The Italian labor market: from dualism to generalized precarization

During the post-war reconstruction period, Italy experienced a long phase of economic growth that ensured welfare state consolidation and expanding employment protection legislation in defense of workers' interests (Brandolini et al., Reference Brandolini, Gambacorta and Rosolia2018). However, the country's highly fragmented income maintenance system, its selective iper-garantismo (term used to indicate the hyper-protection of regular workers, especially in the public sector) and its highly polarized system of welfare created a deep divide in protections and subsidization between workers in core and highly unionized sectors and workers in peripheral ones (Ferrera, Reference Ferrera1996). This inequality in the distribution of protections and social benefits originated in the strengthening of trade unions and consequent expansion of workers' rights which followed the autunno caldo (hot autumn), a wave of strikes, occupations and sit-downs that struck Italy in the fall of 1969. The reaction of employers to the resulting rigidification of the productive process was the transfer of productive activities to a secondary sector filled with marginal workers lacking the rights that unions had secured to their core constituencies (Piore, Reference Piore1980). Hence, the growth of the secondary sector occurred in symbiosis with the increasing protection of the primary sector, so that by the 1980s Italy became the ‘clearest case of labor market dualism in the industrialized West’ (Piore, Reference Piore1980: 384–385).

The extreme dualism of its labor market renders Italy a most-likely case for the play-out of Rueda's insider-outsider partisanship model (Rueda, Reference Rueda2005). The model predicts that, in contexts characterized by labor market dualism, neither social democratic nor conservative governments will promote pro-outsider policies, due to the low political participation of outsiders which renders their representation scarcely beneficial. Coherently, during the years of the First Republic (1947–1992) Italian parties not only failed to mitigate dualism but contributed to widening it. In these years, party competition was structured around three poles, with the DC (Democrazia Cristiana), since 1960s in organic alliance with the Socialist Party, at the center and two anti-system parties, the neo-fascist MSI (Movimento Sociale Italiano) and the communist PCI (Partito Comunista Italiano), at the extremes. This configuration worked against the adoption of universalistic social benefits, promoting competition within the left camp for the support of unionized workers. The result was a ‘workerist’ system of social insurance and labor market reforms benefiting unionized blue-collar workers (see Ferrera and Gualmini, Reference Ferrera and Gualmini2004), coexisting with severe regulations limiting the use of fixed-term contracts and other forms of sub-employment. Besides protecting insiders from competition (see Lindbeck and Snower, Reference Lindbeck and Snower1989), these regulations kept the share of workers in atypical employment low.

This situation gradually changed in the late 1990s and early 2000s, when encompassing labor market reforms were approved to prepare the country's entrance in the European Monetary Union. Landmarks in this process were the Treu law (1997) and the Biagi reform (2003), adopted by center-left and center-right governments. The former provided an institutional framework for the resort to temporary employment and reduced disincentives to the use of fixed-term contracts. The latter further liberalized atypical employment via the institutionalization of new forms of atypical work arrangements.

These reforms marked a shift away from the traditional concern for the security and income-protection of insiders, but left the core of protected workers virtually untouched. It was only in the aftermath of the financial crisis that retrenchment of insiders' rights begun. Following recommendations from the European Central Bank, the technocratic administration headed by Mario Monti promoted a package of austerity measures which included a structural reform of the labor market (Law 92/2012). With the alleged aim of removing structural obstacles to growth via the elimination of labor market rigidities, the first target of the reform was the loosening of insiders' protections. A first draft envisaged the elimination of the extensive short-term work (STW) schemes that, by ensuring the subsidization of temporarily redundant employees in large firms, have long been considered the major source of the iper-garantismo of Italian insiders. This proposal clashed with the fierce opposition of labor unions backed by the Democratic Party, and STW schemes for employees of large firms with more than 15 employees were maintained. Nonetheless, insider protections were significantly loosened via the weakening of legislation protecting workers with open-ended contracts from unfair dismissal (see Picot and Tassinari, Reference Picot and Tassinari2017). In addition, the reform further liberalized atypical employment, eliminating the obligation to justify the stipulation of temporary contracts up to 1 year of duration.

In this phase, the actions undertaken by the Democratic Party were roughly in line with the predictions from Rueda's model. Despite its embrace of the flexibilization agenda in the 1990s, the party continued flanking labor unions for the protection of insiders' long-standing rights. This changed radically under the leadership of Matteo Renzi, elected party leader in December 2013. Renzi promoted a comprehensive reform, named Jobs Act, that substantially reduced insiders' protections by decentralizing the bargaining process and weakening firing rules. Most notably, it eliminated the obligation to the reinstatement of workers in case of unfair dismissal and limited the possibility to resort to those same STW benefits for the protection of which the Democratic Party had mobilized in 2011.

The approval of liberalizing reforms at the hands of the Democratic Party raises the question of why a social democratic party should promote market-oriented policies that undermine the interests of its core constituency. This puzzle is neither new nor limited to the Italian case, as similar policies have been implemented by social democratic governments across Western Europe since the 1980s. Different explanations have been proposed. Following Klitgaard (Reference Klitgaard2007), the shift of social democrats toward the promotion of market-oriented policies is aimed at the protection of their main political weapon, the welfare state, when its legitimacy is threatened by the persistence of poor economic performance and high levels of unemployment. Alternatively, Beramendi et al. (Reference Beramendi, Häusermann, Kitschelt and Kriesi2015) attribute the shift to a change in the preferences of their voters away from social consumption toward social investment, change driven by the growing weight of highly educated voters in their electorates (see also Gingrich and Häusermann, Reference Gingrich and Häusermann2015). The latter mechanism, partially modified, reasonably applies to the Italian case. Educated voters came to constitute a core constituency of the Democratic party (see Maraffi, Reference Maraffi2018), and the Jobs Act has been interpreted as part of a broader strategy by the Democratic Party to gather the support from highly educated and upper classes. Although the liberalization of the labor market was promoted in combination with consumption-oriented policies, these policies were targeted to gather support from upper groups (Vesan and Ronchi, Reference Vesan and Ronchi2019).

Summing up, the precarization of the Italian labor market occurred in two stages. At first, reforms took place at the margins, institutionalizing the use of atypical work arrangements, enhancing employers' capability to resort to temporary contracts, and expanding unemployment protection coverage to the widening group of atypical workers. In this phase the Democratic Party, in joint action with labor unions, defended insiders' interests, which is in line with the original formulation of the insider-outsider partisanship model. By 2012, however, the technocratic administration of Mario Monti and the democratic administration of Matteo Renzi caused a step change: the liberalization of atypical employment continued, but it was sided by the loosening of pro-insider employment protection legislation. This two-stage process emerges clearly in Figure 1, which displays trends in the strictness of employment protections for permanent (left panel) and temporary (right panel) contracts. It shows that, in the time frame 1990–2018, the restrictions to employers' usage of temporary contracts declined steadily, while the strictness of the dismissal regulation of workers in permanent employment remained unchanged for most of the period, declining from 2012 onward.

Figure 1.

Strictness of employment protections, 1990–2018.

Source: OECD (Reference OECD2020). The OECD Employment Database. The strictness of employment protections for permanent contracts is the average of four indicators that measure the strictness of the dismissal regulation of workers in permanent jobs (procedural requirements, notice and severance pay, regulatory framework for unfair dismissal and enforcement of unfair dismissal regulation). The strictness of employment protections for temporary contracts is the average of indicators that measure the restrictions to employers' usage of temporary contracts (e.g. the valid cases for the use of fixed-term contracts, maximum number of contract renewal and temporary contracts maximum (cumulated) duration). Both measures are constructed on a scale 0–6. The dotted line (drawn at 2012) signals the start of the second stage in the precarization of the Italian labor market.

This combination of an increasing number of workers in atypical employment and lowering protections for workers with permanent contracts contributed, in joint action with an economic crisis which fostered job destruction especially within younger cohorts (see Liotti, Reference Liotti2020), to the shift of the Italian labor market from dualism to generalized precarization. Following Hipp (Reference Hipp2016), when the incidence of temporary employment is high, the more stringent the dismissal protections the higher the odds that workers do not worry about keeping their jobs. This evidence suggests that, in Italy, the weakening of the dismissal regulations and the concomitant liberalization of atypical employment created a fertile ground for the spread of perceived precarity. Following the reforms, not only the share of workers trapped in atypical and insecure jobs increased (among others, see Barbieri and Scherer, Reference Barbieri and Scherer2005; Brandolini et al., Reference Brandolini, Gambacorta and Rosolia2018; Baldini et al., Reference Baldini, Gallo, Lusignoli and Toso2019; Boeri and Garibaldi, Reference Boeri and Garibaldi2019; Liotti, Reference Liotti2020), but permanent employment ceased to be a guarantee of security. In the dual system, a clear-cut division existed between the majority of workers in permanent employment largely insulated from the risk of unemployment, and the minority of the unemployed and atypically unemployed; but as employment protections for workers with open-ended contracts loosen and the reserve army of the atypically employed and unemployed grows, the insiders are no longer sheltered by the actual and perceived risk of unemployment. Coherently, while at the beginning of the century only a marginal share of Italian workers feared job loss (16%), in 2018 concerned employees outnumbered their unconcerned counterparts (51%) (ITANES, 2001, 2018). This dramatic increase has not only been driven by the rising share of employees with temporary or atypical work arrangements (from 12% to 21% in the sample), but also by the spreading perception of precarity among permanent workers. In 2018, the 43% of employees on open-ended contracts declared to have feared job loss, against the 12% in 2001.

Precarity and voting in the 2018 election

The shift toward generalized precarization calls into question core assumptions in the insider-outsider literature, and challenges the explanatory power of the model in contemporary Italy. First, as open-ended contracts are no longer guarantee against the actual and perceived vulnerability to unemployment, formal employment status might no longer be a reasonable predictor of policy preferences. Second, individuals in condition of occupational precarity can no longer be assumed to be a minority whose representation is not electorally beneficial. This raises the question of whether and how precarity influences voting behavior in this newly changed context.

Concerning the first issue, studies that investigate the relationship between precarity and voting tend to focus just on formal employment status (for a review, see Schwander, Reference Schwander2019). This is justified on the grounds that different kinds of employment contracts and work arrangements entail a different vulnerability to the risk of unemployment which in turn has an influence on policy preferences (see Rueda, Reference Rueda2005; and Lindvall and Rueda, Reference Lindvall, Rueda and Emmenegger2012). This argument is reasonably sound when employment and income protection schemes grant insiders with high levels of job and income security. However, when employment protections decline permanent contracts no longer automatically shelter from the actual and perceived risk of unemployment, especially in a context of economic crisis and rising unemployment rates. It follows that perceptions of precarity and policy preferences can no longer be inferred by relying on information on workers' contractual arrangements (see also Lewchuk, Reference Lewchuk2017). In this context, ‘the employment type describes the work (labor) organization of special-type employees (for example, self-employed) or relations with the employer (for example, informal relations) but, admittedly, says nothing about the instability of the employee's social and economic situation’ (Gasiukova and Shkaratan, Reference Gasiukova and Shkaratan2019: 118). It follows that operationalizations of precarity based on formal employment status are bound to suffer a severe validity bias, especially when used by scholars interested in the impact of precarity on policy preferences and voting behavior. In this context, I contend, the perception of insecurity concerning one's employment status is a better-suited indicator of precarity than formal employment status, and can be expected to hold more explanatory power in models of vote choice.

Concerning the second issue, the shift from dualism toward generalized precarization has important consequences for the representation of precarious workers. In the insider-outsider literature, the interests of labor market outsiders are assumed to be under-represented in the arena of policy making due to their low number and to the difficulties entailed in their mobilization, factors which render their representation hardly beneficial (see Rueda, Reference Rueda2005). In a context of generalized precarization, however, these obstacles are lifted. First, as precarity spreads the number of voters who can be attracted via the promotion of pro-outsiders policies increases. Second, when precarity is no longer prerogative of marginalized workers in peripheral sectors the interests of hardly mobilizable workers align with those of workers in unionized and core sectors. As a result, the electoral weight of precarious workers increases, creating incentives for political parties to represent their interests.

Not all parties, however, are equally equipped for, or likely to, mobilize the interests of precarious workers and attract their electoral support. Whether and how precarity influences vote choice is conditional on precarious workers considering some parties as responsible for their condition, and/or more willing or capable than others of improving their situation. In this respect, challenger parties seem to enjoy a comparative advantage over mainstream parties. Following Antonucci et al. (Reference Antonucci, D'Ippoliti, Horvath and Krouwel2021), the insecurity stemming from a vulnerable position in the labor market might pull voters away from mainstream toward anti-establishment options (symbolic mechanism). Precarious workers might also refrain from mainstream parties because, by embracing the flexibilization agenda and converging on neoliberal and austerity policies, these parties are directly responsible of their increased insecurity. Conversely, they might turn to right-wing challengers because these parties' welfare chauvinistic rhetoric attracts insecure voters who feel threatened by migrants' competition in the labor market, and to left-wing challengers because these parties' radical stances against austerity measures and commitment toward redistribution directly address their economic insecurity (instrumental mechanism).

The instrumental and symbolic mechanisms theorized by Antonucci and co-authors can be reasonably expected to play out in Italy, although with some specifications. As for mainstream parties, since the 1990s both center-left and center-right parties have converged on the flexibilization agenda that contributed to spread insecurity among workers. This trend was exacerbated by the onset of the sovereign debt crisis, that severely affected the legitimacy of incumbent governments and mainstream parties. Across Southern Europe, the crisis prompted an ‘electoral epidemic’ marked by low levels of turnout, declining support for the incumbents, and diminishing trust in political parties (Bosco and Verney, Reference Bosco and Verney2012), that soon turned into a government epidemic characterized by extreme electoral volatility and government instability (Bosco and Verney, Reference Bosco and Verney2016; Hutter et al., Reference Hutter, Kriesi and Vidal2018). This loosened the ties between mainstream parties and their electorates (Marcos-Marne et al., Reference Marcos-Marne, Plaza-Colodro and Freyburg2020), and opened a niche for challenger parties to capitalize discontent against austerity and voice the need for a regeneration of the political system (Hutter et al., Reference Hutter, Kriesi and Vidal2018).

In this context, it is reasonable to expect a pull-out effect of precarity on support for mainstream parties. However, while the symbolic mechanism theorized by Antonucci et al. (Reference Antonucci, D'Ippoliti, Horvath and Krouwel2021) should work against the mainstream left and the mainstream right, the instrumental mechanism should work more strongly against mainstream left parties who hold the paternity of the most recent liberalizing reforms and are most exposed to blame-attribution due to their incumbent position at the time of the election. Based on these considerations, I expect that:

As for challenger parties, the convergence of mainstream parties on labor market deregulation can be expected to increase their appeal among precarious workers, which is in line with the symbolic mechanism theorized by Antonucci et al. (Reference Antonucci, D'Ippoliti, Horvath and Krouwel2021). Once again, however, the instrumental link should work most in favor of the left-wing challenger (i.e. the Five Star Movement) as compared to the right-wing challengers.

The M5S was founded in 2009 by the comedian and activist Beppe Grillo and by the web entrepreneur Roberto Casaleggio. Although it is most renowned for its anti-establishment and environmentalist stances, the movement is markedly pro-welfare (see also Vesan and Ronchi, Reference Vesan and Ronchi2019) and has been strongly committed against the precarization of the Italian labor market since its inception. In 2006, Grillo published ‘Modern slaves: The precarious in the Italy of wonders (Schiavi moderni: Il precario nell'Italia delle meraviglie), a collection of short stories where precarious workers share their daily experiences of exploitation and insecurity. The book is an explicit critique against the reforming of the Italian labor market and the consequent precarization of Italian workers:

‘The Biagi law introduced the precariat in Italy. A modern bubonic plague that affects workers, especially at a young age. Before it wasn't there, now it is. It turned work into timed projects. Salaries into charity. Rights into unreasonable claims. Everything became a project in order to apply the Biagi law and create the new modern slaves’ (Grillo, Reference Grillo2007: 6. Translation provided by the author).

In the comedian and activist's blog posts, the references to the emerging precariat were frequent, together with the call for the introduction of a citizenship income granting income security to all. In the ‘Grillo's agenda’ published in late December 2012, the measure ranked high in the 16 items list, second only to the introduction of an anti-corruption law (Grillo, Reference Grillo2012). This commitment against precarity increased during the XVII legislature (2013–2018), when the Movement made a guaranteed minimum income measure named Reddito di cittadinanza (RdC) its flag policy and the pivot of the 2018 electoral campaign.

By centering its program around the RdC, the M5S landed at the 2018 election presenting itself as the representative of precarious workers. Once in government, the movement lived up to its promises and confirmed its commitment against precarity. By 2019 the RdC entered into effect; in the same year, the decreto dignitá (‘dignity decree’) introduced measures limiting employers' capability to resort to temporary contracts, reducing temporary contracts' maximum duration and renewal period and strengthening regulations protecting workers from unfair dismissal. In addition, the Decreto riders was approved, which expanded employment and social protections for food-delivery couriers (see Carella and Marenco, Reference Carella and Marenco2022).

As concerns the right-wing challengers, Antonucci et al. (Reference Antonucci, D'Ippoliti, Horvath and Krouwel2021) theorize that their welfare chauvinistic stances should attract the support of precarious workers. In addition, labor market outsiders have been theorized to overlap with the losers of globalization, and as such to represent one of their core constituencies. However, more recent studies show that such an overlap is largely mistaken (Häusermann, Reference Häusermann2020) and that it is outsiderness, not exposure to international competition, the major driver of income and employment insecurities (Natili and Negri, Reference Natili and Negri2022). Coherently, neither the League nor Brothers of Italy manifested a commitment toward the protection of precarious workers which resembles the one displayed by the M5S. It follows that, while the symbolic mechanism should equally benefit anti-establishment parties regardless of their ideology and programmatic positioning, the instrumental link should work especially in favor of the M5S. I therefore expect that:

Finally, the impact of precarity on voting behavior might be moderated by the individual's financial situation. Albeit sharing a condition of vulnerability in the labor market, precarious workers differ on a wide array of dimensions, including personal and household income. This creates differentials in the extent to which precarity is experienced as a limiting condition, and it is relevant in shaping vote choices. I hence hypothesize that:

Data and operationalization

In order to test the above formulated hypotheses, I rely on survey data collected by ITANES in occasion of the 2018 general election and run multinomial logistic regression models testing the effect of perceived precarity on vote choice. I classify parties in four groups based on their ideology and mainstream or challenger status, relying on the definition of challenger developed by De Vries and Hobolt (Reference De Vries and Hobolt2020) which includes parties without recent government experience. Based on these two criteria, I identify four party families: the mainstream left (the Democratic Party and other parties in the center-left coalition); the mainstream right (Berlusconi's Forza Italia and other parties in the center-right coalition); the challenger left (the Five Star Movement); and the challenger right (Brothers of Italy and the League, the latter being categorized as challenger due to the radical restructuring the party experienced under the leadership of Matteo Salvini).

The key explanatory variable in each model is the respondents' perceived precarity in the labor market. I operationalize it by relying on the survey question Over the last year, have you been afraid of losing your job?. I classify respondents as precarious if they declared to have been very or somewhat afraid, as not-precarious otherwise; the unemployed, to whom the question was not addressed, are all assigned to the ‘afraid’ category. This item is well suited for measuring perceived precarity in that the fear of job loss is not an indicator of perceived job insecurity only, which refers to the perceived probability of job loss. It is also not an indicator of perceived employment insecurity only, which refers to the perceived availability of alternative jobs. Rather, it is the result from a combination of both types of insecurity, of the anxiety related to potential job loss and the concomitant fear of not finding a new job (see Hipp, Reference Hipp2016).

In each model, I control for formal employment status by relying on a categorical variable with four levels: self-employment (reference-category); permanent employment; temporary employment; and unemployment. This allows to isolate the effect of perceived precarity from the effect of formal employment status, and test whether perceived precarity and / or formal employment status influenced vote choice. Additionally, I control for a set of socio-demographic and attitudinal variables. As for socio-demographic factors, I control for age, gender, education level, region of residence and union membership. The inclusion of age, gender and education is necessary since younger, female and less educated individuals are over-represented among precarious workers. The region of residence is included because differences in vote trends across Italian regions exist, with the M5S being more successful in Southern regions where unemployment, poverty and social distress are more widespread (Maraffi, Reference Maraffi2018; Brancaccio et al., Reference Brancaccio, Mete and Tuorto2019; Tuorto, Reference Tuorto2019). Finally, union membership is controlled for because of its well-established correlation with employment security and left-wing voting. As for attitudinal factors, I control for respondents' self-positioning on the left-right scale, attitudes toward the European Union and immigration, trust in political parties, populist beliefs, and judgement of the performance of the incumbent government (the Democratic administration of Paolo Gentiloni) on economic issues. Controlling for the respondents' judgment of the Gentiloni administration allows to rule out the possibility that the relationship between perceived precarity and vote choice is driven by an incumbent effect: among individuals dissatisfied with their employment situation the incumbent gets the blame and loses, the opponent thrives.

Finally, economic hardship is operationalized as a dummy scoring 1 if the respondent declared to be facing difficulties in living with the family income, 0 otherwise. This variable is at first included as a control and, in a second stage, it is interacted with perceived precarity.

Analysis and results

Prior to testing hypothesis, I examine the relationship between precarity and turnout. Mainstream theories of representation in dualized societies link the under-representation of outsiders by hands of political parties to outsiders' low levels of political participation. Therefore, it is crucial to investigate whether this argument still holds true in present-day Italy. To accomplish this, I utilize a binomial logistic regression model (Model 1) to estimate the impact of precarity on voter turnout. Results are reported in Table 1 as odds ratios, which indicate the relative likelihood to vote for the party considered in the estimation as compared to the party used as reference category (odds ratios greater than 1 indicate a higher likelihood, odds ratios smaller than 1 indicate a lower likelihood). They show that precarity increased the likelihood of voting: specifically, the predicted probability that an average individual in the sample voted increases by 5.3 percentage points as they shift from being not afraid to being afraid, a positive effect which is not conditional on economic hardship, as proven by the non-significant interaction term in Model 7 (Table 2, page 26). This result challenges one of the key assumptions in the insider-outsider literature, and exposes the relevance of shedding light on the relationship between precarity and vote choice.

Table 1.

Models of turnout (1) and vote choice (2 and 3); coefficients are presented as odds ratios.

Reference categories in brackets. Significance levels: *P < 0.1; **P < 0.05; ***P < 0.01.

Table 2.

Models of turnout and vote choice, with interaction; coefficients are displayed as odds ratios.

Reference categories in brackets. Significance levels: *P < 0.1; **P < 0.05; ***P < 0.01.

To test the hypotheses, I focus on respondents who voted and run multinomial regression models to estimate the effect of precarity on the probability to support a challenger over a mainstream party. Model 2 estimates the relative likelihood to support the M5S and a right-wing challenger compared to any mainstream party; Model 3 the relative likelihood to support a mainstream left and a mainstream right party compared to any challenger party. The coefficient estimates, reported in Table 1 as odds ratios, point in the expected direction, and indicate that precarity increased the odds of voting for a challenger relative to a mainstream party. Results from Model 2 show that the odds of voting for a left-wing challenger (i.e. the M5S) over a mainstream party are 1.68 times greater for precarious than for non-precarious respondents, and the odds of voting for a right-wing challenger 1.21 times greater, although the latter coefficient fails to reach statistical significance. Results from Model 3 show that the odds of voting for the mainstream left over a challenger are 0.54 times smaller for a precarious than for a non-precarious respondent, and those of voting for a mainstream right party 0.96 times smaller, although this coefficient is also not statistically significant.

The results from Models 2 and 3 allow to draw conclusions on the relative likelihood to support challenger as compared to mainstream parties, suggesting that precarity fosters support for challenger over mainstream parties although the effect is only significant for the challenger and mainstream left. However, they do not allow to draw conclusions on the effect of precarity on the probability to support each party per se. To address this issue, I run a multinomial model (Model 4) where I disaggregate the dependent variable vote choice in four categories (regression coefficients are displayed in the Appendix, Table A1). I then calculate how the predicted probability of an average voter to support each party changes as he shifts from not precarious to precarious (Appendix, Figure A1), and display the change in these probabilities in Figure 2.

Figure 2.

Differences in the predicted probability to support each party family following a shift from not precarious to precarious.

The results show that precarity had a significant effect on the probability of supporting the Five Star Movement, increasing it, and the Democratic Party, decreasing it. However, precarity did not have a significant effect on the probability of supporting either mainstream or challenger right-wing parties. These findings support hypotheses 2 and 3 and align with the instrumental mechanism proposed by Antonucci et al. (Reference Antonucci, D'Ippoliti, Horvath and Krouwel2021). They show that precarious workers did not turn to challenger parties indiscriminately, but were drawn to the party that advocated for policies that directly safeguarded their interests. This indicates that it is not challenger and mainstream status per se that attracted or repelled precarious workers (the symbolic mechanism proposed by Antonucci and colleagues), but rather instrumental considerations.

Concerning the effect on formal employment status, the results indicate that being in a permanent employment relationship increases the probability of supporting the M5S over a mainstream left party, although this effect is only statistically significant at the 0.10 confidence level. This result may seem counter-intuitive at first, as employees in standard employment relationships have been long considered the core-constituency of social democratic parties. However, it is in line with the consideration that the security of permanently employed individuals has been undermined by reforms promoted by the Democratic Party since 2014. As a result, it is reasonable for these voters to turn to the M5S, which has spoken out against the trend toward flexibilization. In addition, neither unemployment nor atypical employment status increased the probability of supporting the M5S over any other party, a finding that holds even when perceived precarity is excluded from the models (Model 5, Table A2 in the Appendix).

The lack of any significant effect of formal employment status on vote choice in Model 5 corroborates the expectation that perceived precarity, not formal employment status, is a meaningful predictor of vote choice in contemporary Italy. However, formal employment status might be significant in that the effect of perceived precarity might vary across employment categories. To account for this possibility, I run an additional model which includes an interaction term between perceived precarity and formal employment status (Model 6, Table A3 in the Appendix). The interaction term is not statistically significant, with one exception: the positive effect of permanent employment on the probability to support a mainstream right party over a mainstream left party is stronger for those who perceive their condition as precarious than for those who do not. This result is in line with the consideration that the reforms promoted by the PD since 2014 have directly undermined the employment security of the permanently employed, which makes it reasonable for the negative effect of permanent employment on mainstream left support to be stronger for those who perceive their condition as insecure.

In conclusion, I test whether the impact of precarity on voting behavior is conditional on income security, that is, stronger among the economically vulnerable and weaker among the financially sound, by running a model which includes an interaction term between precarity and economic hardship (Model 8). Results, displayed in Table 2, contradict hypothesis 3: interaction coefficients are non-significant, which indicates that the effect of precarity on vote choice is not conditional on economic hardship. This suggests that precarity has become relevant in shaping vote choice regardless of whether it is accompanied by financial distress.

Robustness checks

In order to increase confidence in the results, I replicate the analysis employing a series of alternative model specifications. First, I replicate models 2 to 4 while coding the League as a mainstream right party. This allows testing whether Brothers of Italy, the only party other than the M5S without government experience, attracted the support of precarious workers. Results, displayed in Tables S1 and S2 in the Online Appendix, are consistent with the results from the main models. The effects of precarity on support for the M5S and the mainstream left remain significant, while no significant effect emerges for the challenger or mainstream right. This reinforces the conclusion that precarious workers did not turn indistinctly to anti-establishment and opposition parties, but to the one challenger that committed itself to the promotion of policies that directly safeguard their interests.

Second, I replicate models 1 to 4 on the sub-sample of respondents excluding the unemployed (Tables S3 and S4 in the Online Appendix). This is necessary because the decision to include the unemployed in the ‘afraid’ category is based on the assumption that unemployed individuals perceive their condition as precarious, but entails the risk that the correlation between precarity and vote choice may be driven by the inclusion of the unemployed in the ‘afraid’ category. Once again, however, coefficient estimates retain direction and significance.

Finally, I replicate the analysis while including respondents' recall of their vote choice in the 2013 general election among the controls (Tables S5 and S6 in the Online Appendix). Albeit with reduced significance, the main effects are confirmed after this rather demanding robustness check.