Basic Human Values and Attitudes Towards a Universal Basic Income in Europe

2.1 Support for UBI

Before the 2016 wave of the ESS, there were cross-national surveys including a question about a basic income. For instance, International Social Survey Programme (ISSP) 1987 and 1992 and Eurobarometer 56.1 (Sept–Oct 2001) had the following question: “The government should provide everyone with a guaranteed basic income.” In these surveys, however, a definition of the basic income was not provided at all; it is thus hard to assume that people are aware of what a basic income is. By contrast, Dalia’s (2017) recent survey on UBI, which asked 11,021 respondents across 28 EU member states what they think of UBI, presented a rather concrete definition of UBI. However, this survey did not include various covariates that can be used for empirical analysis exploring the relationships with UBI support. Of course, a great deal of opinion polls or surveys about UBI have been carried out so far within a country. As such, the ESS in 2016 has been the unique survey, with which cross-national comparative research can be conducted, providing a variety of variables with a relatively elaborate definition of UBI:

1. The government pays everyone a monthly income to cover essential living costs.

2. It replaces many other social benefits.

3. The purpose is to guarantee everyone a minimum standard of living.

4. Everyone receives the same amount regardless of whether or not they are working.

5. People also keep the money they earn from work or other sources.

6. This scheme is paid for by taxes.

As Vlandas (2019, p. 3) pointed out, however, the definition above has some gray areas. Specifically, the level of UBI remains abstract; there is no indication of whether or how taxes will be increased; it is not clear to what extent social benefits will be replaced. The findings from a recent UK survey obviously indicate that both an increase in taxes and a cut in welfare spending to finance UBI significantly decrease support for UBI (Smedley, 2017). Accordingly, support for UBI should be interpreted with caution since it highly depends on how UBI is designed and financed. In other words, different UBI models may lead to different levels of support, which could also change the associations between support for UBI and its predictors.

Theoretical tools to interpret UBI support are currently more underdeveloped than theoretical expectations of preferences for redistribution or welfare attitudes. On the one hand, there are common grounds between UBI and other redistributive policies in the welfare state. UBI is a powerful policy not only for achieving universalism but also for a broader equalization of opportunities through redistribution (Mays & Tomlinson, 2019).^[1] Attitudes towards UBI can also be regarded as “an indicator of support for universalistic welfare policies and the associated redistributive aspirations” (Bay & Pederson, 2006, p. 432). Furthermore, several expectations from the welfare state literature are consistent with the recent findings on support for UBI; for example, the explanatory power of the variables relating to self-interest is significant in building preferences for both welfare state policies and UBI (Vlandas, 2021).

On the other, UBI is widely considered a radical alternative to the existing welfare state as the principles of UBI contradict the underlying logic of the welfare state (Bay & Pederson, 2006; Roosma & Van Oorschot, 2020; Widerquist, Noguera, Vanderborght, & De Wispelaere, 2013). Universality and unconditionality are the core principles of UBI, hence it explicitly challenges, first, means-tested benefits and job-seeking obligations attached to benefits on which the liberal welfare regime is based; second, wage-related insurance-based benefit systems on which the corporate welfare regime is based; and third, stressing on full employment and activation of the unemployed on which the Scandinavian type of welfare regime is based (Roosma & Van Oorschot, 2020). Consequently, attitudes towards UBI are expected to be different from attitudes towards the welfare state policies.

2.2 Basic Human Values and UBI Support

It is widely assumed that policy preferences are formed at both individual and country levels (Blekesaune & Quadagno, 2003; Blekesaune, 2007). If this is the case, in principle, individual and contextual factors may play a significant part in shaping preferences for UBI; moreover, the moderating role of contextual factors cannot be excluded. At the individual level, it has long been suggested that not only self-interest or group identifications but also values or beliefs play a role in judging the desirability of policies (Feldman, 1988). This article focuses on human values as explanatory factors rather than self-interest or group identifications that are taken as controls. Roosma and Van Oorschot (2020) emphasize the role of welfare attitudes, which are also considered controls in the current study; however, these attitudes are distinct from values.

Values refer to general standards on which specific evaluations are based, but attitudes are considered evaluations of specific objects (Feldman, 2003, pp. 480–481). The consensus has been reached that “values hold a higher place in one’s internal evaluative hierarchy than attitudes” (Hitlin & Piliavin, 2004, p. 361). Thus, more general value systems as well as specific attitudes would influence policy preferences (Blekesaune & Quadagno, 2003). If values are working quietly behind the scenes, a lack of attention to values may hinder explanations of social phenomenon of interest (Hitlin & Piliavin, 2004).

According to the Schwartz (1992) theory of human values, using 54 value items, the number of fundamental human values can be reduced to 10: self-direction, stimulation, hedonism, achievement, power, security, conformity, tradition, benevolence, (individual) universalism. Inglehart and Baker (2000) and Inglehart and Welzel (2005) have also developed a values paradigm, but Schwartz’s framework is considered superior to Inglehart’s for theoretical, empirical, and practical reasons (Miles, 2015). It cannot be said that Schwartz’s conceptualization of value structure is immaculate and completely validated. However, Schwartz and Boehnke (2004) confirmed the 10 values by conducting confirmatory factor analysis with data from two sets of 23 samples from 27 countries, which is the first statistical test of the Schwartz theory.

Schwartz and his colleagues have also confirmed the basic categories and structure of values thereafter with data from hundreds of samples in 82 countries across the globe, which provides evidence for the validity of the Schwartz theory across countries. (Schwartz, 2012; see also Bilsky, Janik, & Schwartz, 2011; Davidov, Schmidt, & Schwartz, 2008; Fontaine, Poortinga, Delbeke, & Schwartz, 2008). Meanwhile, all rounds of the ESS have adopted the 21 survey questions of the value items recommended by Schwartz (2003). Davidov et al. (2008) have found evidence that the ESS values scale demonstrates configural and metric invariance, which is essential for investigating relationships among values and other variables cross-nationally. These are sufficient reasons for the present study to employ the Schwartz human values in exploring the role of values on UBI support.

Schwartz has classified the 10 human values into a circular continuum illustrating their relationships, as shown in Figure A1 of the Online Appendix. The figure illustrates four higher-order value types along the two orthogonal dimensions: self-transcendence versus self-enhancement and openness to change versus conservatism. Among them, this article focuses on the two competing value types: self-transcendence and self-enhancement. The self-transcendence value type emphasizes “acceptance of others as equals and concern for their welfare (universalism and benevolence)”, and the self-enhancement value type stresses “the pursuit of one’s own relative success and dominance over others (power and achievement)” (Schwartz, 1994, p. 25). This value dimension has been taken as it is more relevant to the distribution of material resources (Barnea, 2003), which is an essential feature of UBI. Schwartz (1994, p. 40) argues that this dimension is pertinent to people’s stance towards economic egalitarianism and hypothesizes that individual universalism and benevolence are most positively associated with egalitarian orientations, while power is most negatively related to it.

More specifically, individual universalism as a self-transcendence value, which should not be confused with welfare universalism, stresses “Understanding, appreciation, tolerance and protection for the welfare of all people and for nature” (Schwartz, 2003, p. 268). Benevolence as the other self-transcendence value emphasizes “Preservation and enhancement of the welfare of people with whom one is in frequent personal contact” (Schwartz, 2003, p. 268). These values might be compatible with the idea of UBI as UBI is advocated as a way of improving everyone’s well-being, not to mention the well-being of people surrounding his or her. Accordingly, it is hypothesized that the two self-transcendence values are positively associated with support for UBI.

With respect to the self-enhancement values, the central motivational goal of power is “Social status and prestige, control or dominance over people and resources” (Schwartz, 2003, p. 267); the goal of achievement is “Personal success through demonstrating competence according to social standards” (Schwartz, 2003, p. 267). These values may have a negative effect on support for UBI as they emphasize one’s own material payoffs rather than other-regarding preferences that are altruistic or inequality-averse. The self-enhancement values are also placed in diametrical opposition to the self-transcendence values in the value circle (Caprara, Schwartz, Capanna, Vecchione, & Barbaranelli, 2006, p. 8; see also Figure A1). Therefore, it is predicted that the two self-enhancement values are likely to be negatively associated with support for UBI.

2.3 Other Determinants of UBI Support

Apart from the basic human values, the recent literature on support for UBI employed a range of predictors that can be classified into four groups: individual characteristics, specific beliefs or preferences, particular attitudes, and contextual factors. As for the individual characteristics, age, gender, employment status, income, and ideological position were widely used as individual controls. It is expected that persons who can benefit from UBI, such as individuals exposed to high labor market risk and on low incomes, are more likely to be supportive of UBI, as Vlandas (2021) already provided its evidence. Economic individualism and economic fairness as beliefs about economic disparities as well as preferences for redistribution were used in Roosma and Van Oorschot (2020).^[2] Regarding specific attitudes, Parolin and Siöland (2020) used welfare chauvinism, and Roosma and Van Oorschot (2020) employed a number of welfare attitudes. Last, various contextual factors were used: income inequality, unemployment rate, social expenditure, social protection, or material deprivation.

Beliefs about economic disparities need further explanation, which Roosma and Van Oorschot (2020) have not provided. In the welfare state literature, there are “two dominant social ideologies economic individualism or the work ethic and social equality or collective responsibility” (Hansenfeld & Rafferty, 1989, p. 1029). Economic individualism assumes that people’s income disparities reflect differences in their talents and efforts. It is thus conjectured that people who are favorably inclined towards economic individualism are less likely to support UBI because UBI is an unconditional cash transfer regardless of individuals’ talents and efforts. By contrast, it is hypothesized that people who see a highly unequal society as unacceptable are likely to be in favor of UBI. For a just society, people may expect that differences in people’s standard of living should be small, which UBI pursues by guaranteeing everyone a minimum standard of living. Moreover, preferences for redistribution are expected to be compatible with support for UBI since UBI is widely considered a more effective and egalitarian system of redistribution (Widerquist et al., 2013; Wright, 2010).

As Roosma and Van Oorschot (2020) did, it would be interesting to see how individuals’ evaluations of current welfare policies affect their opinion on UBI. If people perceive the existing welfare system as satisfying and well-functioning, it would be less likely for people to support UBI as it is expected to replace many other social benefits. Seven survey questions on social benefits and services from the ESS might be relevant to this study. Two items among them question whether social benefits and services alleviate poverty and make a society more equal. It is supposed that those who agree with these views are more likely to support UBI since the views are in accord with the goals of UBI. Other two items ask about the burden of social benefits and services on businesses and the economy. It is hypothesized that those who have more concerns about high taxes and economic strain to sustain generous welfare policies are less supportive of UBI because UBI is often considered too costly. Another question is about a widespread thought that welfare makes people lazy. This laziness argument is one of the strongest ones for the opponents of UBI. Therefore, it is predicted that those who agree with the argument are more likely to oppose UBI. The next to last item asks whether social benefits and services should target the poorest people. It is assumed that support for targeting is inversely related to support for UBI since targeting is seemingly against a universalist approach on which UBI is based. The last question is about welfare chauvinism that might be associated with a low level of support for UBI. This is not only because anti-immigrant attitudes are generally linked to a low level of support for broad redistributive policies (Alesina, Miano, & Stantcheva, 2018), but also because one of the principles of UBI that everyone receives UBI if certain criteria are fulfilled goes against an idea that immigrants should never get the same rights to welfare benefits and services as native-born citizens. It should also be noted that attitudes and values or beliefs may be endogenous to each other; that is why different models excluding each other blocs of variables have been tested in the current analysis.^[3]

Regarding contextual factors, first, income inequality is expected to have a positive effect on attitudes towards UBI. UBI is well known as a redistributive measure to redress economic disparities, and increasing income inequality may lead people to prefer more redistribution, according to Meltzer and Richard (1981). Second, unemployment rate might be positively associated with support for UBI as it indicates a country’s socioeconomic condition that signals degrees of social needs, which UBI may satisfy by guaranteeing the unemployed a regular income. Third, it is interesting to see if levels of welfare development influence UBI support because there exist conflicting hypotheses about that, particularly, in a developed welfare regime: on the one hand, we may expect that people in advanced welfare states are receptive to a generous welfare policy as they have already got used to it; on the other, there might be no demand for a further redistributive policy in a country providing comprehensive social protection, as the ‘growth to limits’ hypothesis (Jæger, 2006) or the saturation hypothesis (Pfeifer, 2009) suggests.^[4] Last, it is also meaningful to examine how the degree of welfare universalism in a country is linked to support for UBI, which has never been considered in the recent literature on UBI support. It might have conflicting effects on UBI support. According to Bay and Pederson (2006), positive attitudes towards UBI may reflect the popularity of the universal welfare state; on the contrary, people in a highly universalist welfare state may not demand a more universalist policy such as UBI.

Furthermore, it is worth exploring whether or not the associations between the basic human values and support for UBI are conditional on contextual factors. Seligman and Katz (1996) suggest that contextual factors may affect value priorities. Kulin and Svallfors (2013) present empirical evidence that the associations between human values and attitudes are stronger in a more materially secure class or society. Thus, there is a possibility that a certain contextual factor, such as the level of welfare development, moderates the associations between the basic human values and support for UBI. We can expect that, for instance, the basic human values may be more strongly associated with support for UBI in a more developed welfare state.

2.4 Data and Methods

The dependent variable is support for UBI that is dichotomous: 1 is assigned to the responses ‘strongly in favor’ or ‘in favor’; 0 is assigned to the responses ‘strongly against’ or ‘against’. This item is included in the ESS Round 8,^[5] which was conducted in 23 countries: Israel, Russia, Switzerland, and the European Union and the European Economic Area, including Austria, Belgium, Czech Republic, Estonia, Finland, France, Germany, Hungary, Iceland, Ireland, Italy, Lithuania, the Netherlands, Norway, Poland, Portugal, Slovenia, Spain, Sweden, and the United Kingdom.

The independent variables are basic human values; the data were collected from the ESS Round 8. The basic human values include self-transcendence values (individual universalism and benevolence) and self-enhancement values (power and achievement). To compute the scores of these values, the ESS basic human values scale items were used. Each value is based on multiple value items: individual universalism (‘equality’, ‘broad-minded’, and ‘protecting the environment’), benevolence (‘helpful’ and ‘loyal’), power (‘wealth’ and ‘social power’), and achievement (‘capable’ and ‘successful’) (Schwartz, 2003; see also Table A1 in the Online Appendix). The response categories for each value item range from 1 to 6. The original coding of the items was reversed so that higher scores signify that the value is more important. Following the recommendation of Schwartz (n.d.), they are treated as continuous, and then the score of each value was calculated in this way: First, the mean of the items for each value is computed; second, each individual’s mean score over all 21 value items is calculated; and third, the centered score of each value is taken by subtracting the overall mean score from each value’s averaged score. Additionally, the country-level measures of the basic human values were calculated by averaging individuals’ value scores in each country.

Regarding the individual-level controls obtained from the ESS Round 8 (see Table A1 for the survey questions), first, economic individualism, economic fairness, and preferences for redistribution are used. These are dichotomous variables: 1 is assigned to the responses ‘agree strongly’ or ‘agree’ with each statement. Second, the welfare attitudes variables that are also binary (preventing poverty, promoting equality, increasing laziness, burden on business, strain on the economy, support for targeting, and welfare chauvinism) have a value of 1 that represents ‘agree strongly’ or ‘agree’, except for support for targeting (1 is assigned to the responses ‘in favor’ or ‘strongly in favor’) and welfare chauvinism (1 is assigned to the response ‘they should never get the same rights’). Third, the other controls include age group, gender, employment status, left-right self-placement, household’s total net income after tax and compulsory deductions, and feeling about household’s income.

A number of contextual variables are also used: net income inequality, unemployment rate, welfare development indicators (social expenditure, social protection, and material deprivation), and an index of welfare universalism. The net income inequality data were collected from the Standardized World Income Inequality Database (SWIID) Version 8.0 (Solt, 2019). Unemployment rate can be measured as a percentage of the unemployed of total labor force, and the data were taken from the World Development Indicators (World Bank, 2019). The level of social expenditure as a percentage of GDP is used to measure the relative development of the welfare states; the public social expenditure data expressed as a percentage of GDP were taken from the OECD Social Expenditure Database (SOCX) (OECD, 2019). A social protection indicator can also be used to measure welfare state development because it reflects the level of protection that people actually experience (Lee, 2018); alternatively, we can use an indicator of material deprivation that is measured as percentage of people who are severely materially deprived.^[6] The social protection data as a logarithm of social protection benefits in PPS (Purchasing Power Standards) per head and the material deprivation rate were collected from Eurostat (2019). The welfare universalism index that is a geometric mean of the percentage of non means-tested social protection benefits of the total social protection benefits and the percentage of public social expenditure of the total social expenditure, which is similar to the index constructed by Jacques and Noël (2018); the data came from Eurostat (2019) and OECD (2019). Most of the contextual variables data were collected in 2016, but in some cases where the data are not available, observations in 2015 were used instead. For Israel, data on social protection, material deprivation, and welfare universalism are missing; for Russia, data are missing on social expenditure, social protection, material deprivation, and welfare universalism.

Given the fact that the dependent variable is binary and that country-level factors and their cross-level interactions are taken into account in the main analysis, I begin with a multilevel logistic regression as a primary modelling framework in estimating each model. In this multilevel analysis, however, there are some concerns over the higher-level sample size that is, at best, 23. According to a simulation study of Bryan and Jenkins (2016), there should be at least 30 countries for multilevel logistic models to guarantee reliable estimates, although this threshold is simple rules of thumb rather than hard-and-fast rules. Thus, the results from the multilevel analysis should be (Möhring 2012) cautiously interpreted. To address this issue, in the sensitivity checks, following Möhring’s (2012) recommendation, alternative estimation strategies with country fixed effects are taken:^[7] a logistic regression and an ordered logistic regression using an ordinal measure of support for UBI.

3.1 Descriptive Analysis

Before moving on to statistical estimations, descriptive findings are presented on the country-level support for UBI and the associations between the main predictors and support for UBI. Figure A2 illustrates public support for UBI across the countries, using a country-level aggregate measure in each country that was produced by computing the percentage of respondents who support UBI (‘in favor’ or ‘strongly in favor’). In post-communist countries, the level of support is very high; for instance, 80.4% in Lithuania and 73.6% in Russia. By contrast, Norway and Sweden, which are the Scandinavian welfare states, show noticeably low support for UBI: 33.9% and 39.4%, respectively.

Figure A3 presents the relationships between the country-means of the four basic human values and the country-level measure of UBI support. In this figure, the national average self-enhancement values (power and achievement) are positively related to support for UBI, while the national average self-transcendence values (individual universalism and benevolence) are negatively linked to support for UBI. Meanwhile, if we focus on the four Nordic countries (Finland, Iceland, Norway, and Sweden), a different pattern of relationship emerges. In particular, Norway and Finland serve as clearly opposing cases, although the two countries share the Nordic tradition of universal welfare policies (Kildal & Kuhnle, 2005).^[8] Norway features a lower level of individual universalism but higher levels of power and achievement than Finland, and support for UBI is greater in Finland than in Norway: 55.5% and 33.9%, respectively. These imply that Finish people who are more egalitarian-oriented and less inclined to stress one’s own material payoffs are more supportive of UBI than Norwegian people.

Additionally, regarding the relationships among the basic human values, as expected in the Schwartz theory, individual universalism and benevolence are positively and moderately related to each other, while they are negatively and moderately linked to power and achievement, as shown in Table A3(b). Table A3(c) provides the correlation coefficients between the country-level support for UBI and the contextual factors. Social protection and material deprivation are strongly related to support for UBI, and net income inequality is moderately related to that: r = − 0.68 ,   0 . 82 ,   0 . 42 , respectively.

3.2 Estimation Results

Table 1 presents the results of the multilevel analysis. Column (0) is the empty model. The intraclass correlation (ICC) for this model is 0.067. This indicates that 6.7% of the chances of individual support for UBI is explained by between-country differences; conversely, 93.3% is explained by differences between individuals in a country. Column (1) includes the control variables of individual characteristics. Support for UBI consistently decreases with age. For instance, we can see 56% decrease in the odds of an individual aged over 70 being in favor of UBI, compared to an individual aged between 15 and 19.^[9] Women are slightly less likely to support UBI, although it is marginally significant, which is an unexpected result. According to Van Parijs and Vanderborght (2017), women are expected to be more supportive of UBI than men. As predicted, those who are unemployed are more supportive of UBI than those who are employed; people whose political position is close to right are less likely to support UBI; high-income earners and those satisfied with their income are also less likely to support UBI. All the effects of the individual characteristics are robust to different models in all the tables.

Column (2) adds the basic human values. As predicted, individual universalism as a self-transcendence value is positively related to support for UBI; however, benevolence that is the other self-transcendence value is not positively associated with support for UBI. This finding is interpreted that being more interested in caring for everyone is in accord with the dimension of universality in UBI, while being more helpful and loyal to people around his or her may conflict with that. Regarding the self-enhancement values, both power and achievement are positively linked to support for UBI, which contradicts the hypothesis. Column (3) adds beliefs or preferences instead of the basic human values. Those who prefer economic fairness and redistribution are more likely to support UBI, but those who support economic individualism are less likely to support UBI, as expected.

Column (4) adds welfare attitudes instead of the basic human values. Preventing poverty and promoting equality are positively associated with support for UBI, while increasing laziness, burden on businesses, and strain on the economy are negatively related to support for UBI, as hypothesized. Interestingly enough, those who support targeted welfare policies for the lowest income earners are more likely to support UBI: the odds of those who support targeting being supportive of UBI are 50% higher than the odds of those who do not support targeting being supportive of UBI.^[10] This seems contradictory to the well-known fact that UBI is a policy strongly against targeting. In addition, welfare chauvinism does not seem to be significantly associated with support for UBI, which contradicts one of the main findings of Parolin and Siöland (2020). The different operationalization of welfare chauvinism in the present research from theirs may give rise to this result. They used an original ordinal measure, but this study produced a dichotomous variable on the assumption that accepting the same rights to welfare benefits and services as native-born citizens if certain criteria are fulfilled, on the one hand, is fundamentally distinguishable from denying the same rights in any case, on the other (see Table A1). The results of Column (5) are similar to those of the preceding models.

Table 2 provides the results of the impacts of the country-level factors on individual support for UBI. In addition to the variables in Column (5) of Table 1, Columns (6)–(11) add a different contextual factor in each model, and Columns (12) and (13) include all these factors, except that material deprivation is used as an alternative to social protection as the two variables are strongly correlated. The results indicate that the levels of income inequality, unemployment, social expenditure, and welfare universalism are not significantly associated with support for UBI, whereas social protection or material deprivation is strongly related to UBI support. It is thus reasonable to focus on the levels of welfare state development expressed by social protection or material deprivation rather than social spending.^[11] This leads us to look at the moderating effects of social protection which Table 3 reports.^[12] Columns (14)–(17) show the cross-level interactions with each basic human value and social protection. Among them, the interaction effect between individual universalism and social protection is positive and significant. Column (18) includes all the interactions into a single model. The moderating effect of social protection still remains significant in the relationship between individual universalism and support for UBI.

Figure A4 illustrates the marginal effects of the basic human values on support for UBI by the level of social protection ranging from 8.2, which is close to the lowest level in Lithuania (8.12) to 9.4, which is close to the highest level in Norway (9.41). In the figure, the marginal effect of individual universalism on support for UBI becomes larger and significant as the level of social protection increases above 8.8, while this marginal effect is insignificant where the level is below 8.8. This indicates that individual universalism is not significantly related to support for UBI in Portugal and Spain as well as Post-communist countries in which the level of social protection is below 8.8. Conversely, the relationship is significant in the other European countries. By contrast, the marginal effect of benevolence increases in the opposite direction as the level of social protection increases; however, this effect is not statistically significant (see Table 3). As for power and achievement, their moderating effects appear negligible since there is little change by the level of social protection, as shown in the figure.

Additionally, sensitivity checks assess the robustness of the current findings. In the Online Appendix, Table A4 presents the moderating effects of material deprivation instead of social protection. The interaction effect between individual universalism and material deprivation is consistently negative and significant in Columns (1) and (5); the other interaction effects are not significant in Column (5) where all the interaction effects are integrated. This finding indicates that the association between individual universalism and support for UBI is positive and significant in countries where the material deprivation rate is low, such as Sweden (0.8) and Switzerland (1.5). By contrast, the relationship is not significant in countries where the material deprivation rate is relatively high. This result is consistent with the finding that the moderating effect of social protection is significant in the association between individual universalism and UBI support. Moreover, Table A5 provides the results using logistic regression with country fixed effects, and Table A6 reports the results using ordered logistic regression with country fixed effects in which individual support for UBI is operationalized as an ordinal variable. The key findings are robust to these alternative estimations.