Abstract
The idea of a universal basic income (BI) is both radical and simple. Obtaining a sufficient citizenship-based income without work obligations is fundamentally opposing the foundations of the welfare systems that are in place nowadays. As BI has gained increasing attention in public debates and among policymakers, questions arise about its social legitimacy. This study is the first to analyse a broad range of explanatory individual and contextual factors that may affect popular support for BI. In addition, we study how BI support is related to support of current welfare provisions, to analyse how radically different people perceive a BI to be. We use a unique survey question – available for 23 European countries, from the recent release of the European Social Survey (2016) –that introduces BI with an extended definition, emphasizing its universal and unconditional character and that it will replace other benefits and services and is paid for by taxes. Results show relatively high, but varying levels of support among European countries and social groups. People who are in a more vulnerable socio-economic position support BI more, as well as political left-wingers, egalitarianists and people who support targeting benefits at the poor. Also, a BI is more supported in countries with higher levels of material deprivation. This pattern of relations on both the individual and contextual levels seems to suggest that it is not the universal character or its unconditionality that makes a BI so attractive to a large share of the European population, but the fact that it provides (poor) people with a guaranteed minimum income. We also find that people who support other welfare reforms are more supportive of a BI. This, and the fact that younger people are more pro-BI might give hope to BI advocates who present the proposal as a social system of the future.
Introduction
A universal basic income (BI) is an idea that over time has fascinated citizens, policymakers, politicians and scientists alike. The idea is radical and seemingly simple, but rests upon complex philosophies about society and solidarity, while it has a long history in political thought (Widerquist et al., 2013). Although a BI has been presented as a simple and unified idea, concrete BI policy proposals take many different forms, varying in goals, coverage, benefit levels, entitlement conditions and implementation trajectories (De Wispelaere and Stirton, 2004). In its most full-fledged form, a BI has a universal coverage (provided to every citizen), its level guarantees independent subsistence to each individual, and it is unconditionally provided (no obligations are attached). Obtaining a sufficient income without having to work, without showing the willingness to work or demonstrating that one is not (or no longer) able to work, is fundamentally opposing the foundations of the common welfare systems that are in place nowadays. Clearly, a full-fledged BI is a ‘radical departure from traditional welfare state policies’ (Widerquist et al., 2013).
The idea of a BI has long historical roots (e.g. it is discussed in Thomas Paine’s 1796 pamphlet on Agrarian Justice), but in the last decade, BI seems to have moved beyond a ‘political pipedream’ and has now entered political debates as a serious policy proposal (Organisation for Economic Co-operation and Development (OECD), 2017). Experimental trials with a variety of BI forms have been debated and introduced in Finland, the Netherlands, France, Canada and the United States (OECD, 2017). As BI has gained increasing attention in public debates and among policymakers, questions arise about what the general popular support for introducing a BI actually is, how it is related to individuals’ social positions and ideologies, and to the economic, institutional and cultural contexts in which they live.
From the research literature, we do know quite a bit about the individual and contextual determinants of welfare attitudes generally (Roosma et al., 2013; Svallfors, 2012), but understanding support for BI cannot be simply deduced from this literature. A BI is introduced as a radical alternative for the current social income provision in European welfare states, but the proposal is still not implemented in any of these countries, resulting in people having no experience with it and possibly having limited knowledge about the consequences of its possible implementation (that can take various forms). Understanding BI support is therefore not self-evident, that is, the usual theoretical explanations for differences in welfare attitudes that focus on self-interest and ideological motivations (Meier Jæger, 2006) are difficult to apply directly to explaining differences in BI support. For instance, a BI may not only be positively experienced by the most vulnerable, but also by those who welcome better opportunities for combining work and care; the BI is not an exclusive ‘leftist’ idea, since liberal right-wingers have pledged for a simple and minimal social security system (Widerquist et al., 2013) and a BI may have different social impacts in different welfare systems. Moreover, deservingness theory predicting that benefit schemes that are targeted at groups that are considered to be more deserving are more popular (Van Oorschot et al., 2017), has a complex relation to BI support as well; BI is aimed at both deserving and undeserving individuals. For instance, the scheme will also benefit those who are not in need of a benefit (as a BI is universal) and those who do not reciprocate (as a BI is unconditional). In addition, the existing studies that focus on explaining BI support from an individual’s characteristics are scarce and mostly lacking analytical depth due to data restraints. An important aim of this article is therefore to explore and understand the different aspects and predictors of BI support. For these reasons, our analyses will have an explorative rather than a hypothesis testing character.
We aim to answer the following research questions: (1) What is the level of support for a BI in different European countries? (2) Which groups of citizens are more or less in favour of it? (3) What are the country characteristics that influence BI support? In addition, to get a better understanding of whether and how people’s support for BI is embedded in their other welfare attitudes (which could give information on whether they actually perceive and evaluate BI as a radical alternative to the current welfare system), we add the question: (4) How is popular support for a BI associated with other welfare attitudes, such as people’s support for welfare provision, and their perceptions of the performance and outcomes of their welfare state?
As mentioned, ours is not the first study on support for a BI, but it is the first to analyse cross-nationally a broad range of explanatory individual and contextual factors that may affect it, which has become possible by the recent release of data from the European Social Survey (ESS) 2016, which is available for 23 European countries. Moreover, this is the first study that was able to use a cross-sectional survey question that introduces a comprehensive idea of BI and its different aspects: its guaranteed minimum income, its universal character, its unconditionality, that it replaces other benefits and services and that it is paid for by taxes. Using such a comprehensive, yet precise definition with different aspects has of course methodological implications that we will discuss in our methods section. Yet the clear advantage is that the abstract concept is now introduced in detail and refers directly to the most general definition of a BI that is used in the public policy debates about the proposal. As such, our findings may function as a frame of reference for studies that use less comprehensive definitions of BI.
In the next section, we first discuss previous findings about public support for a BI, the possible individual and contextual determinants for this support and the relations with other welfare attitudes. After introducing our data and methods, we present the results of our analyses and discuss them in the ‘Discussion and conclusion’ section, paying particular attention to the issue of how respondents seem to have interpreted the ESS BI question.
Public opinion on BI
Previous findings
Studies that asked people about their support for various forms of BI are scarce, and scattered over countries and time periods. Given the fact that BI opinions are a very recent subject of study, we will discuss their findings here briefly by way of overview of the state of knowledge.
As for levels of support, the existing studies learn that larger parts of national populations of European countries show support of around 50–80 percent for introducing a form of BI (Andersson and Kangas, 2002; Bay and Pedersen, 2006; Ipsos, 2017; Liebig and Mau, 2002; Stadelmann-Steffen and Dermont, 2017; Ullrich, 2002). 1 Ullrich (2002) reports that in a selection of the OECD countries, averages of above 50 percent agree with the general idea that the government should provide everyone with ‘a guaranteed basic income’. In Sweden, support levels of 45 percent are found for alternative BI schemes (Andersson and Kangas, 2002), while in other Nordic countries, support levels for BI schemes tend to be higher. Different studies report support in Finland ranging between 50 (Stadelmann-Steffen and Dermont, 2017) and 63–79 percent for various BI schemes (Andersson and Kangas, 2002), while support in Norway peaks around 66 percent (Bay and Pedersen, 2006). In a sample of employed people in Germany, 77 percent of respondents support a guaranteed minimum income (Liebig and Mau, 2002), and in the United Kingdom, 48 percent of the population said they would support an unconditional, universal BI at subsistence level (Ipsos, 2017). A European Union (EU)-comparative online survey study found that on average 68 percent of working age Europeans across 28 EU member states reported that they would vote for the introduction of a BI (Dalia, 2017). So, BI seems to be popular over time and across countries. However, under more detailed specifications, a BI might drop substantively in popularity. Especially, the suggestion that not only natives, but also immigrants would be covered by a BI lowered support levels (Bay and Pedersen, 2006; Linnanvirta et al., 2017; Stadelmann-Steffen and Dermont, 2017), and reference to the possible (cost) consequences of a BI, like increased taxes or reduced (other) benefits, had an eroding effect in the United Kingdom (Ipsos, 2017).
The question about which considerations motivate people to be in favour or against a BI is an important issue for understanding (lack of) public support. Of Dutch BI supporters, 79 percent report a practical reason, that due to a BI’s administrative simplicity there would be fewer opportunities for abuse (Van Oorschot, 1998). The main reasons mentioned against it were concerned with the work ethic and with deservingness considerations (it would also be given to people who do not need it and people do not have to reciprocate; Van Oorschot, 1998). The Ipsos UK study found as most important reasons to support a BI that it would offer more income security, reward unpaid work, guarantee an income in time of unemployment due to technology and automation, and simplify the present system and bureaucracy (Ipsos, 2017). Also, this study found support for a BI to be lower when the BI was presented as not being restricted to the needy and working or work-willing only. The EU-comparative study of Dalia (2017) found as most popular reasons in favour of a BI that it would reduce anxiety about financing basic needs, and create more equality of opportunity, while the reasons against include that a BI might encourage people to stop working, would attract migrants and should not been given to those who are not in need of it. Note that the differences in (types of) motives found by these studies are related to the fact that each asked interviewees to respond to a different set of pre-structured alternatives.
Possible individual characteristics influencing BI support
With Martinelli (2017), we assume most generally that, as in the case of other welfare attitudes (Meier Jæger, 2006), BI opinions are affected by the interests people have in the provision, and by their ideological stances. The two empirical studies with analyses of individual-level determinants align in finding that people with weaker socio-economic positions and people with a more left-wing political stance tend to be (somewhat) more in favour of a BI (Andersson and Kangas, 2002; Stadelmann-Steffen and Dermont, 2017). This suggests that (perceived) personal need and insecurity, as well as a preference for social equality affects BI opinions, aligning with the results from welfare attitude studies generally (Meier Jæger, 2006; Roosma et al., 2014). However, Linnanvirta et al. (2017) find that among their sample of poor Fins, the level of support for a BI, as well as the motives for and against it, do not differ substantively from the Finnish general population. Also, from a theoretical perspective, it is not immediately clear what kind of interest indicators and ideological positions would play which role.
As for interests, there are suggestions in the literature that specific groups would benefit relatively most from a BI, and would therefore be more supportive of it. BI would advantage especially people in more economically precarious situations, such as workers with more insecure working patterns, low-paid and part-time workers and people on (means-tested) minimum income benefits (Martinelli, 2017). Also, Van Parijs and Vanderborght (2017) argue that the ‘precariat’ should support a BI most, because in an economic sense, it would ‘gain most’ of it (p. 183). As people with lower education have a higher risk of ending in a more precarious position, it could be expected that they would be more in favour of a BI. Women, who in general work fewer hours, are paid significantly less, and are often dependent on a breadwinner, are expected to benefit from a BI in both financial and emancipatory terms (Van Parijs and Vanderborght, 2017). 2 We therefore include indicators of socio-economic position as subjective income level, work status, educational level and gender in our explorative analysis.
A fully fledged BI covers the whole (adult) population, however, and as such, affects the economic situation of all citizens. Thus, it can be argued that people in a lower structural position would not favour a benefit structure that benefits the rich as well. Especially, working-class people supporting new right-wing parties might argue that a BI is a typical instrument of egalitarians to benefit non-deserving people (De Koster et al., 2013). In this respect, we can also refer to deservingness theory that argues that people are less in favour of benefit schemes that benefit undeserving social groups (Van Oorschot, 2006). Important criteria to judge the deservingness of social groups are the neediness of the group, whether or not they reciprocate to society and whether or not they can be held responsible for their situation of neediness (Van Oorschot et al., 2017). As BI is both universal and unconditional, it will benefit social groups that are not needy (‘the rich’) and that do not reciprocate or take responsibility (‘the lazy unemployed’). The fact that BI also benefits undeserving individuals may alter support, also of those who do have a direct economic interest in it.
Moreover, there are other than economic interests to support a BI. Psychological effects, as for example, experiencing more freedom in one’s life, or social effects, as for example, being able to pay more attention to social life, care and personal development, may be valued as well, and not only or mostly by the lower classes or by women. It could be expected that for this reasoning people with children at home, would favour a BI. Age may also be a factor in supporting BI, although it is not immediately clear in which the direction the effect would go. People later in their career may value a BI to create the possibility of working fewer hours, while people at the start or in the middle of their working life could benefit from using free time for investing in personal development. We include variables as whether or not people have children at home and also age that may be a factor in valuing free time and personal development.
As for the influence of ideological stances, expectations are not self-evident either, because in social debates the BI is advocated on the basis of a range of different ideologies. For instance, from an egalitarian perspective, BI is advocated as an income equalizer and a strong remedy against poverty (Baker, 1992; Van der Veen and Van Parijs, 1986), and as a socialist critique on capitalism as well (Wright, 2010). But, a BI has also been advocated from a liberal perspective in setting all people free to make their own decisions in life (Jordan, 1992). And, both business leaders and liberals have been attracted by ‘its simple, non-bureaucratic, trap free, market-friendly operation’ (Van Parijs and Vanderborght, 2017). Among BI supporters is also free market advocate Milton Friedman who argues that a BI will ‘free the poor from a cumbersome, paternalistic welfare system’ (Widerquist et al., 2013: 2). From a communitarian perspective, the BI is promoted as offering people possibilities of opting for caring or participating in other socially meaningful ways in society, or contributing to gender equality, social cohesion and social trust (Etzioni and Platt, 2008; Jordan, 1992). Political parties as such, however, never carried the idea very strongly though, with some exceptions that can particularly be found in the green parties (Van Parijs and Vanderborght, 2017). Support for a BI may therefore not be directly linked to a specific type of ideological stance or may be linked to multiple ideological positions. To explore the relationship with ideological variables, we include political left–right stance, egalitarianism and meritocratic values as individual predictors in our analyses.
Possible contextual factors influencing BI support
In the literature addressing popular support for a BI, it is commonly suggested that such support may be influenced by characteristics of the national context that people live in, especially by the general institutional design of a country’s welfare state or regime type. As BI is presented as a radical alternative to the current welfare state, introducing it would directly contradict the underlying logic of different welfare regimes. A BI clearly challenges the liberal welfare regime that is strongly based on means-tested benefits for the poor only and where benefits contain strict job-seeking obligations (Panitch, 2011). It is also fundamentally different from the social insurance-based benefit systems in corporate welfare regimes, which rely on wage-related contributions and benefits (Van Parijs and Vanderborght, 2017). And although a BI system seems to be most close to the universal Scandinavian type of welfare regime, with its broad coverage and encompassing benefit levels, it nevertheless conflicts with the regime’s historical focus on (full) employment and ‘activation’ of unemployed people (Bonoli, 2010). Bay and Pedersen (2006) suggest in this respect that the high work ethic of Scandinavian citizens may be a strong cultural barrier for BI support. Therefore, welfare regime theory cannot give us clear guidance in predicting in which welfare state regimes BI would be most supported.
In addition to welfare regime type, the influence of other contextual factors is hinted upon in the literature (Martinelli and De Wispelaere, 2017; Van Parijs and Vanderborght, 2017). Suggestions regard socio-economic factors that signal degrees of social needs, like unemployment rate, inequality and poverty. The idea is that there would be more support for a BI in countries (and times) with higher levels of need for social support. In countries with higher levels of unemployment and higher levels of material deprivation, people would favour a BI more as a means to ensure a guaranteed minimum income level for people in need. Higher levels of inequality could increase support for a redistributive measure under the condition that BI is perceived as such. A similar mechanism finds support in welfare attitude studies, where support for more generous welfare benefits and more redistribution is higher in countries where unemployment, poverty and inequality are higher (Blekesaune and Quadagno, 2003; Jeene et al., 2014). Related to this, one could expect that in countries where social need is addressed more comprehensively, as in countries where social spending is higher and even so in countries that are wealthier, a BI may be (seen as) less necessary to provide in social support and may therefore be less popular. This may be related to the fact that the so-called path dependency of welfare institutions is stronger in more developed welfare states (Boone et al., 2018; Pierson, 2000), with the result that it will be more difficult to gather support for a radical alternative such as BI.
There are also, however, suggestions of cultural factors playing a role. A stronger cultural norm of the importance of work ethic would be associated with lower support for an unconditional BI (Bay and Pedersen, 2006) as a BI gives people the opportunity to refrain from working life. In countries where the work ethnic norm is high, there could be more resistance to the element of unconditionality in the definition of BI. In addition, we look at the influence of the power of trade unions in national economies. It is suggested that trade unions hinder the introduction of a BI, as they emphasize the importance of paid work (Vanderborght, 2006). In countries where trade union density is high, rejection of the BI proposal could be stronger, as public debates are more influenced by the resistance of trade unions. We also include trade union membership at the individual level.
There are very limited cross-national studies in which the effects of contextual factors as we discussed here are actually analysed and tested. By our knowledge, there is one research note that examines correlations between contextual characteristics and the mean BI support in countries (Lee, 2018). However, this study does not take into account the individual-level variation and composition effects. Therefore, also with regard to context factors, we adopt an exploratory approach. In our analysis of country differences in BI opinions, we will include the following factors mentioned above in a multilevel analysis: unemployment rate, material deprivation rate (poverty measure), income inequality, social expenditure, countries’ wealth (gross domestic product (GDP) per capita), work ethic and trade union density.
Relations between support for BI and support for the welfare state
BI is often depicted and argued for as a ‘radical alternative’ for the welfare state as we know it today. This suggests that public support for BI as a way of welfare provision could be radically different from support for the welfare state, an example being that stronger supporters of BI are more critical about various aspects and dimensions of current welfare states and their functioning. By exploring whether and in which way, support for BI is (radically) different from support for the traditional welfare state, we will gain more insight in what theories could apply to, or should be developed for understanding BI support in more detail. For this purpose, we will relate BI support to a range of commonly measured popular attitudes to welfare provision (Roosma et al., 2013), as regards its goals, the role to play for government, the provision’s redistributive design and scope, its implementation, and its (un)intended outcomes and consequences, and analyse whether support for BI aligns with these welfare attitudes or deviates from it. It is explicitly not our goal to explain BI attitudes with welfare attitudes, we rather aim to assess the associations between them. The ESS (2016) wave data allow us to operationalize attitudes towards the different dimensions, and relate them to people’s opinion on BI. Below, we discuss our initial expectations about the association between attitudes towards the dimensions (that are available in the data set) and support for BI.
As for the general goals of welfare provision (reducing income differences and poverty) we assume that people who support these, are also supportive of BI, since both systems of welfare provision align in them (Baker, 1992). Regarding the role of government in providing welfare benefits and services, the relationship is more ambiguous. BI is favoured for its simple design and providing individuals with the financial means to make their own choices (Widerquist et al., 2013), while social benefits strongly rely on complex entitlement structures and conditional benefits. The state being responsible for providing conditional benefits might be perceived as different to the provision of an unconditional income. In addition, the welfare state was criticized for not addressing adequately new social risks with the existing benefits and services (Morel et al., 2012). People might believe that these risks are better covered by a BI.
Considering the redistributive design of the welfare state, a central feature of it is targeting benefits at certain groups who are seen as deserving of support (Van Oorschot et al., 2017), while a BI is in essence a universal benefit provided to all citizens. We expect that individuals who favour targeting more strongly are less in favour of a BI. In the same line of arguing, it could be expected that people who fear mistargeting in terms of abuse of welfare provisions, or underuse of benefits by people who are deserving, are more supportive of a BI, since in a system of universal entitlement abuse and underuse of benefits would be non-existent.
We believe that people who are dissatisfied with the outcomes of the welfare state could in general be more likely to support BI as they might see it as a more fair and effective system of redistribution (Widerquist et al., 2013). People who see the positive consequences of the welfare state in terms of preventing poverty and tackling inequality could see these positive outcomes as well for the BI (but not necessarily). On the other hand, people who recognize negative economic consequences of the welfare state (in the sense that it harms the economy) would also not be very enthusiastic about the BI as it may have an even stronger redistributive impact. Also, people who believe that the welfare state makes people lazy will be even more against an unconditional BI. But people who claim that the welfare states crowd out interpersonal solidarity may be more in favour of BI as it provides more time to combine work with voluntary work and family life (Van Parijs and Vanderborght, 2017).
Data and methods
Data
Studying support for BI, especially in a comparative perspective, was long hindered by the lack of availability of high-quality survey data. Moreover, many questions hinting at a BI were ambiguous, for example, asking about support for a ‘guaranteed minimum income’, which could be interpreted not only as support for a BI, but also as support for social assistance. The ESS, round 8 (2016), measured support for BI – in 23 European countries
3
– after providing respondents with an extensive introduction. The question is stated as follows: A basic income scheme includes all of the following: the government pays everyone a monthly income to cover essential living costs. It replaces many other social benefits. The purpose is to guarantee everyone a minimum standard of living. Everyone receives the same amount regardless of whether or not they are working. People also keep the money they earn from work or other sources. This scheme is paid for by taxes. Overall, would you be against or in favour of having this scheme in [country]?
The following answer categories were provided: strongly in favour, in favour, against and strongly against. The data set contains 44,387 respondents; 44,144 respondents answered the question about BI of which 3552 people choose the ‘Don’t know’ option (8.04%). These respondents were coded missing and robustness checks show that these ‘missings’ are overall random. 4 In our explorative, multivariate analyses, we use a dummy variable indicating either support (1) ((strongly) in favour) or no support (0) ((strongly) against).
As mentioned in the introduction, this is the first cross-sectional study that introduces the idea of BI with an encompassing definition of its different aspects: its universal and unconditional character, the fact that it guarantees a social minimum, that it replaces other benefits and services and that it is paid for by taxes. In debates about BI, many other varieties of BI proposals are discussed, including proposals for a negative income tax, partial BIs, conditional BIs and BI systems that replace all benefits and services. Unfortunately, with this survey question, it is not possible to analyse to what elements people actually respond when they express their support or rejection, and we do not know how they interpret the different elements. At the same time, the question as it is formulated refers directly to the most common and encompassing definition of a BI (Widerquist et al., 2013). It can therefore function as a point of reference for future surveys and studies.
For our independent variables on the individual level, we use the following indicators. Subjective income is measured on a 4-point scale, indicating whether respondents were ‘living comfortably’ (4) or ‘coping’ (3) on present income or having it ‘difficult’ (2) or ‘very difficult’ with their present income. Work status is measured with dummy variables, including being in paid work (reference category), unemployed, in education, permanently sick or disabled, retired, doing housework or community service. Education is measured on a 5-point scale, running from primary education (1), to lower vocational training (2), medium vocational training (3), higher vocational training (4) or bachelor’s and master’s (5). Children measures whether people have children living with them at home. Age is measured in years. For gender, we use male as the reference category. Trade union membership indicates with dummy variables people who were previously or currently a member of a trade union or a similar organization or never have been a member (reference category). Political affiliation is measured on an 11-point left–right self-placement scale (higher score indicating more right-wing). Egalitarianism is measured in answer to the question ‘For a fair society, differences in standard of living should be small’ measured on a 5-point scale. Meritocracy is measured on a 5-point scale in answer to the question ‘Large differences in income acceptable to reward talents and efforts’.
The welfare attitudes are measured with the following items. Support for the goal of the welfare state is measured as support for the statement that the government should reduce income levels. The support for the role of the government is measured with three items (11-point scale), asking to what extent the government is responsible for providing childcare, a reasonable standard of living for the old and for the unemployed. Two other variables concern support for welfare policies tackling new social risks: support for benefits that allow parents ‘to combine work and family life, even if this means higher taxes and support to ‘spend more on education for unemployed at cost unemployment benefits’. Considering support for redistributional design, there is one item that measures support for targeting social benefits only to people with lowest incomes. Two variables are indicating perceived mistargeting of social benefits: a measure of people’s perception of abuse (‘many people manage to obtain benefits to which they are not entitled’) and a measure of perceived underuse (‘many people get fewer benefits than they are legally entitled to’) of welfare benefits. People’s opinions about the outcomes of the welfare state are measured with eight items, asking, ‘Do social benefits and services lead to a more equal society (item 1) and less poverty (item 2)?’ ‘What do you think is the standard of living of the old (item 3, 11-point scale) and of the unemployed (item 4, 11-point scale)?; ‘Does the welfare state place a too great strain on the economy (item 5), and costs businesses too much in taxes and charges (item 6)?’ ‘Does the welfare state make people lazy (item 7) or less willing to care for one another (item 8)?’ All items are measured on a 5-point scale, unless indicated otherwise, and are recoded such that a higher score measures more support for the statement.
As contextual-level measurements, we use from OECD Statistics the unemployment rate as percent of labour force in 2015; the Gini coefficient in 2014, trade union density in percentages (the ratio of wage and salary earners that are trade union members, divided by the total number of wage and salary earners, calculated using administrative data adjusted for non-active and self-employed members otherwise (with some exceptions 5 )) in 2015 and GDP per capita in US dollars at current prices and purchasing power parities (PPPs) from 2015. From Eurostat, we use total expenditure on social protection as percentage of GDP in 2015 (for Poland we use data from 2014) and material deprivation (measured as percentage of people who are severely materially deprived, have living conditions severely constrained by a lack of resources) in 2015. From the World Value Study, we obtain a measure for work ethic – the percentage of people in a country that agrees (strongly) that it is humiliating to receive money without having to work for it – for 2008. For Israel, data are missing on social expenditure, material deprivation and work ethic. For Russia, data are missing on social expenditure, material deprivation and union density.
Methods
We first present several descriptive statistics (design and population weights are applied) and then, second, continue with our multivariate analyses presenting several multilevel logistic regression models in which we add three sets of individual-level indicators step by step (including first the socio-economic characteristics, then the ideological characteristics (including trade union membership) and finally, the welfare attitudes). In the third step, we include the contextual-level measures. Because of a limited number of level 2 units (countries), we only include one variable in Models 4–10 and two contextual variables in Model 11. For measuring the explained variance, we follow instructions from Snijders and Bosker (2012: 305–307). We compute the linear predictor for Ŷij and use its observed variance. This variance is divided by the total variance, which consists of the variance of the linear predictor, the variance of the random intercept and the fixed value of the level-one residual variance (π2/3).
Results
Descriptive statistics
Figure 1 presents an overview of support for BI in the different countries. We see ample variation in support for BI with the highest level of support in Latvia (over 80% indicates to be (strongly) in favour of a BI) and the lowest level of support in Norway (over 66% indicates to be (strongly) against a BI). Nevertheless, it is remarkable that support for such a radical alternative to the current welfare system gains so much support in European countries. In total, about 56 percent of the respondents in this selection of countries indicates support, while 44 percent is against a BI. In 20 out of 23 countries, support is higher than 45 percent. Overall, it seems that populations in Eastern European countries are more in favour than their counterparts in Nordic and Western European countries, although there are some exceptions (for instance, Estonia being more against BI and Finland more in favour compared with their respective region).
Support for BI in European countries – 2016.
Effects of individual-level indicators
In the next step, we relate support for BI to three sets of individual-level variables. The results are presented in Table 1. Model 0 is the empty model. The intra-class correlation (ICC) for this empty model is small: 0.067, which means that only a small proportion of about 7 percent of the differences between individuals is related to the group level (countries) (which suggests that the influence of country characteristics may be small). In Model 1, we include all variables that measure people’s social position. The results show that there is a relatively strong and positive relation with having a lower subjective income and being unemployed, on the one hand, and supporting a BI, on the other. This confirms previous results (Andersson and Kangas, 2002; Stadelmann-Steffen and Dermont, 2017) and may be an indication that people primarily see BI as a way to secure themselves of a guaranteed income. Higher educated people, people having children at home and elderly people are less in favour of a BI. As BI is often depicted in the literature as a way to make it easier to combine work and family life, it is remarkable that having children has a small but negative relation with BI support. Interestingly, the effect of age shows that younger people are more in favour of this ‘radical alternative’ to the current welfare state. This might be because younger people – or people from younger cohorts – see BI as an opportunity to invest in personal development, or are more open to new ideas, or see advantages of a new system of welfare support that addresses new social risks better. There is no significant effect of gender, despite arguments that females would benefit relatively more from a BI as they would gain in economic independence.
Multilevel logistic regression model on support for BI – individual characteristics.
SE: standard error.
p < 0.10; **p < 0.05; ***p < 0.001.
In Model 2, we add the individual-level items that measure ideological and political position, and trade union membership. It shows that despite of some advocates among philosophers and politicians from the political right under the general public, BI is mostly carried by the political left. Current or previous trade union membership has no significant relationship with BI support. Only when we control for all welfare attitudes in Model 3, the negative effect of current trade union membership becomes borderline significant. The critical position of the trade unions against BI does not seem to largely affect those who are or were members of it.
The level of explained variance in these first two models is fairly low. Yet, it must be noted that the R2-value is considerably lower for logistic regression than for ordinary least squares (OLS) regression with continuous outcomes (Snijders and Bosker, 2012: 306).
In Model 3, we add the variables that measure various dimensions of welfare support. The explained variance almost triples to 0.11. The relations with welfare attitudes provide interesting insights in possible reasons for BI support. We see a strong positive relation between support for BI and support for the goal to reduce income differences. Support for the role of government (range) is, as expected, not automatically related to support for a BI. The effects are relatively small and there is even a negative relation with those who support the government to be responsible for childcare. This could be because supporters of a BI would rather have more time (and financial means) to take care of their children, instead of relying on state/organized childcare. Interestingly, the relations with support for social investment type of roles of the government (investing in education of unemployed and benefits to combine work–family life) are very strong: people who are more in favour of such social investment arrangements are also more in favour of a BI. This could suggest that by some BI is indeed perceived as a modern, innovative idea as an alternative for the current welfare state.
Surprisingly, there is a strong positive relation between support for targeting benefits to the poor only and BI support. As targeting benefits exclusively to the poor is directly opposing the universal character of the BI, this result raises questions about whether respondents have really understood what a BI actually entails: do people really understand that ‘everyone receives the same amount’, means including the middle classes and the rich as well? Or, can this result be interpreted that people support a BI and targeting benefits only at the poor, for the same reason: improving the conditions of the poor. Whether this is by introducing a BI or by targeting benefits only to the poor might be irrelevant to them.
This interpretation of the results would also explain the relations between BI support and perceived abuse and underuse of benefits. Abuse of welfare would be more or less solved when a BI is introduced. Because every citizen is entitled, abusing benefits would be impossible. Yet, the results show that people who see a lot of welfare abuse are more against introducing a BI. This may result from the fear that benefits are targeted at people who do not really need them or abuse these benefits (Van Oorschot and Roosma, 2015). People who see a lot of underuse of benefits are instead very supportive of a BI. This could suggest that those who are worried that people who are deserving of benefits do not get assistance are more likely to support a BI that provides a guaranteed minimum income for everyone, including those who now are not receiving what they are entitled to. This may add to an overall interpretation that a BI is supported for its ability to provide a guaranteed minimum and address issues of poverty rather than to overcome the major flaws of our current welfare system.
Inspecting the relationships with the items representing the perceived outcome of the welfare state would lead to a similar interpretation. People who are very critical about the outcomes of the welfare state are not more in favour of the alternative BI. Relations are small and insignificant instead. Only those who believe that welfare benefits will reduce equality and poverty are more in favour of a BI. This also seems to suggest that a BI is seen as an alternative way to address the needs of the poor. The fact that people who believe that the welfare state makes people lazy are more against a BI shows one of the major pitfalls of social legitimacy of BI: the unconditional aspect of BI goes against the fundamental deservingness criterion of reciprocity (Van Oorschot et al., 2017).
Contextual effects on BI support
In Table 2, we present Models 4–11 in which contextual-level factors are added, one in each model. Most indicators are not significant and the explained variance does not increase much compared with Model 3 with all individual-level variables included. Only two indicators have a significant effect. The degree of material deprivation in a country significantly influences support for BI positively (Model 5). Such deprivation seems to the most important contextual factor, since in this model, the explained variance increases substantially to 0.146. Also, the level of social spending has a significant effect (Model 10). The higher the social spending in a country, the lower support for BI. In generous welfare states, there seems less reason to support an alternative social model. However, this effect adds not so much in additional explanatory value. If material deprivation and social spending are both included (in Model 11), the effect of material deprivation remains significant, while the effect of social spending turns insignificant instead and the increase in explained variance compared with Model 5 is only marginal.
Multilevel logistic regression models on support for BI – contextual factors controlled for all individual-level items.
GDP: gross domestic product; SE: standard error.
p < 0.10; **p < 0.05; ***p < 0.001.
Discussion and conclusion
In the past years, advocates of an unconditional universal BI have entered the public debate and tried to influence the political and social policy agenda by proposing a radical alternative to our current welfare state. As we know that European welfare states are both supported for their goals and policies, as well as criticized for their mistargeting and outcomes (Roosma et al., 2013), it is not only interesting to study to what extent the general public supports a BI, but also if they see a BI as a possible alternative for the welfare state or flaws of it.
A first conclusion of our article is that support for BI is high across Europe. Considering the fact that the proposal would have major consequences for our current welfare systems, both in redistributional effects, as in social and economic outcomes and in terms of (administrative) implementation, support for BI seems to be rather overwhelming. In 20 out of 23 countries, support is higher than 45 percent. The high support confirms results from previous studies (Dalia, 2017; Ipsos, 2017), but in the ESS question, for the first time, BI was explained and introduced to the respondents by providing an elaborate definition. Not only the universal and unconditional character were highlighted, but also the fact that the scheme would be paid for by taxes.
This overwhelming support might therefore raise some doubts whether respondents fully understand the impact of introducing a BI. Or, more likely, it may be that they express support for different reasons. A BI is a complex proposal with many aspects of which some could be praised by respondents, while others could be criticized. Although we could not test the motivations of people to express support (or disagreement) with BI, the individual respondent characteristics that affect BI support, the relationships between BI support and attitudes towards various welfare state dimensions, and the effects of contextual-level factors on BI support give us some indications that point in a similar direction. That is, people in a worse social–economic position, people supporting egalitarian values and people leaning to the political left are more in favour of a BI. Also, people who want the government to reduce income levels and people who support targeting benefits at the poor (!) are more in favour of a BI. In countries where material deprivation is high, support for BI is significantly stronger. This all may suggest that people in general express support for the BI proposal for the reason that they want to improve the conditions of the people who are worse off in their country (which may include themselves). If this would be the case, then it is not the universal character or its unconditionality that makes a BI so attractive to a large share of the European population, but the fact that it provides (poor) people with a guaranteed minimum income. It could also implicate that a BI is not a politically neutral idea. That is, academic or social supporters of the BI may have different political and philosophical affiliations, among the general public, the idea of BI is carried mostly by the political left. This would of course require further study and the development of a theoretical framework that distinguishes different mechanisms for explaining support for different elements of BI. We suggest that deservingness theory would provide a relevant starting point to formulate expectations regarding support for specific elements as the universal and unconditional character of BI, as particularly these aspects violate the popular deservingness criteria of need, reciprocity and control (Van Oorschot et al., 2017).
Critique on the current welfare state, regarding its mistargeting of benefits or negative economic, moral and social consequences, is no reason to support a radical alternative. In fact, people who expect abuse of welfare benefits or moral hazard are more likely to reject a BI. It suggests that a BI might deal with the same legitimacy issues as the welfare state: a fear for inactiveness and freeriding.
An interesting part of the results is the strong positive relation with support for social investment policies (in educating the unemployed and in combining work–family life). People who are open to such welfare reforms might also be more open towards reform by means of a BI. Also, the fact that younger people – or this might be people from younger cohorts – are more supportive of a BI might give some hope to BI advocates who present the proposal a social system of the future.
The major drawback of our study is that our interpretations could only be based on modest quantitative relations between survey items of a large pooled sample of Europeans. The question whether or not the proposal of BI has social legitimacy is therefore not definitely answered. We need more (especially qualitative) studies to ask people about their particular understandings of BI, which elements of the idea they support and reject, and their argumentations to be in favour or against it. This will not only tell us more about whether or not the introduction of BI can be successful in the future, but also – if the BI proposal remains a futuristic idea only – what elements of our current welfare system need improvement according to the public.
Footnotes
Funding
The author(s) received no financial support for the research, authorship and/or publication of this article.