2.1. Surveys on attitudes towards basic income

High-quality evidence on attitudes towards basic income is limited. In the 2000s, a few surveys investigated public opinion on basic income in Nordic countries.Footnote ¹ Nationwide and representative opinion surveys on basic income were conducted in Finland and Sweden based on the same questions in 2002 (Andersson and Kangas, Reference Andersson, Kangas, Kildal and Kuhnle2005). The item on attitudes towards basic income was as follows: “What do you think about a system that would automatically guarantee a certain basic income to all permanent residents?” Approximately two-thirds of the Finnish respondents responded in favour of basic income compared to only 46% of Swedish respondents. The authors interpreted this difference in terms of political power of the social democratic party, which is generally against basic income and was stronger in Sweden, and Swedish people’s belief that the Swedish welfare state is the best example of a universalistic welfare regime. A Norwegian study conducted in 2003 included the same question to collect opinions on basic income (Bay and Pedersen, Reference Bay and Pedersen2006). The survey revealed that two-thirds of the Norwegian electorate favoured a basic income system, although approximately 30% of the electorate exhibited negative attitudes towards it, which the authors ascribed to the Norwegian oil economy, strong culture of solidarity in the country, and social policy traditions of Norway that emphasise flat-rate benefits and general taxation.

Recently, multiple public opinion surveys on UBI have been conducted in European countries.Footnote ² Most of these surveys, however, have not yielded data suitable for in-depth analysis, but the ESS Round 8 Data (2016) are suitable for exploring various factors associated with individual UBI preferences. The survey asked respondents in 21 European countries, as well as in Israel and Russia, about their attitudes towards UBI by explaining the concept, which is based on the definitions provided by Basic Income Earth Network (2019) and Citizen’s Basic Income Trust (2019). The survey results indicated that, on average, almost half of the respondents supported the introduction of a UBI scheme in the 21 European countries, although there were distinct differences in approval ratings across countries (Lee, Reference Lee2018). A limitation of the questionnaire items was that they did not specify the quantum of monthly basic income and the extent to which existing welfare benefits will be replaced with UBI. This ambiguity made it difficult for the respondents to expect who would benefit under the new system. Nevertheless, the question clearly indicated that the mentioned UBI scheme is a new social security scheme based on universality, unconditionality, and individuality and that it aims to improve poor people’s standard of living and enhance income redistribution.

2.2. Expected effects and simulations of UBI schemes

Haagh and Rohregger (Reference Haagh and Rohregger2019) reported that UBI schemes can commonly be expected to reduce economic inequality, increase self-motivation for labour market participation, improve administrative efficiency, and strengthen social protections for workers vulnerable to labour market transformations. These expected effects can attract the attention of the precarious workers addressed in this study. First, given that most people in precarious work face economic inequality, they would, at the very least, be likely to favour the potential redistributive effect of UBI. Second, many people could more voluntarily take up part-time jobs while relying on UBI. Third, if UBI schemes can increase administrative efficiency, every person can experience reduced bureaucratic red tape when receiving welfare benefits. Last, because a considerable proportion of non-standard workers are not entitled to social security benefits (Matsaganis et al., Reference Matsaganis, Özdemir, Ward and Zavakou2016), UBI schemes can be considered a possible alternative to protect workers who are outside the social security net in terms of welfare reform. Therefore, we examine the relationship between each type of precarious work and UBI preferences by considering the extents of interest of various groups of precarious workers in the expected effects.

Although UBI experiments have been conducted in many countries, it is difficult to find reliable scientific evidence to prove any of the reported effects. However, multiple microsimulation studies on diverse types of UBI schemes have demonstrated that such schemes are expected to have a stronger redistributive effect than existing income security systems.Footnote ³ Kela’s (2016) microsimulation in Finland demonstrated that full UBI models with generous payment levels, such as €1,000 and €1,500 per month, can substantially reduce income inequality and the number of low-income households, and partial UBI models can be effective for narrowing income gaps, unless the level of payment is lower than the current basic social transfers. Especially, it was predicted that the introduction of UBI schemes can benefit low-income wage earners in Finland (Kela, 2016). Torry (2014; Reference Torry2017) and Martinelli (Reference Martinelli2019) estimated the distributional effects of UBI models, which can be considered in the UK, by employing EUROMOD, a tax-benefit microsimulation model for the European Union. Torry (Reference Torry2017) revealed that partial UBI schemes with modest payments have a stronger positive effect on income redistribution than the existing benefits system.Footnote ⁴ In another study, Torry (Reference Torry2014) demonstrated that full schemes with relatively low payment levels could encourage people to additionally participate in employment by reducing means-tested benefits, although low-income households would incur losses. Martinelli (Reference Martinelli2019), who explored three UBI models—a full scheme with a moderate payment level, a generous full scheme, and a partial scheme with a modest payment level—found that all of these schemes could potentially reduce the poverty rate and income inequality. Moreover, the microsimulation indicated that all the three models would improve working-age poverty rate on average, although a majority of working-age households without children would likely experience income loss (Martinelli, Reference Martinelli2019). As a result, microsimulation studies are in line with public expectations on the redistributive effect of UBI schemes (Haagh and Rohregger, Reference Haagh and Rohregger2019).

It would be very difficult for ordinary people to anticipate the specific and extensive results calculated using these microsimulation models and to precisely estimate their personal benefits without the specifications of UBI schemes. Nevertheless, survey participants can decide whether they are for or against UBI without such precise information by relying on the aims and features of the new system provided by survey items.

2.3. Precarious workers’ preferences for UBI schemes

Both part-time positions and temporary employment tend to increase income and job insecurity and poverty risks (Burgoon and Dekker, Reference Burgoon and Dekker2010; Horemans and Marx, Reference Horemans, Marx, Fritz and Koch2013; Van Lancker, Reference Van Lancker, Fritz and Koch2013). In addition, people in these types of work are likely to face difficulties when accessing social security benefits, such as unemployment benefits and pension, in most EU member states owing to frequent marginalisation from the labour market or very few working hours (Matsaganis et al., Reference Matsaganis, Özdemir, Ward and Zavakou2016). Schulze Buschoff and Protsch (Reference Schulze Buschoff and Protsch2008) pointed out that social insurance systems protect atypical workers to a lesser extent than they protect standard employees. Because of such economic insecurity, part-time workers and temporary workers can express stronger tendencies towards the expansion of redistribution, and exclusion from the social safety net can motivate their desire for restructuring the existing social security systems. Although UBI is not the only alternative for realising their demands, overall, the factors would cause these workers to take a positive view of UBI. Especially because temporary employees are more likely to become unemployed than permanent workers, they may regard unconditional and periodic UBI payments as a reliable support system that can support their uncertain future. Part-time employees, who usually earn relatively low hourly wages, have shorter job tenures, and have access to limited job opportunities (Horemans and Marx, Reference Horemans, Marx, Fritz and Koch2013), could experience lower levels of stress and worry because of not having full-time jobs by relying on the additional income provided by UBI schemes. In consideration of all these points, our hypotheses related to part-time workers and temporary employees are as follows:

The literature indicates that unskilled or low-skilled workers tend to be more supportive of government redistribution and welfare policies than highly skilled or administrative workers (Jæger, Reference Jæger2006; Linos and West, Reference Linos and West2003; Svallfors, Reference Svallfors2004; Wren and Rehm, Reference Wren, Rehm and Wren2013). This trend is understandable because these groups of employees are the most vulnerable in all countries (Häusermann et al., Reference Häusermann, Kurer and Schwander2016). Furthermore, as jobs in the service sector that require lower skill levels become increasingly vulnerable to labour market transformations (Oesch, Reference Oesch2013), low-skilled service employees are likely to earn low wages and experience poor working conditions. This could incentivise low-skilled service workers to support UBI given its potential for redistribution. In terms of the demand for welfare reforms, however, low-skilled service workers would be less likely to demand the reorganisation of social protection than other types of precarious workers because they can generally qualify for all welfare benefits as other standard workers, unless they are on a fixed-term contract or in a part-time position. Considering that there are many alternative redistribution measures other than UBI and that the desire of this group of workers for the overall restructuring of social security is not strong, they are unlikely to support UBI more actively than other workers. Therefore, the hypothesis regarding low-skilled service employees is as follows:

Studies have demonstrated that a substantial proportion of solo self-employed workers have suffered from financial hardships. Halleröd et al. (Reference Halleröd, Ekbrand and Bengtsson2015) revealed that considerable numbers of the working poor in Europe are self-employed, and most of them do not have other employees. Schulze Buschoff and Protsch (Reference Schulze Buschoff and Protsch2008) demonstrated that solo self-employed workers tend to change their employment status frequently and are at a high risk of becoming unemployed. Contrary to self-employed individuals running stable businesses, solo entrepreneurs tend to start their businesses involuntarily; often rely on more irregular, potentially lower income; and are less likely to be adequately covered and protected by earnings-related social insurance or pension schemes (Schulze Buschoff and Protsch, Reference Schulze Buschoff and Protsch2008; Dekker, Reference Dekker2010; Jansen, Reference Jansen2016; Pedersini and Coletto, Reference Pedersini and Coletto2009). These characteristics of solo self-employment, which seem similar to those of temporary employment, can induce favourable attitudes towards UBI. However, Jansen (Reference Jansen2016) revealed a stark contrast in pro-welfare attitudes between solo self-employed workers and temporary workers. This may be ascribed to the fact that self-employed workers are assumed to prefer “free markets and a low level of social protection because they depend on flexible labour markets and often on relatively low-paid workers” (Iversen and Soskice, Reference Iversen and Soskice2001, p. 883). Jansen (Reference Jansen2016) argued that because solo self-employed workers are potential employers, they would consider temporary workers as an important source of employment for their businesses and be reluctant to extend social security. Consequently, their negative attitudes towards welfare policies may neutralise their favourable impressions towards UBI, which can be ascribed to their economic vulnerability and low level of social protection. Hence, the hypothesis for solo self-employed workers is as follows:

We expect that part-time workers’ and temporary employees’ income insecurity and job uncertainty would play an important role in making them support UBI. This means that these two factors would function as intermediate variables between those types of precarious work and UBI preferences. Thus, the following hypothesis is formulated to verify whether income and job insecurity are parts of the mechanism in the relationships between the dependent and independent variables:

2.4. Unemployment benefit generosity and UBI preferences

Nowadays most welfare states operate unemployment insurance schemes to protect workers from income insecurity in the event of unemployment, but the introduction of a UBI scheme could probably transform unemployment insurance systems. Because most workers—even the self-employed in some countries—have unemployment insurance, whether voluntary or compulsory, they are likely to be approximately aware of the possible unemployment benefits that they could receive. Thus, the characteristics of existing unemployment insurance schemes might affect their UBI preferences.

Importantly, information about the specifications of UBI schemes is vague, and debates on the topic are still underway in all European countries. In other words, the situation is uncertain in this regard. Kahneman and Tversky (Reference Kahneman and Tversky1984) showed that people tend to favour stability over change and try to avoid losses over acquiring possible gains. It seems likely that workers would be reluctant to support the introduction of UBI schemes if they feel that the current level of unemployment benefit is satisfactory. Otherwise, it is possible that they would be in favour of the new system. In addition, Jæger (Reference Jæger2006) demonstrated the relative level of unemployment benefits to be negatively associated with individual preferences for redistribution. This can be interpreted to mean that generous unemployment benefits would undermine people’s support for the expansion of a redistributive policy. Therefore, we predict that the more generous an unemployment benefit scheme, the less supportive would workers be about the introduction of a UBI scheme.

When measuring the generosity of unemployment benefits, focusing only on the level of benefits in terms of their income replacement rate is inadequate, although many economic analyses of the generosity of such benefits tend to concentrate only on this indicator. The conditions of the entitlement, duration of benefit payments, and, especially, the qualifying period, are powerful indicators as well (Kuitto, Reference Kuitto2018; Scruggs, Reference Scruggs, Siegal and Clasen2007). First, the replacement rate refers to the extent to which unemployment benefits replace recipients’ income from employment. Thus, the higher the replacement rate, the more generous is the unemployment benefit. Next, the payment duration indicates the period for which unemployment benefit is paid to an unemployed person. This duration varies substantially across countries, even when their unemployment insurance systems exhibit similar income replacement rates. The longer the payment duration, the more generous is the unemployment benefit. Lastly, the qualifying period for unemployment benefit is the period of employment or the contribution required to gain entitlement. This implies that the longer the qualifying period, the less generous is the benefit. Consequently, the hypotheses associated with the generosity of unemployment benefits are as follows:

To sum up, Figure 1 displays our study framework.

FIGURE 1. Analytical framework: individual- and country-level determinants of universal basic income preferences.

4.1. Descriptive findings

Table 1 demonstrates standard and precarious workers’ approval ratings by country. European standard workers’ opinion on the debate about the introduction of UBI schemes is very strained because, in the 21 countries, their average approval rating for UBI is 49%. The proportion of part-time workers who are in favour of UBI is 49.5%, which is only marginally different from that of standard workers, although a higher proportion of part-timers are supportive of the scheme than standard employees in Finland, France, Iceland, the Netherlands, Portugal, Slovenia, and Sweden. By contrast, in all countries except Belgium, temporary employees’ approval ratings for UBI are higher than those of standard employees, and the average number across countries is 55.4%, which is higher than the figure for standard workers. Low-skilled service employees’ and solo self-employed workers’ average approval ratings are the same at 51.2%, which is only 2.2 p.p. higher than that of standard employees. A comparison of the public opinion in each country reveals that low-skilled service workers have considerably higher approval ratings for UBI in the Czech Republic, Estonia, Iceland, and Slovenia and the solo self-employed in Estonia, Germany, Iceland, Lithuania, and Switzerland. In addition, the results demonstrate that in 10 of the 21 countries, a higher proportion of solo self-employed workers were in favour of UBI schemes than standard workers, while the opposite is true in the other 11 countries.

TABLE 1. Standard and precarious workers’ approval ratings for UBI by country, 2016

Source: Authors’ calculations by using data from the ESS round 8 (2016)

Note: The calculations were conducted with only the samples in paid work for the 21 countries from the data. Post-stratification and country weights were applied to the calculation and the values that correspond to “Don”t know” and “Refuse to answer” were excluded from the calculation. Standard workers mean employees who have a permanent full-time contract.

Table 2 displays the correlation coefficients between country-level variables and their variance inflation factors (VIFs) in the multilevel logistic model including all individual-level variables. The highest correlation is found between the log of GDP per capita and public social expenditure and its coefficient is approximately 0.6, whereas most other correlations are considerably low. Moreover, since all the VIFs are lower than 5, it is found that country-level items are not highly correlated to each other in the model. Thus, it does not seem that multicollinearity between them would be problematic.

TABLE 2. Correlation coefficients between and VIFs of country-level variables

Note: The VIFs were calculated based on the final logistic model including all individual-level variables.

4.2. Regression findings

Table 3 displays the results of our logistic models with country fixed effects. Model 1, which includes only the explanatory and control variables, provides a statistical basis for testing the hypotheses related to the effects of different forms of precarious work (H1a, H1b, H1c, and H1d). For the hypothesis on mediation (H2), Models 2 and 3, which include household income and the likelihood of unemployment variables, respectively, in addition to the variables in the first model, and Model 4, which includes both aforementioned variables in addition to the variables in the first model, were estimated. The hypotheses associated with unemployment benefit generosity (H3a, H3b and H3c) are tested using the full model summarised in Table 4.

TABLE 3. Logistic regression on UBI preferences with country fixed effects

Note: The results in Models 2 and 4 are robust, when the models include no response category to household income variable.

Ref. = reference category; BIC: Bayesian Information Criterion.

Odds ratios and robust standard errors (in parentheses).

*p < 0.05; **p < 0.01; ***p < 0.001.

TABLE 4. Random intercept model on UBI preferences: null and full models

Ref. = reference category; BIC: Bayesian Information Criterion.

Odds ratios and robust standard errors (in parentheses).

*p < 0.05; **p < 0.01; ***p < 0.001.

Model 1 indicates that among the four types of precarious work, only temporary employment is significantly positively associated with UBI preferences, while the other precarious positions do not have a significant impact. Therefore, H1b (i.e. temporary employees are more likely to favour UBI schemes than permanent employees), H1c (i.e. low-skilled service sector employees do not show significant differences in attitudes towards UBI schemes from other workers.) and H1d (i.e. solo self-employed workers do not show significant differences in attitudes towards UBI schemes from permanent employees) are supported by the findings, but H1a (i.e. part-time workers are more likely to favour UBI schemes than full-time workers) does not receive support. These results are robust when we fit ordered logistic models and linear probability models with the original ordinal variable of basic income preferences. However, logistic models without weights indicate that both part-time employment and temporary work have significantly positive effects on UBI preferences, thus lending weak support for hypothesis H1a.

Models 2 and 3 demonstrate that the mediating variables, household income and the subjective likelihood of unemployment, are significantly associated with UBI preferences and attenuate the coefficient of temporary employment.Footnote ⁷ Consequently, H2 (i.e. being a part-time worker or temporary employee increases income insecurity and subjective unemployment risk, which, in turn, positively influence support for UBI schemes) is supported only for temporary employees. Model 4 indicates that household income and the likelihood of unemployment are significantly associated with UBI preferences even when they control for each other: the higher the household income, the weaker is the support for UBI, and the higher the subjective unemployment risk, the stronger is the support for UBI. Moreover, being an employer and of a higher age are found to be associated with a negative opinion about UBI. A comparison between the coefficients of temporary employment in Models 2 and 3 reveal the mediating impact of employment insecurity to be stronger than that of income insecurity.

Table 4 displays the estimates of random intercept models. The full model reveals the net replacement rate and payment duration to be significantly negatively associated with UBI preferences and qualifying period to significantly positively influence the dependent variable. In addition, the estimates based on the two-step approach confirm the findings (see Appendix Table A5). Consequently, the three hypotheses pertaining to the generosity of unemployment benefits are supported by the findings. Multilevel models, including random slopes on the individual-level explanatory variables, do not differ substantially from the random intercept models. Especially, the variances of random slopes on part-time work and temporary employment are nearly zero.Footnote ⁸

In addition to the hypothesis tests, country-level control variables throw up a few interesting results: the findings demonstrate that public social expenditure is significantly positively related to workers’ UBI preferences, whereas GDP per capita and population size are significantly negatively related to workers’ UBI preferences.