysis, yet the lion's share of research focused on mobility across rather than within generations.

2. ^We examined a variety of cut-points ranging from bottom decile to the 60^th percentile and results are substantively similar.

References

Acevedo, M., and Krueger, J. I. (2004). Two egocentric sources of the decision to vote: The voter's illusion and the belief in personal relevance. Polit. Psychol. 25, 115–134. doi: 10.1111/j.1467-9221.2004.00359.x