Farewell to the rightist self-employed? ‘New self-employment’ and political alignments

As mentioned, the social class literature is preoccupied with the assumption that the “petty bourgeois” prefer a limited role of governments vis-a-vis markets, and would have right-wing political attitudes. By using the term “petty bourgeois” class theorists refer to the group in-between “capitalists”, who own the means of production and hirer workers, and “workers”, who do not own the means of production and sell their labor power to capitalists (De Witte and Klandermans, 2000, p. 17). The petty bourgeois are self-employed persons who own small-scale capital and who produce and trade goods often without employing others, except sometimes unpaid family labor (Scase 1982). According to Arum and Müller (2004, p. 430) traditional petty bourgeois self-employment comprised “craftsmen, small shop-owners, proprietors and others working on their own in skilled occupations.” Typically, the self-employed would be associated with intergenerational inheritance and a high degree of stability over the life course (Blau and Duncan, 1967, p. 41), giving them a clear interest in conservative politics aimed at preserving the status quo of capitalist markets.

Class voting studies have put forward that the self-employed are also at the heart of extreme-right ideologies (Lipset, 1959; Norris, 2005). The radical tendencies among the petty bourgeois is sometimes attributed to their “contradictory” class position (Wright, 1997, p. 122): being distinct from both capitalists and wage-laborers they would be “trapped between the threat of big business and manufacturing industry, on the one hand, and the collective strength of organized labor, on the other” (Norris, 2005, p. 131). Discussing the social basis of fascist and other reactionary movements in Europe, Manza and Brooks (1999, p. 69) note that “many interpretations of right-wing movements have suggested that these movements draw disproportionate support from a ‘frustrated’ petty-bourgeoisie whose members are often among the first to suffer from economic downturns.” The self-employed are also prominent in more recent studies on anti-statist movements, such as the Tea Party in the United States. Skocpol and Williamson (2012) argue that among Tea Partiers, business is often idealized and that many Tea Party members are small business people.

With its recent decline, the “petty bourgeois” no longer accurately represent self-employment in modern democracies. To interpret the changing nature of self-employment, scholars have increasingly used the label “new self-employment”. Although the label has been used in different ways, as a common denominator it usually refers to self-employed workers which are believed not to correspond to the traditional profile of business-owners. A group, according to Buschoff and Schmidt (2009), that works “on their own account and without employees, often in professions with only low capital requirements. A growing share of these workers can be found on the one hand in ‘modern’ service-sector branches (such as education, health, financial and enterprise services) and on the other hand in the construction industry (via outsourcing and subcontracting)” (2009, p. 147).

In social stratification research, the term “new” self-employment has been used to denote the growth of self-employment in new occupational areas. This is most systematically done by Arum and Müller (2004) who distinguish two new types from traditional self-employment², i.e., unskilled self-employment and self-employed professionals and managers. Instead selling goods, self-employed workers in these new types would increasingly provide services and sell their labor power as freelancers or independent contractors (Kösters et al, 2013). Since the mid-1990s, the majority of people in self-employment work in either one of the two categories (Arum and Müller, 2004). Self-employment has therefore become more diverse and polarized in terms of economic well-being and social status. Compared to other categories, unskilled self-employment is associated with the highest labor market risks, as it is generally more instable and more poorly paid (McManus, 2000; Arum and Müller, 2004; Lofstrom, 2013), and – because many immigrants in self-employment hold lower-skilled jobs (Kanas et al, 2009) – also most ethnically diverse. Like manual workers in wage-employment, most self-employed workers do not own property and sell their labor power at a wage-like rate. Therefore, this form of self-employment is sometimes considered to be part of, or at least closer to, the working class instead of the petty bourgeois (Form, 1982). By contrast, those in professional and managerial occupations (typically including liberal professions) are generally assumed to be highly educated and profitable (Barbieri, 2003).

To account for the heterogeneity of self-employment, this study further differentiates the Arum-Müller classification. In doing so, inspiration is drawn from “new class theories”. As a single sub-category, professional-managerial self-employment misrepresents fundamental occupational differences between “technocratic” jobs (i.e., the “old middle class”) and a new class of “social-cultural specialists” (Güveli, 2006; Oesch, 2008). Unlike the former, people in the latter category hold occupations that involve providing social services and/or using knowledge on social-cultural issues and conducting tasks that are relatively difficult to monitor. Social-cultural specialists, compared to rightist-oriented technocrats, have developed a strong preferences for leftist, and in particular new-left and green politics (Güveli et al, 2007; Oesch, 2008; Jansen et al, 2011). In self-employment, social-cultural specialists may be found in occupations that are characterized by a combination of relative autonomy, specialized social-cultural knowledge, and tasks that are not directly instrumental to profit-maximization (e.g., authors, journalists and other writers, or creative and cultural workers, and self-employed health professionals). Although working on their own account, they often provide services to the (semi-)public or non-profit field or other domains that receive state support. With these characteristics, the new group of social-cultural specialists hold self-employment jobs that strongly deviate from traditional “petty bourgeois” self-employment. These differences can be expected to be greater than for self-employed technocrats (e.g., independent accountants and other financial and legal specialists, civil engineers and self-employed ICT professionals), who, like the petty bourgeois, drive businesses that are more directly related to profit-maximization and whose tasks do not involve providing social services or social-cultural knowledge.

Inside the domain of industrial relations studies, “new” self-employment has largely been examined through the lens of precarious work, linked to the flexibilization of labor markets (e.g., Stanwoth and Stanworth, 1995; Buschoff and Schmidt, 2009; Kalleberg, 2011). Compared to standard employees (i.e., with secure employment contracts), freelancers, sub-contractors and others in quasi-independent employment relationships are exposed to greater risks because they – sometimes involuntary – bear responsibility for market changes, enjoy lower levels of social protection, and have less capacity for savings, insurance and pensions (Buschoff and Schmidt, 2009; Dekker 2010a; Kautonen et al, 2010; Pedersini and Coletto, 2010). What is even more explicit in this stream of literature, is that risks and uncertainties cut across occupational lines. Also highly educated professionals may work in semi-autonomous, low-paid, or precarious self-employment jobs, without legal protection (King, 2014).

Building on the insider–outsider literature, insecure forms of self-employment may be thought of as “outsiders” on the labor market. To study insecurities associated with self-employment, indicators may be borrowed from the insider/outsider-model by Rueda (2007, p. 15), who focusses on the precarious nature and involuntary nature of the work relationship, i.e., Like precarious wage-employment, self-employed precariousness should be characterized by insecure and unstable work. In the case of self-employment that is most likely to be reflected through irregular orders and income, or little financial buffers. Second, most precarious workers are assumed to aspire more secure jobs. For people in self-employment that might be the case when someone prefers to have a job in wage-employment. A third aspect, however, may be added: the dependent nature of the (self-)employment relationship. Some workers, although formally in self-employment, work in hierarchical subordination to a single firm, on which they are economically dependent (Muehlberger, 2007). These arrangements can be associated with high labor market risks, in particular when self-employed workers are allocated tasks over which they have little control (Standing, 2011, p. 16). In such cases, they bear the entrepreneurial risk without entrepreneurial independence.

Given the increased heterogeneity and presumed precarization of self-employment, the political alignments of the self-employed can be expected to have become more diffuse, and less strongly attached to liberal-conservative politics. Below, two sets of hypotheses are formulated on how disparate types of self-employment may translate differently into political orientations. The first category of expectations departs from the assumption that political preferences are for an important part shaped by self-interest (Reeskens and Van Oorschot, 2011), and that economic insecurities would be associated with left-wing political support. The second scenario builds on the assumption that people with marginal positions on the labor market move away mainstream politics. From this perspective, labor market vulnerabilities might not be associated with left-wing political support, but rather with radical-righ support or political alienation.

Policy and party preferences may be formed by a mixture of normative, social and rational motives (Van Oorschot, 2000; Mau, 2003). With respect to labor market vulnerabilities rational self-interest provides a straightforward potential explanation for left-wing attitudes. This understanding of self-interest involves that welfare consumers (or, at least potential consumers) would be more likely to support welfare policies (Kangas, 1997; Reeskens and Van Oorschot, 2011): People who generally run higher risks of being unemployed or otherwise not being able to provide for themselves are assumed to support welfare and governmental redistribution because they may need protection themselves in the future (Corbetta and Colloca, 2013; Marx, 2014; Marx and Picot, 2013). Along these lines, recent studies comparing temporary and permanent workers have suggested that temporary workers are more likely to support (new) left-wing parties (Corbetta and Colloca, 2013; Marx, 2014; Marx and Picot, 2013; Emmenegger et al, 2015). Following this line of thought, the political preferences of people in new self-employment might be similar to others in “non-standard” work relationships. Rather than stability, which is the driver of petty bourgeois conservatism, “new” self-employment would be characterized by instability and insecurity. It may therefore be expected that self-employed persons with insecure conditions demand more inclusive welfare systems, for example, in the domains of pensions or health care³ (Dekker, 2010b) and generally hold more left-wing attitudes. To examine which segments of the self-employed are the most left-wing, hypothesis 1 assumes that precarious self-employment is (a) insecure rather than secure, (b) involuntary rather than voluntary, and (c) dependent rather than autonomous.

Hypothesis 2 formulates a similar expectation related to occupational categories. This hypothesis differentiates between the “new” self-employment types (i.e., unskilled self-employment and social-cultural specialists), on the one hand, and the “old” petty bourgeois and technocratic professionals, on the other hand. Again, the political preferences of people in unskilled self-employment might be motivated by self-interest, as this category is generally associated with the highest labor market risks. With lower incomes, less capacity for savings, and little insurance against economic hardship the support for welfare arrangements and left-wing politics may be stronger among unskilled self-employed workers, whereas their support for rightist politics may be weaker. For social-cultural professionals, however, left-wing political attitudes may only partially be explained by self-interest. On the one hand, like technocratic professionals, social-cultural specialists may not be the potential users of welfare state services because they tend to work in higher educated jobs and may be economically well-off. Self-interest would therefore not necessarily dictate pro-welfare attitudes. Yet, it should be noted that many in this category work in competitive sectors with low entry barriers (e.g., the media industry or other creative sectors), where pressure on tariffs and hourly earnings reduce the capacity to save and build financial buffers. So, also for this group vulnerabilities may be associated with the support of left-wing parties and policies. Yet, next to rationalist motives, also ideological and normative motives may play a role here. In particular, this category of social and cultural professionals may be drawn to new-left parties, as their tasks involve specialized social-cultural knowledge stimulating non-economic, post-materialist values (Güveli et al, 2007; Oesch, 2008). As a result of their occupational tasks this group may have developed a distinct understanding of the ‘functional and moral necessity’ (Van Oorschot 2000, p. 20) of contributing to common welfare, leading to a left, and in particular new-left political orientation (Jansen et al, 2011). Acknowledging the relevance of task structures for the normative aspects of political preference formation, does not rule out the potential impact of self-interest: Because the services provided by social-cultural specialists are often in the (semi-)public or non-profit domain (such as education and healthcare), their interest to support left-wing politics may still, partially, be based on their role as producers (or co-producers) of social and cultural services. Yet, such different lines of reasoning result in similar expectations about the link between self-employment in social-cultural professions and left-wing preferences.

Alternatively, for precarious self-employment neither a conservative nor a left-wing orientation may be straightforward. On the basis of the insider–outsider model, right-wing conservative parties often oppose labor market policies that are beneficial to “outsiders” [i.e., policies that advance the creation of jobs or compensate potential job loss (Rueda, 2007)], because these policies imply an intrusive role of government and higher taxes. Yet, this is not to say that “self-employed outsiders” would turn leftward. Insider–outsider theories postulate that left-wing parties – and social democrats in particular – defend the interest of insiders rather than the interest of outsiders (Rueda, 2007; Lindvall and Rueda, 2014). People in self-employment do not benefit from strong protection of wage-employment. On the contrary, as long as they depend on, for example, short-term freelance contracts, strict hiring regulations might even prevent their access to work. For precarious self-employed workers this implies that their interest are represented by neither left-wing, nor right-wing parties. This line of reasoning is consistent with the general thesis that economically marginalized groups (e.g., the ‘losers’ of globalization, Kriesi et al, 2008) may become politically alienated, and might abstain from participation in general or turn away from mainstream parties in particular. A movement to the radical right may be expected to be driven by political dissatisfaction (De Witte and Klandermans, 2000). Emmenegger et al (2015), for example, find that labor market disadvantages may lead to support for radical political forces as they influence individual’s perception of the extent to which the political system is responsive to their personal needs. Especially radical-right wing politics are assumed to appeal strongly to those who are most insecure and alienated from mainstream parties. The ideological blend of radical-right parties [i.e., combining a centrist economic program with welfare chauvinism and protectionism (Mudde, 2000)] may be especially attractive to those in vulnerable forms of self-employment. Self-employed individuals in economically and socially precarious conditions, may exhibit a “politics of resentment” by seeking to punish mainstream politics for their relative misfortunes (Wells and Watson, 2005; Standing, 2011). Hence, hypotheses 3 assumes that people in precarious self-employment are outsiders in the political system.

The hypotheses are tested using data from the Netherlands. The Netherlands are a representative case. With approximately 16 per cent the self-employment rate is around the European average (Kösters and Souren, 2014; OECD, 2015), including a typical downward trend in agricultural self-employment, and in retail and hospitality, and a rise in self-employment in service-oriented industries (Eurostat, 2014). Like in other countries, more than half of the Dutch self-employed do not employ others (Van Stel et al, 2014; OECD, 2015), and the share of solo self-employed workers rapidly increased over the last decades. In 1996, approximately 330 thousand people were in solo self-employment, and this number amounted to almost a million in 2014 (CBS, 2014). This growth in solo self-employment is among the strongest is Europe (Kösters and Souren, 2014). Research on the Netherlands confirms that self-employment is a heterogeneous employment type, with substantial diversity not only in occupation, income and education, but also in access to social security and fiscal benefits (Bosch and Van Vuuren, 2010; Blumberg and De Graaf, 2004; CBS, 2014).

Moreover, the Netherland has a typical European multiparty system that allows to investigate political divisions in various ideological directions (Emmenegger et al, 2015). The Dutch political system has two mainstream conservative or liberal parties, the CDA (Christian Democrats) and VVD (Liberal Party). The VVD is most strongly associated with self-employment. After its formation in 1948, the VVD quickly got the name of being an “Entrepreneur’s Party”, mainly due to strong organizational and programmatic ties with the employer’s associations. This reputation largely persisted despite electoral expansion after the 1970s. Lucardi (1987, p. 165) concluded that although “the VVD is not the party of entrepreneurs, it probably is a party for entrepreneurs. The entrepreneur, as it were, remains her ideological example.” Also the CDA has traditional ties to employer’s associations, and in particular to agrarian organizations. Historically, the CDA and its predecessors – drew much supported from the so-called “kleine luyden”, a label used to denote craftsmen, shop owners, farmers and other petty bourgeois (Lucardi, 2007).

Data for this study were collected through the Solo Self-employment Panel (‘ZZP Panel’ in Dutch). The ZZP Panel is an annually repeated survey since 2009 among 3000 self-employed persons without employees organized by the Dutch Ministry of Social Affairs and research company Panteia. The panel is primarily designed to map the management, organization and continuity of solo self-employed businesses. The sample is stratified by 10 economic sectors, and panel members are recruited using an address file which is based on the registers of the Dutch Chamber of Commerce. The panel has 2 waves per year. Each first wave is conducted using a telephone survey and mainly covers a fixed set of questions. The second wave is an internet-based follow-up questionnaire, covering rotating topics. Panel attrition is dealt with by filling up the sample to 3000 once per year. For more information on the panel, see Hoevenagel et al (2015).

For this study, the waves of 2014 are used. First, some background characteristics were retrieved from the first wave. Recruitment through the 2013–1 wave and fresh sampling yielded a response rate of 41 percent (N = 3002). Second, specifically for this study, a tailor-made module on the political preferences was added to the second wave of 2014. This follow-up survey was fielded between mid-December 2014 and mid-January 2015. In total 2498 panel members who participated in the first wave were invited again, of which 851 completed the questionnaire. With 34 per cent the response rate is comparable to earlier follow-up surveys (Hoevenagel et al, 2015). Ultimately, the analyses are based on 815 respondents with valid information on all relevant variables. Descriptive statistics (including bivariate tests) are presented in Appendix A. By and large, the sample corresponds to recent findings by Statistics Netherlands that the solo self-employed are often higher educated, predominantly male and that the majority is between 35 and 55 years old (CBS 2014). A sector weight is applied to retain the same distribution of sectors compared to the first 2014-wave.

Political orientations are measured using two kinds of variables, policy and party orientations First, respondents were asked to rate the importance of particular policy issues on an 11-point scale. Two out of four⁴ issues relate to typical entrepreneurial interests, i.e.; (1) fewer rules and administrative burden for business, (2) lowering taxes for business. The other two issues relate to the alleged precarity of self-employment, i.e., (3) improving access to social security for solo self-employed (4) tackling false self-employment. Second, to measure party orientations vote propensity questions (Van der Eijk et al, 2006) are used rather categorical vote choice questions. Respondents were asked to rate on a scale from 0 (“will never vote for this party”) to 10 (“will certainly vote for this party”) how likely it is that they would ever vote for a party. The advantage of vote propensities is that propensity scores provide the same level of information on each party in the party system, regardless of its actual size or popularity. A multinomial outcome variable would result in low frequencies for particular parties, also theoretically interesting ones. Note that in Table 1 “vote intention” records only 26 respondents with respect to the main left-wing party (PvdA), and only 48 for the populist PVV. Such low number are inappropriate for logit regression. Instead of grouping parties together, propensity scores are more informative, as the continuous nature of the variable allows similar linear regressions for each individual party. As a proxy of political alienation a dichotomous variable for “non-voting” is created by indicating respondents with a low propensity to vote for any party (i.e., reporting a score lower than the scale’s midpoint of 5 for all parties).

Respondent’s occupation is measured in an open question by asking people to describe their business (i.e., what kind of goods they sell or what kind of services/labor they provide). On the basis of this information each respondent was assigned an occupational code (ISCO-08). Following a procedure similar to Arum and Müller (2004) these occupations are classified into traditional self-employment, manual self-employment and professional-managerial self-employment. Yet, based on Güveli’s (2006) distinctions, professionals and managers are further reclassified into sub-groups for self-employed technocrats and self-employed social-cultural specialists. An overview of the specific occupations allocated to each type is presented in Appendix B.

To measure precariousness, several items are used to assess whether the self-employed job is involuntary, insecure and dependent. First, involuntary self-employment is a dummy-variable indicating whether respondents would prefer to work in wage-employment.

Second, insecurity is measure using two variables: a dummy-variable is included for financial vulnerability, measured by a combination of the level of income and the availability of financial back-up (De Vries and Dekker, 2015). A dummy is created indicating whether a low level of income (i.e., <1500 euro) is combined with little to no financial buffers. Income is measured as respondent’s net monthly income derived from their business. The size of the financial buffer is assessed by asking how many months respondents would be able to make a living in case their business would stop generating income. Respondents with low income who would survive only “less than a month” or “1–6 months” are classified as vulnerable. Next, to measure the instable nature of self-employment a scale in constructed based on four items. Respondents were asked to rate on an 11-point scale whether (a) their income is stable (regular [0] vs. irregular [10]); (b) their clients/orders are stable (very fixed [0] vs. very volatile [10]) and whether (c) their income and (d) clients/orders are predictable (foreseeable in advance [0] vs. not foreseeable [10]). With a Cronbach’s alpha of 0.83, a reliability analysis confirmed that these items comprise a reliable scale. An index (range 0–1) was calculated by taking the mean over the four items divided by 10.

Third, to assess the dependent nature of self-employment, respondents were asked what portion of their annual turnover they drew from their most important client. A dummy variable was created to indicate whether a respondent was largely (i.e., for 75 percent or more) dependent on his/her most important customer. Finally, because turnover does not provide any information on a potential hierarchical work relationship, a job autonomy scale is added based on two item: Respondents were asked to rate on an 11-point scale whether (0) they themselves or (10) their client decided on (a) which tasks are carried out, and (b) how tasks are conducted. Again, an index (range 0–1, Cronbach’s alpha of 0.65) was calculated by taking the mean over items divided by 10⁵.

Finally, three control variables are included in the analysis, i.e., age, gender (female = 1), the number of years since in self-employment, and a dummy indicating whether the respondent has other sources of personal income than from his/her business (for example from wage-employment, social benefits, pensions etc.)

Table 2 shows the unstandardized regression effects of the perceived importance of self-employment policies on party orientations. Following the left–right divide, support for reducing taxes is associated with a stronger tendency to vote VVD (0.32), CDA (0.26) and PVV (0.22), and a weaker tendency to vote PvdA (−0.26), SP (−0.17), GL (−0.25). Moreover, support for tax cuts negatively relates to non-voting, but appears unrelated to the propensity to vote D66. Next, support for a reduction of rules and regulations is associated with stronger propensity to vote VDD (0.24), and a weaker tendency to vote PvdA (−0.19), SP (−0.16) or GL (−0.29), but is unrelated to voting CDA, D66 or PVV or abstention.

Support for social self-employment policies (i.e., enables access to social security for self-employed persons without employees), is generally related to a weaker tendency to vote right-wing parties (VVD[−0.36], CDA[−0.23] and D66 [−0.14]) and a stronger tendency to vote left-wing (PvdA [0.13], SP[0.16], GL[0.12]). Hence, the relationship between social policy preferences and party orientations is largely in the expected direction. Moreover, support for social security is positively (0.15) related to non-voting, but unrelated to voting PVV. Finally, preventing false self-employment is generally not associated with any party orientation, except positive effects on supporting CDA (0.11) and SP (0.08).

In the next step, Table 3 shows the unstandardized regression effects using the attitudes towards policy orientations as the dependent variables. Because the precariousness-based and occupation-based variables are to some extent related (see Appendix, Tables A4 and A5), first separate models are shown for both types of variables (Models A and B), before reporting the final model including the two sets simultaneously (Model C). From Table 3 it appears that occupational differences are related to support for self-employment policies. Compared to traditional types of self-employment, the support for typical entrepreneurial policies (i.e., fewer rules, lower taxes) is significantly lower among social-cultural specialists (−0.95 and −0.96) and – with respect to taxes [and albeit at the 90 per cent level of confidence] – manual self-employed (−0.47). Technocrats, as expected, do not appear to differ from traditional self-employed in their preferences for typical entrepreneurial policies. These observations may support hypotheses 2 that compared to traditional types, new types are less strongly oriented towards liberal-conservative politics. Yet, the data in Table 3 do not provide evidence for stronger support for social policies (i.e., opening up social security, preventing false self-employment) among social-cultural specialists or manual self-employed. Although technocrats are generally less supportive of social policies (−1.09 and −1.36) relative to traditional self-employment, other new types are not found more supportive. Like technocrats, social-cultural specialists even reject (−1.10) prevention of false self-employment.

Turning to variables indicating precariousness, the data in Table 3 show that involuntariness, dependence and insecurity are not uniformly associated with weaker liberal-conservative attitudes, or stronger left-wing politics (hypothesis 1). Financial vulnerability and perceived instability, for example, are related to a higher support for access to social security (0.86 and 0.89, respectively) or preventing false self-employment (0.65 for financial vulnerability). Some of these effects, however, partially or fully disappear when taking into account occupational differences. Instability, on the other hand, is also related to stronger support (0.78) for cutting the administrative burden for small businesses. Moreover, involuntary self-employment is associated with higher support for all kinds of self-employment policy, both typical entrepreneurial policies (fewer rules [0.66], lower taxes [0.62]), as well as social policies (on social security [0.78] and false self-employment [0.81]). In opposite direction, job autonomy is generally related to weaker support for all kinds of policies, but in particular regarding social policies.

The propensity to vote D66 shows a different picture, see Model 3. Compared to people in traditional self-employment, both professional-managerial types (that is, technocrats and social-cultural specialists alike) have a higher propensity to vote for this social-liberal party (1.42 and 1.51, respectively). Manual self-employed, on the other hand, are less supportive of D66 (−1.17). However, none of the ‘precariousness’ variables affect the propensity to vote D66. Again, these results indicate stronger support for hypothesis 2, than for hypothesis 1.

Next, Models 4 to 6 provides the regression estimates for the propensity to vote for a left-wing party (PvdA, SP and GL). Relative to people in traditional self-employment, social-cultural specialists have a higher propensity to vote PvdA (1.24) or GL (2.52). Manual workers are neither stronger nor weaker associated with left-wing voting. Compared to traditional self-employed, technocrats are somewhat less likely to vote SP, and somewhat more likely to vote GL. Hence, this results only partially supports hypothesis 2. Moreover, contrarily to hypothesis 1, involuntariness and instability are not associated with a higher propensity to vote left-wing. For GL the reverse seems true, a higher degree of instability is associated with a lower propensity to vote GroenLinks. Also the other results are mixed: Involuntariness is associated with stronger support for the SP (0.79), but not PvdA or GL. Financial vulnerability is associated with stronger support for GL (0.85), but not PvdA or SP. Interestingly, job autonomy is positively related to the propensity to vote PvdA (1.43), SP (0.76) and GL (1.12). This result indicates that – contrary to the expectation – self-employed are not more inclined to vote left-wing when they are more dependent, but rather when they enjoy greater autonomy in work. Dependent self-employment seems associated with a lower propensity to vote GL (−0.73) and, to some extent, SP (−0.58, uncontrolled for occupation).

Finally, in Model 7 and 8 hypothesis 3 is tested stating precarious self-employment is associated with either radical-right wing voting (PVV) or abstention (non-voting). It appears that in their support for the PVV manual self-employed workers do not differ from traditional self-employment. This correspondents to the idea that manual self-employed may share their radical tendencies with the “petty bourgeois”. Social-cultural professionals (−2.04) and technocrats (−1.89), on the other hand, strongly reject the PVV. With respect to precariousness, model 7 shows that only involuntary self-employment is associated with stronger PVV support (0.74). For the other elements of precariousness, hypothesis 3 does not seem to hold. In Model 8 the “propensity” of non-voting is estimated through a logit model. Compared to other types, unskilled self-employed workers are more likely to abstain from voting (1.44). Yet, none of the ‘precariousness’ identifiers is associated with abstention. Dependent self-employment seems even negatively (−0.82) associated with non-voting.