Theoretical framework

Social Investment Left

The main argument of the Social Investment LeftFootnote ⁴ for state intervention in capitalism is that the latter is prone to market failures, for instance, stemming from externalities, public goods, incomplete contracting and information problems of various kinds (Giddens, Reference Giddens1998; Keman, Reference Keman2011; Vandenbroucke, Reference Vandenbroucke and White2001). A pure market logic therefore does not incentivize sufficient long-term investments that yield to personal development and economic growth. The following examples illustrate this argument: although post-industrial economies need skilled and trained workforces, citizens will not always invest in – or just maintain – their human capital but instead often choose to forego training in favour of short-term income if they fear about making ends meet. Another example concerns imperfect capital markets, which prevent individuals from borrowing the individually and socially optimal amount for investments in human capital. Fear of losing basic security also results in short-term and short-sighted behaviours, whereas individuals with income security can develop long-term and, therefore, more stable, strategies. Indeed, “[t]here is more than money at stake in the poverty trap. The income poverty trap is also a security trap” (Haagh, Reference Haagh and Van Parijs2018, p. 82). Equally, insufficient child care provisions might prevent adequate investments in young children’s cognitive developments early in life with long-term adverse effects, in particular, regarding the intergenerational transmission of poverty and disadvantages (Hemerijck, Reference Hemerijck2017; Jenson, Reference Jenson, Bonoli and Natali2012).

What all these examples have in common is the idea that the state could increase the efficiency of markets through well-designed interventions (Giddens, Reference Giddens1998). Instead of promoting aggregate demand for labour via state investments to increase public or private consumption, such policies aim to increase workers’ employability by investing in their skills or increasing incentives to take up work and/or to break the intergenerational cycle of disadvantages (Hemerijck, Reference Hemerijck2018; Jenson, Reference Jenson, Bonoli and Natali2012).

The basic income creates an income floor which helps to motivate long-term investment strategies. UBI can for instance motivate a person to return to education and support long-term labour market integration. In line with the social investment ideology, a basic income, therefore, has a preventive and active function, in particular, as it does not reduce employment incentives but might motivate individuals to look for “better work.” In a life course perspective, social investment policies have the function to enable individuals to prevent and overcome difficult life events and life transitions without losing their capacity to support themselves economically (Kvist, Reference Kvist, Deeming and Smyth2017). UBI will fulfil precisely this function. The basic income is part of a systemic change of social rights in relation to welfare and work incentivizing individuals in a sustainable, long-term and preventive manner and helps to create, strengthen, maintain, re-establish and use individuals’ capabilities and skills (see Garritzmann et al., Reference Garritzmann, Häusermann and Palier2021, p. 6). We, therefore, expect that the Social Investment Left would support a basic income as it allows individuals to develop long-term personal strategies.Footnote ⁵ Table 1 summarizes the main differences of the three strands of left ideology.

Table 1. Variation within the Left.

Data, methods and results

To test our expectations, we rely on the eighth wave of the European Social Survey (ESS) which includes the following question about support for UBI:

Our dependent variable measures support for UBI by coding it 1 if respondents choose “in favour” or “strongly in favour” of the scheme and 0 otherwise. We use the Left–Right self-placement 11-points scale of the ESS to identify left-wing citizens by creating a dichotomous variable: 0–4 on the Left–Right scale is coded 1 to capture the Left, while other values are coded 0. Using this variable, we restrict our sample to left-wing respondents. The ESS does not include specific questions on concerns about capitalism, but it does include questions that allow us to create proxies of these concerns. Specifically, following our theoretical framework that distinguishes three types of concerns about capitalism, we identify a number of questions on attitudes toward redistribution, inequality, freedom and social investment to map out the different dimensions in the attitudes that left-wing individuals display in their criticisms of capitalism:

• • Pro-redistribution We rename the ESS variable gincdif “pro-redistribution” because it captures whether respondents think that governments should reduce differences in income levels.

• • Pro-equality. The second variable smdfslv, which we rename “pro-equality,” asks whether for “a fair society differences in standard of living should be small.”

• • Equal opportunity and being free. The third and fourth variables ask respondents about whether it is important (1) that “people are treated equally and have equal opportunities” and (2) “to make own decisions and be free.”

• • Pro-LGBTQ The fifth variable aims to capture cultural liberalism by the respondents’ agreement to the statement “Gays and lesbians free to live life as they wish.”

• • Pro-childcare and pro-parents. The sixth and seventh variables are based on two questions in concerning childcare and work–life balance, both of which have been shown to be key to the social investment paradigm (Ansell & Gingrich, Reference Ansell, Gingrich, P., S., H. and H.P.2015; Hemerijck, Reference Hemerijck2017; Morel et al., Reference Morel, Palier and Palme2012). Respondents choose value from (1) to (4) about the preferred degree of government responsibility. We call this variable “pro-childcare.” The second variable is about whether there should be benefits for parents to combine work and family even if it means higher taxes. We call this variable “pro-parents.”

• • Pro-unemployed education. Our eighth variable asks respondents whether they would spend more on education for unemployed at the cost of unemployment benefits.

All variables are coded in a way that higher values mean stronger agreement. We run a factor analysis that inductively extracts factors from these different variables thereby allowing us to identify latent ideology dimensions within the Left. The factor analysis on these eight variables identifies three different factors (indicated by the three factors with eigenvalues above 1 in Table 2) which we associate with different left-wing ideologies.

Table 2. Factor analysis on Left-wing respondents.

Not Factor analysis/correlation; Number of observations = 9,727. Method: principal-component factors; Retained factors = 3; Rotation: (unrotated) Number of parameters = 21; LR test: independent versus saturated: χ ²(28) = 4095.54, Prob>χ ² = 0.0000.

Factor 1 loads positively on being favourable to redistribution, equality (of outcomes and opportunities), to a lesser extent on childcare, and the least on being favourable to provide more education to the unemployed at the expense of unemployment benefits. We call this factor the exploitation factor, which speaks to the concerns of the Labourists Left since it captures people with concerns about the exploitative aspects of capitalism.

By contrast, factor 2 loads most positively on opportunities, freedom and supporting LGBT rights, as well as providing education instead of benefits to unemployed, but negatively on redistribution and equality. We call this the repression factor, which speaks to the concerns of Libertarian Left, since it captures people with concerns about the repressive nature of capitalism.

Finally, factor 3 attaches greater weight to childcare, benefits for parent and education for the unemployed at the cost of benefits but loads negatively on freedom, pro-LGBT and equal opportunity. There is also a weak negative loading with redistribution and equality. This captures pro-social investment attitudes. We call this the efficiency factor, which speaks to the concerns of the Social Investment Left. We provide summary statistics in Section A1.1 in the Appendix.

All our models include a standard set of control variables (Chrisp & Martinelli, Reference Chrisp and Martinelli2018; Lee, Reference Lee2018; Roosma & van Oorschot, Reference Roosma and van Oorschot2018): gender, age, education, income, source of income and occupations.Footnote ⁶ As our dependent variable is binary, we use logistic regressions but the results are robust to alternative estimation methods, see Section A3.4 in the Appendix. Since respondents are clustered in countries and there might be unobservable heterogeneity across countries, we include country-fixed effects and rely on robust errors clustered by country.

All controls in our analysis below are included but not shown (see Figure A2.2 in the Appendix for full results): male respondents are surprisingly more supportive, while low-income respondents, especially in elementary occupations, are more supportive of UBI than women and higher income earners. Age and being on a temporary work contract are not statistically significant, which is also the case for most occupational identifiers. Finally, education has a positive and statistically significant effect while most variables capturing sources of income are not statistically significant, except for the “other benefits” variable (see Table A2.3 in the Appendix).

Figure 1 reports the results for four different models. The first model pertains only to exploitation concerns, the second, only to repression concerns, and the third, only to efficiency concerns, while the fourth model includes all three types of concerns together. For each model, we report the point estimate of the semi-standardized coefficients, which indicates the effect of one standard deviation change in the independent variable on the log odds of the dependent variable. Around each point estimate, we report 95% as well as the 90% confidence intervals. When these confidence intervals overlap with the vertical zero line, the variable under consideration has no statistically significant association with support for UBI.

Figure 1. Ideological dimensions and support for UBI among left-wing respondents. Note: The sample has been restricted to respondents who identify as left. This sample was then used to compute the factor analysis and run the above logistic regressions. Country-fixed effects are included, but not shown. Standard errors are clustered by country (90% and 95% confidence intervals), the constant is positive but not statistically significant, Log pseudolikelihood = −4,268. Countries included are as follow: Austria, Belgium, Czechia, Estonia, Finland, France, Germany, Hungary Iceland, Ireland, Italy, Lithuania, Netherlands, Norway, Poland, Portugal, Slovenia, Spain, Sweden, Switzerland and the United Kingdom. All independent variables are standardized so that they have a mean of 0 and a standard deviation of 1, which makes the magnitudes of different coefficients directly comparable. The following controls are included but not shown: age, gender, education, fixed-term contract, income level, source of income and occupation. Full results are shown in graphical form in Figure A2.2, while the full regression results in the tabular format are shown in Table A2.3, both in the Appendix.

The first model suggests that exploitation concerns among left-wing respondents have a positive and statistically significant association with support for UBI. The magnitude of the effect is large in relation to other determinants of UBI support. This finding is at odds with the expectations of political theory work on UBI which we discussed in the theoretical section (see van Parijs, Reference Van Parijs and Van Parijs2018b). Recall that exploitation concerns load strongly on pro-equality and pro-redistribution preferences, which suggest that respondents with these characteristics are more supportive of UBI and hence see this scheme as achieving these aims. This suggests that – unlike many proponents of traditional institutions of the welfare state, which tend to see these institutions as the best protection against exploitation and UBI as representing a minimalist safety net – left-wing individuals who care strongly about equality and exploitation tend to favour a UBI.Footnote ⁷

We suspect that this can be explained – paradoxically – by the universal and highly decommodifying character of the UBI. Given that a UBI would be paid to everyone and without conditions, it achieves a very high level of decommodification, which has been argued by power resource theorists to be necessary to reduce capitalist exploitation of labour, especially its poorest and most precarious members (Korpi, Reference Korpi2006). In a context where many social policies have been retrenched, made more conditional and recommodified (Knotz, Reference Knotz2015, Reference Knotz2018) and where dualization has especially undermined the protection of non-standard workers (Emmenegger et al., Reference P., S., B. and M.2012), the UBI’s universalist and fully decommodifying characteristics might be especially appealing. Indeed, those who are especially concerned about the exploitation of the poor also tend to be supportive, as shown by Roosma and van Oorschot (Reference Roosma and van Oorschot2020)’s finding that support for targeting benefits to the poor and for UBI is correlated.

The second model shows that repression concerns are negatively associated with support for UBI. The effect is statistically significant at the 95% level and – again – at odds with the expectations of the political theory literature on UBI. Recall that repression concerns, characteristic of the Libertarian Left, loads strongly on equal opportunity, being free and cultural liberalism but negatively on pro-redistribution and pro-equality. Given that redistribution preferences are often used as a proxy for general welfare state support, one might therefore speculate whether the two dimensions could also be interpreted as measuring a general pro-welfare state preference dimension, which loads positively with support for UBI (Parolin & Siöland, Reference Parolin and Siöland2020; Roosma & van Oorschot, Reference Roosma and van Oorschot2020; Vlandas, Reference Vlandas2020). This result suggests that those who favour equality of opportunities, equal rights for LGBT and freedom of action are largely opposed to the UBI.

One possible reason for this counterintuitive finding is that equality of opportunity may not be understood as necessitating state intervention. Another reason might be that libertarians’ dislike of state interventionism is stronger than their desire for equal opportunities. However, while this negative correlation can be observed within the Left (ie. between left-wing respondents), repression concerns are not negatively associated with UBI support once we extend our sample to the wider population to include both left- and right-leaning respondents. Indeed, Figure A4.6.1 in the Appendix clearly shows that the coefficient is no longer statistically significant once both left and right-wing individuals are included in the analysis. Thus, repression concerns are not per se associated with lower support for UBI; only left-wing respondents who have particularly high repression concerns seem less likely to support UBI compared to other left-wing respondents.

The third model shows a significant positive correlation between efficiency concerns and support for UBI. This suggests that in line with our expectation, the Social Investment Left is indeed positively related to UBI. In this view, the UBI is seen as the ideal way of promoting individuals’ ability to invest in their human capital. Roosma and van Oorschot (Reference Roosma and van Oorschot2020), whose study also finds a positive relationship between support for social investment attitudes and UBI, suggest that a UBI is perceived as a modern, innovative idea as an alternative for the current welfare state, just as social investment has been found to be perceived in other welfare state studies. The weak negative loading of this social investment factor with pro-redistribution and equality attitudes in our analysis reveals that it is possible for left-wing individuals to support UBI even if they do not care particularly strongly about egalitarianism. Note that this association holds when extending the sample to include all respondents, including both left- and right-leaning individuals (Figure A4.6.1 in the Appendix).

Finally, if we include all three types of concerns in the same model, the effects also remain the same. Note further that these three types of concerns have the same effect when included on their own without any controls and they have strong predictive power. Indeed, when running a logistic regression only with the three independent variables capturing left-wing traditions, the model correctly predicts almost 60% of cases correctly (count R ² = 0.599). In Table A2.3 in the Appendix, we display the predicted probability of supporting a UBI when each independent variable is at its minimum versus maximum value. When exploitation concerns are held at their minimum value, this probability is 18% compared to 74% when it is held at its maximum value. For efficiency concerns, these numbers are 32% compared to 77%. For repression concerns, the effect is negative so that when these concerns are at their minimum value, the predicted probability is 67% compared to 48% when these concerns are at their maximum value.

We carry out a number of robustness checks. First, we rerun our analysis without including LGBTQ rights in our factor analysis. While we strongly believe that attitudes toward LGBTQ are a core part of the Left’s political ideology, we want to provide an analysis with purely economic variables. The results, shown in Section A3.1 in the Appendix, suggest our results are robust to this alternative specification. We find the same three factors, and their effects shown in Figure A3.1.2 are the same.

Second, we reproduce our empirical analysis twice using two distinct deductive operationalizations of our three traditions. Indeed, as we now note in the Appendix, we believe the more inductive logic underpinning our factor analysis represents a more transparent approach to capturing the different traditions. This approach also allows us to test the presence of different traditions in our survey data. However, one could nevertheless question the clarity of the factor analysis, especially since many variables load on the first factor (except pro-unemployed education). We therefore rerun the analysis with a deductive approach. Our deductive index of exploitation is a sum of the pro-redistribution, pro-equality and equal opportunity variables. Our deductive index of repression is a sum of equal opportunity variables, being free and pro-LGBT variables. Our deductive index of social investment approach is a sum of pro-childcare, pro-family benefits and pro-unemployed variables. As can be seen, the three resulting indices have similar means and standard deviations. When regressing our dependent variable on these three new deductive indices while keeping the same controls as before, our key results are unchanged (see Figure A3.2.4 in the Appendix). Note that equal opportunity is included in both the exploitation and repression indices because it seems important to both. However, we can also reproduce the same analysis while keeping this variable only in the exploitation index and excluding it from repression concerns index. The results in Figure A3.2.5 show that the findings for the three intellectual traditions of the Left remain similar.

Third, we consider it most important for our theoretical argument how respondents view themselves rather than how they vote, including whether they vote for left parties which can mean very different things across political and party systems. That being said, we reproduce our analysis using past vote for left-wing parties as an alternative to self-identification. To minimize concerns about comparability, we focus on Western European countries only. A simple cross-tabulation reveals that more than 90% that identified as substantially left-leaning (8, 9 and 10) on the Left–Right self-placement scale voted for a left party. Restricting our sample to left-wing voters in Western Europe, our factor analysis reveals the same three intellectual traditions of the Left: those with exploitation concerns score highly on pro-redistribution, pro-equality and equality of opportunities; those with efficiency concerns score most highly on childcare, family benefits and favouring education while those with repression concerns score highly on being free, pro-LGBT rights and equal opportunities. Rerunning our empirical analysis with the same estimation method and controls yields the same results as before (see Figure A3.5.1 in the Appendix).

Fourth, in Section A3.4 in the Appendix, we replicate our analysis using ordinal logistic regression analysis on the non-binarized version of our dependent variable (Figure A3.4.1 in the Appendix) and an ordinary least squares (OLS) linear probability model (Figure A3.4.2 in the Appendix). The key results remain unchanged.

Finally, we rerun our analysis on a larger sample including both left- and right-wing respondents (Section A4.6 in the Appendix). Figure A4.6.1 shows that exploitation and efficiency concerns continue to be positively associated with higher support for UBI, whereas repression concerns are no longer statistically significant. In Figure A4.6.2, we show that left-wing respondents are indeed more supportive of UBI than right-wing respondents. Including our set of concerns reduces the effect of the Left, consistent with the notion that these concerns capture a part of what it means to be left-wing, but the effect remains significant. Controlling for self-placing on the Left in this wider sample does not change the association between exploitation and efficiency concerns with UBI.

Conclusion and discussion

This article explores the relationship between support for UBI among left-wing individuals and different ideological strands of the political Left, captured by different concerns about capitalism. We argue that the nature of criticism against capitalism shapes the extent to which left-leaning individuals will support a basic income. With this, our study contributes to the emerging literature on the drivers and conditions of support for UBI. The relevance of a Left political ideology on welfare state support is underscored by a large political economy and political sociology literature (Bremer & Schwander, Reference Bremer and Schwander2019; Häusermann et al., Reference Häusermann, Kurer and Schwander2016; Jæger, Reference Jæger2006; Linos & West, Reference Linos and West2003). Indeed, left-leaning citizens are more likely to be supportive of the UBI, in particular, in the rich advanced democracies (Parolin & Siöland, Reference Parolin and Siöland2020; Vlandas, Reference Vlandas2020) but the UBI is nevertheless contested within the political Left (Van Parijs, Reference Van Parijs2018a). Disentangling and explaining this variation in support for UBI within the Left is the contribution of our study.

We base our analysis on the eighth wave of the ESS (2016). A factor analysis reveals three factors that capture exploitation, repression and efficiency concerns as main sources of criticism about capitalism. We then test the extent to which these dimensions are associated with support for UBI. Our results first confirm that not all left-leaning individuals are similar in their support for UBI and the type of concerns about capitalism matters for this variation. However, the relationship between different individuals with distinct views of the state and market appears empirically more complex than much of the theoretical-normative literature has so far implicitly assumed.

Indeed, both efficiency and exploitation concerns are associated with higher support for UBI among left-wing individuals, whereas repression concerns are associated with lower support. Thus, citizens who care about equality and redistribution are more supportive of UBI, despite the fact that political thinkers in the Labourist tradition criticize UBI as an inadequate solution to the exploitative nature of capitalism (see for instance the more critical contributions in van Parijs, Reference Van Parijs2018a). It seems that – in contrast to many arguments of Labourists’ political elites – many left-wing citizens consider the UBI as a desirable mean to fight poverty and inequality. Adherents of social investment who favour state interventions to address market failures and increase the efficiency of markets tend to be more favourable to UBI. We suggest that both social investment policies and UBI are perceived as a modern, innovative idea which represents an attractive alternative to the current welfare state institutions (see also Roosma and van Oorschot Reference Roosma and van Oorschot2020). The results for Libertarian Left do not conform to prior expectations (see van Parijs Reference Van Parijs2018c). In sum, we find support only for the expectation from the welfare state literature that Social Investment Left should be associated with UBI but the opposite of what the political theory literature on UBI would make us expect: Labourists appear supportive, whereas Left Libertarians appear less supportive of UBI. In that sense, our study challenges the normative literature on UBI: how can the opinions of Labourist’ elites and citizens diverge so strongly?

Our findings are also relevant for welfare state scholars and from a policy-making perspective: without the support of the political Left, it seems unlikely that any government would implement UBI. Yet, given the heterogeneous support for UBI within the Left, one wonders about other possibly cross-cutting political coalitions that would enable an implementation of the policy. For instance, one might envisage social–liberal coalitions, which have successfully implemented social investment reforms in countries where the political Left alone did not achieve it (see for instance Kübler, Reference Kübler2007) but have not attracted much attention in the welfare state literature. Such a cross-cutting coalition, however, entails the risk that only a meagre version of UBI would be implemented.

Our findings call for more extensive research on the sources of variation of UBI support within the political Left, the specific characteristics of UBI that trigger support or rejection by individuals espousing different ideological dimensions of the Left and the potential coalition dynamics for a larger-scale introduction of UBI. We are also fully aware of the limitations of our analysis. It is entirely possible that part of our unexpected findings are due to imperfect empirical measurement of our main explanatory concepts for which we rely on the information given by the ESS. Yet, to the best of our knowledge, no alternative surveys offer more precise information about support for UBI and about different criticisms of capitalism, let alone in a cross-national survey to capture differences in the structure of the welfare state and the economic system. Nevertheless, we acknowledge limitations with respect to our measurements. Our measurement of exploitation concern for instance is partly problematic because it appears to be closely related to general support for welfare state intervention. Thus, further research should include data collection effort and develop questions more closely linked to repression concerns, including variables that control for general welfare state support in a distinct way.