Introduction
Attention for universal basic income (UBI) policy has peaked in recent years: the radical idea of universal and unconditional social policy reached media and political agendas in countries like Finland (De Wispelaere et al., Reference De Wispelaere, Halmetoja and Pulkka2018), the Netherlands (Groot et al., Reference Groot, Muffels and Verlaat2019) and Canada (Forget et al., Reference Forget, Marando, Surman and Urban2016). The OECD (Browne and Immervoll, Reference Browne and Immervoll2017) expressed its interest in a recent policy brief on the feasibility of basic income. This renewed interest has not eluded academic proponents, who have argued that basic income has become a legitimate policy alternative (Reed and Lansley, Reference Reed and Lansley2016).
In response to this popularity, students of welfare legitimacy seek to explain why such a policy would attract popular support. First, representative survey studies have shown that popular support for UBI is most strongly motivated by redistributive interests and the principled reduction of inequalities (e.g. Delsen and Schilpzand, Reference Delsen, Schilpzand and Delsen2019; Roosma and van Oorschot, Reference Roosma and van Oorschot2020), lending credence to the idea that UBI is mainly perceived as a ‘free lunch’ (Friedman, Reference Friedman1975). Chrisp et al. (Reference Chrisp, Pulkka and García2020) and Rincón García (Reference Rincón García2021) find that taxation mechanisms have the strongest impact on support for UBI. Vlandas (Reference Vlandas2021) similarly argues that the political left, including labour unions, plays a pivotal role in coalition formation surrounding UBI. Schwander and Vlandas (Reference Schwander and Vlandas2020) find that leftist support for UBI is grounded primarily in a ‘labourist’ ideology, suggesting that concerns of over-exploitation of the working classes comprise an important driver of UBI support. This is contrasted with UBI support amongst politicians, where especially the green parties and activist movements endorse UBI (Liu, Reference Liu, Caputo and Liu2020; Perkiö, Reference Perkiö2020). Other studies focus on anti-immigrant sentiments as the main obstacle to UBI support (Bay and Pedersen, Reference Bay and Pedersen2006; Stadelmann-Steffen and Dermont, Reference Stadelmann-Steffen and Dermont2020). Parolin and Siöland (Reference Parolin and Siöland2020) also point to the importance of welfare chauvinism as an objection to UBI. Rossetti et al. (Reference Rossetti, Roosma, Laenen and Abts2020) find that objections to UBI are grounded in the (un)deservingness of the unemployed. Overall, existing survey studies have been fairly successful in explaining UBI support using established theories of welfare ideology.
However, existing opinion studies rely almost exclusively on survey data, a method with drawbacks. In the best case, constructing survey items based only on expert knowledge risks overlooking important aspects of UBI discussions (e.g. Laenen et al., Reference Laenen, Rossetti and van Oorschot2019). In the worst case, empirically ungrounded surveys risk capturing framing effects, ‘creating’ preferences rather than measuring pre-existing ones. These concerns are especially relevant in the case of UBI because the proposal is flexibly defined (Chrisp and Martinelli, Reference Chrisp, Martinelli and Torry2019), and arguments used to justify its implementation evolve (e.g. Perkiö, Reference Perkiö2020). Moreover, particularly in the case of UBI, the general public is often underinformed, further fuelling fears of unreliable support measures (Rossetti et al., Reference Rossetti, Roosma, Laenen and Abts2020, see also Dermont and Weisstanner, Reference Dermont and Weisstanner2020; Roosma and van Oorschot, Reference Roosma and van Oorschot2020).
Two related kinds of literatures already suggest that the UBI debate adopts arguments not included in current survey studies on UBI support. First, the social justice literature provides a range of deductive arguments for why unconditional and universal welfare is justified. UBI is defended as an emancipatory policy that provides ‘real freedom’ and autonomy (Van Parijs, Reference Van Parijs1991; Birnbaum, Reference Birnbaum2012), to extend our notion of work beyond productive employment (Offe, Reference Offe, Widerquist, Noguera, Vanderborght and De Wispelaere2013), towards a broader concept of work that includes valuing unpaid labour such as childcare and volunteering (Jordan, Reference Jordan, Widerquist, Noguera, Vandeborght and Wispelaere2013). Second, qualitative framing studies show how the policy is portrayed in media and politics. These studies generally indicate a much broader discussion in politics and media than suggested by survey research. For example, Perkiö (Reference Perkiö2020) shows that Finnish politics is most concerned with the activating potential of UBI, while newspapers across various countries predominantly frame UBI in the context of automated labour and reducing the need for full-time employment (Perkiö et al., Reference Perkiö, Rincon and Van Draanen2019). However insightful, these studies do not cover the public response to UBI policy.
In response, we aim to confirm, correct, and broaden the scope of survey studies on UBI support through a content analysis of peak discussions on Dutch Twitter. Twitter has been heralded for providing researchers with relatively open access to a unique source of data: a platform that hosts public debate on policy issues without formal or technical restrictions (Mutz, Reference Mutz2006; Mejova et al., Reference Mejova, Weber and Macy2015). The medium has been leveraged to study online political phenomena such as political campaigns (Conway-Silva et al., Reference Conway-Silva, Filer, Kenski and Tsetsi2018), political polarisation (Barberá et al., Reference Barberá, Jost, Nagler, Tucker and Bonneau2015), and social movements (Ince et al., Reference Ince, Rojas and Davis2017). The platform is also particularly relevant to study the debate surrounding UBI policy in the Netherlands: Dutch Twitter uniquely hosted multiple viral discussions of UBI in the period 2014-2016. The online attention led political elites to engage with UBI – especially on the municipal level – and facilitated the setup of experiments with unconditional social assistance (for an extensive discussion see Groot et al., Reference Groot, Muffels and Verlaat2019). Moreover, the range of arguments discussed on Twitter is likely to be broad. Even though Twitter discussion is often triggered and structured by political elites and mainstream media coverage (e.g. Russell Neuman et al., Reference Russell Neuman, Guggenheim, Mo Jang and Bae2014), the Twitter audience is free to engage within these boundaries, leaving space for interpretations of UBI outside the established welfare policy discourse. The analysis of arguments used on Twitter should therefore complement existing survey studies on support for UBI policy.
In the remainder of this article, we first elaborate on the context of the UBI debate in the Netherlands and our methodological procedures. Next, we discuss the identified controversies in detail, particularly in relation to existing studies on welfare legitimacy and UBI. Finally, we present our conclusion, the discussion of limitations, and our recommendations for future studies.
UBI debate in the Netherlands
Compared to most other countries, the Netherlands has witnessed a relatively active and consequential revival of the UBI discussion. First, the debate followed an earlier peak of interest in the 1980s (Groot and van der Veen, Reference Groot, van der Veen, van der Veen and Groot2000). The idea featured in a report by a think tank advising on government policy (WRR, 1985). Mainstream political parties have however never been strong proponents of UBI policy. Labour parties and Christian democrats rejected these proposals for being too low for those who need them and superfluous for those that do not. Particularly the labour party (PvdA) preferred work-based social security mediated by labour union negotiations. The Green Left and liberal democrats (D66) have shown sympathy towards UBI proposals, but never consistently argued for them.
In the most recent wave, illustrated by figure 1, renewed interest for UBI was initially driven by ‘fringe’ journalism, amplified by Twitter audiences. Rutger Bregman (Reference Bregman2013, Reference Bregman2014) initially published an article and a book on the online journalist platform ‘the Correspondent’ with UBI as a flagship proposal. Based on Bregman’s work, the future affairs program Tegenlicht (2014, 2015) aired several documentaries on Dutch national broadcast television. This drew public attention to the idea especially on Twitter, where the two subsequent episodes became trending topics. All three episodes use the threat of automated labour to argue for the necessity of UBI policy. If unaddressed, the automation of labour would increase economic inequalities and create an insecurely employed underclass. A second tenet criticises the existing welfare state as inefficient and unable to protect citizens from poverty or guide them towards employment. The most notable reasons mentioned to implement UBI policy are (a) providing income security in the face of automation, (b) allowing the working poor to invest in education, childcare, and basic goods, and (c) increasing happiness, social trust, and reducing poverty stress. We provided summaries of the documentaries in Appendix B.
Daily number of tweets and replies mentioning Universal Basic Income
After the third episode aired, local politics had caught up with the hype. In response to the attention for UBI and the lobby for experimenting with such policy, a dozen of Dutch municipalities started experiments with unconditional social assistance. It seems the public enthusiasm for UBI came at the right time: the decentralisation of social assistance benefits from the national to the municipal level was accompanied by financial struggles, and some municipal councils doubted the effectiveness of activating incentives (Groot et al., Reference Groot, Muffels and Verlaat2019). While these experiments were ‘inspired’ by the UBI discussion (p. 280), they were framed as ‘trust experiments’ to avoid the political controversy surrounding UBI policy. The trust experiments investigated whether removing ‘stick-and-carrot’ incentives attached to social assistance benefits would improve the well-being of social assistance beneficiaries and their chance of finding a job. Activating incentives were theorised to generate additional stress (e.g. Mani et al., Reference Mani, Mullainathan, Shafir and Zhao2013) and distrust towards welfare institutions (Bohnet et al., Reference Bohnet, Frey and Huck2001), consequently impeding the ability of welfare recipients to find a job. Inversely, trust in welfare recipients would foster cooperative attitudes and enable them to find a job faster. Uncoincidentally, as we show below, these arguments were also put forward in the Twitter debate.
Content analysis
This article employs content analysis (see Krippendorff, Reference Krippendorff2018) of tweets to identify and classify emergent arguments used in favour and against UBI policy in public debate. We take an inductive or emergent approach to content analysis (e.g. Mayring, Reference Mayring, Bikner-Ahsbahs, Knipping and Presmeg2015). In this section, we elaborate on the procedures underlying the development of our coding scheme: the data and sample selection, coding strategy, and validation procedure.
We employ a qualitative content analysis rather than a computational text analysis approach. Computational approaches either cluster words into topics based on their co-occurrence across documents (for an introduction see Blei and Lafferty, Reference Blei, Lafferty, Srivastava and Sahami2009) or training an automated classifier based on labelled input data (for an overview see Kowsari et al., Reference Kowsari, Jafari Meimandi, Heidarysafa, Mendu, Barnes and Brown2019). While the unmistakable advantage of these tools lies in their ability to deal with enormous datasets, qualitative content analysis still outperforms such computational approaches in two ways. First, content analysis yields a more comprehensive and fine-grained set of arguments compared to text clustering approaches. Topic models are well suited to identify broad categories such as ‘politics’, ‘economy’ or ‘entertainment’, but are less able to distinguish distinct arguments within a single topic, especially in relation to the broader socio-cultural context (Zamith and Lewis, Reference Zamith and Lewis2015: 312; Krippendorff, Reference Krippendorff2018: 210). Human coders, instead, interpret and validate tweets to derive the arguments they contain, and therefore achieves more reliable and more nuanced results, albeit on a much smaller scale. Second, text classification requires large volumes of labelled training data – often based on content analysis – to produce accurate results (e.g. van Smeden et al., Reference van Smeden, Moons, de Groot, Collins, Altman, Eijkemans and Reitsma2019). Labelling the required amount of training data for policy debates quickly becomes unfeasible, because the range of arguments is broad (eventually we identified fifty-five unique arguments) and some arguments are much more frequent than others. Moreover, automated classification would add only argument frequencies without identifying new arguments, which is not the main focus of this article. When the arguments are many and language is varied and nuanced – as is often the case – qualitative content analysis produces an equally valid range of arguments.
Regardless, a content analysis of tweets comes with its own limitations. Twitter datasets often contain spam and unrelated tweets, and debates on Twitter are often difficult to capture completely because of search restrictions and participants deleting their tweets and accounts over time (Karpf, Reference Karpf2012; Ruiz-Soler, Reference Ruiz-Soler2017). Moreover, when aiming to study representative public opinion, it must be noted that Twitter users are not representative of the general population (e.g. Barberá and Rivero, Reference Barberá and Rivero2015). The political left is somewhat overrepresented on Dutch Twitter (Wiergina et al., Reference Wiergina, Gorzeman, van Geenen and Schäfer2018) and those engaging in political debates are more likely to be male and higher educated (van Klingeren et al., Reference van Klingeren, Trilling and Möller2021). In addition, content analysis can only include a limited number of tweets due to the time-intensive analytical process. In light of these restrictions, we aim to identify a large and comprehensive variety of arguments. Argument frequencies are not representative of the societal debate and must be interpreted with care. Moreover, qualitative analysis generally relies on the interpretation of researchers, to ascertain that arguments are used. While misinterpretation cannot be prevented entirely, it is minimised through establishing intercoder reliability and a transparent report of coding processes and results.
Data and sample selection
Tweets are gathered first by searching for ‘basic income’ (‘basisinkomen’) in the Twitter search engine (https://twitter.com/search-home). The Dutch term is exclusively used to refer to UBI, also including variations such as ‘universal basic income’. We used the Twitter API to identify replies to the tweets found by the search engine. To capture full conversations, we include up until the ninth reply to tweets mentioning UBI. We decided to stop here because the number of ninth reply tweets is already negligible (1.1 per cent of tweets) and the discussion tends to become redundant or off-topic.
We purposively select all tweets posted on three essential days: the airing of the second documentary (2014-09-21), the third documentary (2015-04-12), and the day of the announcement of the trust experiments (2015-08-05), amounting to a total of 5687 tweets. Although a random sampling strategy is common in the content analytic approach, we have several reasons to opt for a purposive sample supplemented by a random sample. First, the key events on our purposefully selected days attracted the most attention from our Twitter audience and illustrate their importance for the broader discussion. To take a purely random sample of days would be to ignore the event-centred nature of (social) media, where some days are simply more important for some topics than others. Second, with this sampling strategy, we aim to maximise the variation in employed arguments. Particularly opponent arguments are more equally represented on high-attention days. Third, social media audiences actively engage with arguments presented in traditional media. The purposive sample thus reflects an active public response to the most influential (but selective) media events surrounding UBI policy in the Netherlands.
While guided by these events the arguments here go beyond a mere rehashing of the documentaries. First, a supplementary analysis based on a random sample of ten days reveals no alternative arguments left uncaptured by the purposive sample (see Appendix C). Second, Appendix C shows roughly the same arguments are used on different random days, pointing to their broader relevance to the Twitter debate. Third, we find tweeted arguments (especially counterarguments) that remain unaddressed in the documentaries that inspired the debate (see Appendix B). These reflections suggest that the analysis includes a comprehensive (but probably not exhaustive) set of arguments employed on Twitter.
The purposive sample also excludes some peak events, most notably the first documentary and the response to the downvoted Swiss referendum. This exclusion is first based on the practical limitation of qualitatively analysing many tweets. After reaching a point of saturation most arguments are included, as also demonstrated by the supplementary random sample in Appendix C. Moreover, the second documentary seems to have been the starting point of a period of heightened attention on Twitter, perhaps because of its more pragmatic emphasis on the implementation and comparison to existing welfare arrangements (see Appendix B for a summary and comparison of the episodes). The sheer number of responses makes the second documentary more relevant to analyse. Finally, we also exclude the response to the Swiss referendum for its limited relevance to the Dutch debate. Excluding this event is thus also a matter of methodological purity.
Coding strategy
We take an inductive approach to coding, where the aim is not to be led by existing theory, but rather to gain ‘an understanding of the material in terms of the material’ (Mayring, Reference Mayring, Bikner-Ahsbahs, Knipping and Presmeg2015: 374). The development of a coding scheme is an iterative procedure where initial categories are evaluated, combined, or split as more data are added (Tashakkori and Teddlie, Reference Tashakkori and Teddlie2010). The coding process roughly moves from open coding (where categories are discovered) to axial coding (the merging and splitting of categories according to their similarities and distinctions) and selective coding (the application of a crystallised coding scheme) (Corbin and Strauss, Reference Corbin and Strauss1990, see also Cho and Lee, Reference Cho and Lee2014).
Up to three categories are coded per tweet. When no category could be discerned, tweets are assigned to a miscellaneous category (e.g. unclear, spam, unrelated, emotional expression). 34.1 per cent of tweets fall into one of these miscellaneous categories. This shows that ‘shitposting’, casual conversation and emotional expression are common on Twitter. The 65.9 per cent of relevant tweets, however, also show that the platform can occasionally function as a deliberative space (see Rogers, Reference Rogers, Weller, Bruns, Burgess, Mahrt and Puschmann2014).
Tweet content is often difficult to understand in isolation, especially when it is part of a larger conversation. When the argument in a tweet is unclear, the coder first reads the conversational thread in which tweets are embedded and then the other tweets of that (anonymised) author. Tweets containing links to newspaper articles or blogs are coded based on the main argument(s) in the article title or abstract. The minority of tweets linking to video and audio material is not coded substantively. In the case of quoted tweets, we code only the comment of the author.
We assessed the reliability of the final coding scheme using intercoder reliability. The sample was constructed using a stratified random sample: for each category (including the miscellaneous category) we randomly selected nine tweets, amounting to a total of 414 tweets (7.3 per cent of the full sample). Two coders agreed on 96.7 per cent of labels. Correcting for agreement by chance yields an average reliability of κ=.430 across categories. Although guidelines vary, this reliability can be considered ‘fair to good’ (Fleiss, Reference Fleiss1981: 218). While intercoder reliability estimates of content analyses on Twitter are usually somewhat higher – if reported at all – we expected and accept a less than exceptional reliability due to the unusually extensive coding scheme (55 categories) and a relatively ambiguous type of content: substantive arguments rather than speech acts or broad topics.
Results
Careful analysis of the data shows that the UBI discussion follows four central controversies: economic redistribution, welfare conditionality, welfare state efficiency, and structural unemployment. Within these controversies, we recognise debate on policy principles (Jaeger, Reference Jaeger2006), welfare critiques (Roosma et al., Reference Roosma, van Oorschot and Gelissen2016), and recipient deservingness (van Oorschot et al., Reference van Oorschot, Roosma, Meuleman and Reeskens2017). We will elaborate on each of these lines of debate in the following section.
Most fundamentally, participants try to frame UBI as either redistributive, efficient, or liberating. Opponents stress the redistributive aspect of UBI, aiming to frame the policy either as a tax burden or as a form of welfare retrenchment. Some proponents, on the other hand, actively reject the redistributive frame and instead emphasise the importance of unconditionality. Others still focus on fixing the shortcomings of the existing welfare system. We thus recognise different dimensions of controversy, pointing to the relevance of frames for the legitimacy of UBI (e.g. Bay and Pedersen, Reference Bay and Pedersen2006; Perkiö, Reference Perkiö2020, see also Chong and Druckman, Reference Chong and Druckman2007). We also recognise the discursive tension between liberal and egalitarian versions of the UBI proposal (e.g. De Wispelaere, Reference De Wispelaere2016; Chrisp and Martinelli, Reference Chrisp, Martinelli and Torry2019), which seems especially emphasised by opponents.
Figure 2 shows the occurrence frequency of each controversy. Arguments regarding economic redistribution are most frequently adopted, followed by welfare conditionality, welfare state efficiency, and structural unemployment. Especially discussion surrounding the affordability of UBI contributes to the topical dominance of the redistributive fault line. All the arguments within each fault line, each including a brief description, can be found in Appendix A. The frequency of each argument can be found in Appendix C.
Observed frequency of each controversy in the manually coded data
Economic redistribution
Redistributive justice is a key aspect of welfare politics, and this is also the case with UBI. What constitutes a fair distribution of the tax burden is captured in two opposing ideologies: egalitarian and liberal (e.g. Jaeger, Reference Jaeger2006). Framing of UBI as redistributive policy occurs both by proponents (who defend the social right to income) and opponents (who want to ‘expose’ UBI as redistributive policy or, inversely, as welfare state retrenchment). In the debate, the principle of providing minimum income assistance is barely contested. Instead, this discussion revolves mainly around the level of income assistance and the question of taxation.
Most basically, participants present UBI as ‘a human right’ Footnote 1 referring at least in part to Marshall’s (Reference Marshall1950: 149) concept of social rights. A basic tenet of UBI is indeed to be a ‘minimum living wage’ Footnote 2 , ‘effective against poverty’ Footnote 3 . Some arguments take this idea a step further into the domain of economic equality, by presenting UBI as the ‘solution to income inequality’ Footnote 4 , or the ‘answer to the excesses of capitalism’ Footnote 5 . Granting the benefit to everyone would ‘prevent stigmatization’ Footnote 6 of the unemployed, in contrast to socially divisive targeted benefits (see also Larsen, Reference Larsen2008).
On the opposing side, we first find the ‘labourist’ opposition (e.g. Navarro, Reference Navarro and Van Parijs2018, see also Schwander and Vlandas, Reference Schwander and Vlandas2020), contending that UBI will retrench welfare and increase poverty. In one apt summary, UBI is described as ‘anti-social or unaffordable’ Footnote 7 . On the one hand, UBI would not be able to sufficiently support those unable to work, so ‘those who cannot do something on the side face a position of poverty’ Footnote 8 . On the other hand, an untargeted policy such as UBI is regressive and thus would unfairly benefit the rich: ‘why should those that can take care of themselves receive it? Let the people that can’t get it’ Footnote 9 . In combination with suspicion towards liberal proponents (‘it is a liberal invention’ Footnote 10 ), the socialist opposition concludes that UBI ‘leads to the demolition of social security’ Footnote 11 .
On the other side of the opponent spectrum, we also recognise the liberal opposition, warning that UBI will lead to higher income tax and extensive redistribution of wealth and income: ‘it is a complete illusion to think that UBI represents the end of ‘justice-driven redistribution’’ Footnote 12 . More concretely, this is expressed in questions of affordability (‘Who will pay for it?’ Footnote 13 ), or in explicit references to income tax – ‘’free money’ has to be stolen (taxed) from someone’ Footnote 14 or UBI is ‘nothing but redistribution’ Footnote 15 , and ironically a quote from UBI proponent Friedman (Reference Friedman1975) himself: ‘there is no such thing as a free lunch’ Footnote 16 . More extreme statements compare UBI to ‘disguised Marxism’ Footnote 17 that ‘smells like Lenin and his comrades’ Footnote 18 , referring both to the redistributive and utopian aspects of UBI.
In contrast, some participants actively contest the redistributive frame. UBI is ‘not redistributing between rich and poor’ Footnote 19 ; ‘people who earn money do not gain or lose anything’ Footnote 20 . Questions of ‘who is going to pay’ are often resolved by referring to the costs saved by reforming the existing welfare bureaucracy Footnote 21 and reduced health costs due to increased well-being Footnote 22 . From this point of view, redistributive issues are subordinate to the unconditional provision of income.
In sum, there is much disagreement regarding the redistributive character of UBI. Some argue it is redistributive, others as regressive, and others still dismiss the redistributive frame entirely. This controversy highlights the relevance of the redistributive ideology in the case of UBI. Moreover, some opponents invoke the (un)deservingness of the rich and needy to oppose the universal aspect of UBI.
Welfare conditionality
The second line of discussion surrounds the work obligation or the degree to which the provision of social rights should be contingent on labour market participation. The two opposing positions on this axis are those arguing for freedom versus those arguing for responsibility.
In the freedom perspective, unconditional income such as UBI serves primarily to release the work obligation (see e.g. Vallentyne, Reference Vallentyne, Vallentyne and Steiner2000). This perspective views capitalism as ‘a repressive system that forces individuals to commodify themselves on the labour market’ (Schwander and Vlandas, Reference Schwander and Vlandas2020: 240). The left-libertarian critique deems the work obligation unethical, and the government apparatus that enforces the work obligation repressive. Releasing this obligation amounts to ‘real freedom’ (Van Parijs, Reference Van Parijs1991; see also Birnbaum, Reference Birnbaum2012: 32), greater individual wellbeing, and social participation.
From this point of view, freeing people from the obligation to work is a matter of social justice (see also Dahrendorf, Reference Dahrendorf, Holme and Elliott1988: 118). UBI is conceptualised as ‘liberating [and] emancipating’ Footnote 23 , providing ‘freedom of choice’ Footnote 24 , and representing ‘the end of wage slavery’ Footnote 25 . The explicit radical idea of ‘money without [work] requirement’ Footnote 26 – or the more provocative ‘free money’ Footnote 27 – is deemed a ‘human right’ Footnote 28 . The freedom perspective rejects the work obligation and instead argues that UBI constitutes the ultimate form of individual freedom. The concept of justice as freedom goes beyond the traditional principle of economic equality (Marshall, Reference Marshall1950) because it explicitly requires unconditional benefits in addition to universal access to welfare.
The freedom gained by UBI serves to improve individual well-being and self-actualisation. Society is thought to ask too much from its workers (‘why do we work ourselves to death?’ Footnote 29 ), resulting in ‘idiotic work pressure’ Footnote 30 and ‘antidepressants and sleeping pills’ Footnote 31 to cope with it. Contrarily, a society that implements UBI is ‘much less stressful’ Footnote 32 because ‘the fear to lose [your] job disappears’ Footnote 33 . Consequently, with a basic income ‘people become happier and healthier’ Footnote 34 , preventing ‘psychological conditions’ Footnote 35 and encouraging ‘self-development’ Footnote 36 .
The freedom perspective is also critical towards the performance of targeted activation policies. First, activation policies are deemed ‘inhumane’ Footnote 37 and ‘restrictive’ Footnote 38 ‘bully-policies’ Footnote 39 , that ‘force people to enter humiliating trajectories to look for non-existing jobs’ Footnote 40 . This coincides with studies finding that cultural individualists oppose commodifying welfare reforms (Achterberg et al., Reference Achterberg, van der Veen and Raven2014). Second, targeted welfare policy is considered stigmatising and socially divisive (see e.g. Larsen, Reference Larsen2008). Instead, a UBI would ‘make a big commune of the Netherlands’ Footnote 41 , restoring ‘social capital’ Footnote 42 and ‘repairing the connection between individuals and society’ Footnote 43 . This position thus directly opposes the ‘enabling’ or ‘workfare’ approach to welfare, claiming that the employment incentives embedded in welfare policy result in social exclusion and isolation rather than social inclusion and participation (see also Calnitsky, Reference Calnitsky2016).
In defence of welfare conditionality, the work obligation is seen as a responsibility to society (White, Reference White2006). Individuals are deemed responsible to provide for themselves as best they can, and social support – rather than a social right – is only justified in cases of need. The ‘principle of individual responsibility’ Footnote 44 forms a moral objection against UBI. People should ‘generate income by working’ Footnote 45 because ‘we aren’t toddlers that need care from mommy and daddy’ Footnote 46 . This principle is given further credence by invoking deservingness objections (‘everyone a [UBI] and no-one who deserves it’ Footnote 47 , ‘a [UBI] for paupers, jerries and sloths’ Footnote 48 , it only attracts ‘workshy scum’ Footnote 49 etcetera). Such expressions accord with much prior work that highlights the importance of moral objections to unconditional welfare (e.g. Groot and van der Veen, Reference Groot, van der Veen, van der Veen and Groot2000; Sloman, Reference Sloman2018) and its relation with perceived recipient deservingness (van Oorschot et al., Reference van Oorschot, Roosma, Meuleman and Reeskens2017).
The reciprocity objection also elicits concerns about the consequences of UBI. Some view it as a ‘perverse incentive’ Footnote 50 which ‘lowers the incentive to work’ Footnote 51 : with a UBI, people will ‘go to Thailand’ Footnote 52 . Others apply the framework of ‘welfare chauvinism’ (De Koster et al., Reference De Koster, Achterberg and van der Waal2013) and ‘welfare magnetism’ (Bommes and Geddes, Reference Bommes and Geddes2000). Basic income is supposed to ‘attract even more immigrants’ Footnote 53 , ‘bringing the whole world to here’ Footnote 54 , and creating ‘a huge attraction [of immigrants] from within the EU’ Footnote 55 . These findings confirm the continued relevance of welfare chauvinism for the legitimacy of UBI (Bay and Pedersen, Reference Bay and Pedersen2006; Parolin and Siöland, Reference Parolin and Siöland2020; Stadelmann-Steffen and Dermont, Reference Stadelmann-Steffen and Dermont2020).
In sum, the work obligation is the second main controversy concerning support for UBI policy. For some, this obligation is repressive, while others consider it a social or individual responsibility. In addition to these principles, the consequences of releasing the work obligation are hotly debated. Each side of the spectrum has its activation logic, with some contending that work incentives crowd out intrinsic motivations and others that extrinsic incentives are needed to prevent widespread idleness. Opponents in particular object to UBI by invoking the (un)deservingness of the unemployed, the needy, and immigrants.
Welfare state efficiency
Concerns over the efficiency and effectiveness of welfare state policy generate two opposing positions, regarding trust and incentives, both in favour of UBI. Both aim to remove work disincentives from existing policy and reduce costly activation policies, to increase labour market participation. They are opposed, however, in their approach. From a trust position, removing ‘stick-and-carrot’ incentives will build the cooperative attitude needed to find a job. The incentive position, in contrast, argues that work participation increases when a below subsistence level benefit is offered.
The inefficiencies of the existing welfare system are universally recognised. The idea that inefficient and ineffective policies undermine welfare legitimacy is well known in the literature (Roosma et al., Reference Roosma, van Oorschot and Gelissen2016). The current system of tax supplements and deductions is referred to as a ‘deduction circus’ Footnote 56 and ‘circulation machine’ Footnote 57 , ‘administrative hell’ Footnote 58 , and ‘benefit Moloch’ Footnote 59 . Recurrently, the government bureaucracy is referred to as ‘work provision’ Footnote 60 for public officials, stating that ‘it is a societal choice to keep people employed like this’ Footnote 61 . In essence, the problem is that ‘the current system of social provisions is inefficient’ Footnote 62 because it is considered to create ‘useless, empty costs’ Footnote 63 with large investments in ‘senseless courses and reintegration programs’ Footnote 64 , while its complexity makes it ‘sensitive to fraud’ Footnote 65 . The main selling point of basic income from an efficiency point of view is to simplify the social security system. One of the main public advocates of this position speaks of ‘an alternative to the circulation machine’ Footnote 66 . The focus is to ‘replace all benefits with a basic income’ Footnote 67 , so that ‘with basic income unemployment benefits, social assistance and retirement funds cease to exist’ Footnote 68 and ‘deductibles and supplements will be abolished’ amounting to ‘just a simpler system of taxation’ Footnote 69 . In contrast to the other arguments, the efficiency argument heavily accentuates cost-effectiveness.
In contrast to prior lines of discussion, the trust and incentive positions also agree on the idea that UBI should reduce welfare dependency by motivating claimants to accept paid employment: UBI ‘stimulates employment’ Footnote 70 , ‘catalyses activity’ Footnote 71 and ‘increases economic activity and earning potential’ Footnote 72 . Work disincentives in the welfare system are also a shared concern. UBI would remove ‘paralyzing’ Footnote 73 work disincentives, mainly referring to the poverty trap. ‘Social assistance benefits do not motivate people to work’ Footnote 74 , because of ‘the fear to lose their benefit’ Footnote 75 . ‘Earning extra next to the benefit is punished [with] a reduction [of the benefit]’ Footnote 76 . With UBI ‘the impulse to work is greater because your benefit is not reduced’ Footnote 77 . Thus, the restrictive rules surrounding (part-time) work form an obstacle that prevents people from re-entering the labour market.
Despite all this consensus, there is a fundamental difference in the perceived proper method to activate the unemployed. First, from a trust perspective, removing work incentives is supposed to provide mental space and improve government-client relations, which in turn motivate the unemployed to seek paid employment. With UBI ‘people work from their own motivation rather than to survive’ Footnote 78 . One participant notes that ‘when you can choose how to live your life, you will achieve more and feel valued’ Footnote 79 . Without employment incentives, ‘people will work out of ideals or for extra luxury’ Footnote 80 . The cooperative ‘positive’ Footnote 81 view of human nature lies at the foundation of this argument. Activation requires trust in the cooperative attitude of others, to return to ‘a society based on trust’ Footnote 82 or ‘a more social society’ Footnote 83 . This reasoning also strongly resonates in the UBI pilots in the Netherlands, aptly named the ‘trust experiments’ (Groot et al., Reference Groot, Muffels and Verlaat2019). More generally, the idea that stick-and-carrot incentives crowd out pro-social behaviour is central to the work of Bowles (Reference Bowles2016) and this form of UBI is closely related to ideas of social investment (Hemerijck, Reference Hemerijck2018).
The incentive position, on the other hand, argues that employment can be incentivised by a below subsistence level benefit. Providing a ‘low UBI [with] unrestricted work on the side’ motivates paid employment, referred to as a ‘pepper-in-the-butt system’ Footnote 84 . The benefit ‘should be too low to live from, because you want to stimulate people to do something on the side’ Footnote 85 . People must ‘earn on the side to supplement their insufficient [UBI]’ Footnote 86 . On the supply-side, low-paid jobs can be created by deregulating the labour market. For example, abolishing ‘the market-disruptive minimum wage’ Footnote 87 would create jobs because ‘labour costs will drop substantially’ Footnote 88 . In this way, ‘you can create jobs for large groups now unwanted by employers’ Footnote 89 . Although the argument is framed in terms of efficiency, this form of activation assumes retrenchment concerning both de-commodifying and redistributive policy.
To summarise, the existing welfare bureaucracy is a focal point of critique for both liberal and egalitarian proponents. Even though both see UBI as motivating paid employment, their reasoning is inversed, addressing either extrinsic or intrinsic motivations to work. Herein we see the core disagreement between liberal and egalitarian versions of UBI policy (e.g. De Wispelaere, Reference De Wispelaere2016; Chrisp and Martinelli, Reference Chrisp, Martinelli and Torry2019).
Automation and structural unemployment
The last controversy holds that structural unemployment is increasing due to the automation of labour, or that the number of available jobs is limited even in present-day society. This argument is fairly new to the welfare discourse and pops up in the UBI debate in multiple countries (e.g. Perkiö et al., Reference Perkiö, Rincon and Van Draanen2019). The prospect of automation prompts a reflection of the ‘productivist welfare post-war consensus around full employment’ (Noguera and Widerquist, Reference Noguera, Widerquist, Widerquist, Noguera, Vanderborght and De Wispelaere2013: 261). In particular, the prospect of automated labour and high structural unemployment seems to threaten the principle of work-based social security. In the automation narrative, the main justification for UBI is the lack of jobs, rather than the right to refuse work. In contrast to earlier lines of discussion, the automation narrative addresses a motive for welfare legitimacy that avoids the moral controversy surrounding the employment obligation – only the validity of the problem is challenged.
The automation narrative depicts a future where much work is done by robots: ‘robotization leads to the disappearance of jobs’ Footnote 90 , the ‘overcapacity of labour supply’ Footnote 91 , ‘increasing productivity’ Footnote 92 etcetera. In addition, concerns of current structural unemployment are mentioned (‘there aren’t enough jobs for everyone’ Footnote 93 , ‘there is no work now, let alone in the future’ Footnote 94 ). There is some evidence suggesting that unemployment reduces support for welfare conditionality (Buß et al., Reference Buß, Ebbinghaus, Naumann, van Oorschot, Roosma, Meuleman and Reeskens2017). We distinguish two underlying reasonings: a systemic critique on the principle of work-based social security and invoking the deservingness of the unemployed in a world without work.
First, the prospect of high unemployment challenges the system of work-based social security. In the face of automation, observers conclude that ‘full employment [is] unrealistic’ Footnote 95 or ‘an illusion’ Footnote 96 , and consequently we ‘need to search for a new economic model’ Footnote 97 . From this point of view, UBI is primarily seen as a solution to structural unemployment. The ‘disappearance of jobs [can be] cushioned by [UBI]’ Footnote 98 , or more generally, UBI is ‘the future system of social security’ Footnote 99 . The legitimacy of the welfare state rests on the availability of ample jobs, and the UBI discussion challenged the (future) tenability of that requirement. Since the necessity of work is usually beyond discussion, it is striking to see it so openly questioned.
Secondly, a lack of jobs justifies unemployment, because the work obligation loses its relevance. In the future ‘not everyone can work for money because the work is done by robots or jobs have been cut due to costs’ Footnote 100 . Simply put, ‘there is no other option if there are jobs available for only 20% of the people’ Footnote 101 . Consequently, the unemployed become deserving of social support: ‘millions of people will lose their jobs’ Footnote 102 , so ‘we cannot stimulate people to take jobs that don’t exist’ Footnote 103 . This relates to the perceived control over welfare dependency (e.g. van Oorschot et al., Reference van Oorschot, Roosma, Meuleman and Reeskens2017), and suggests that the importance of deservingness criteria varies over time (Buß et al., Reference Buß, Ebbinghaus, Naumann, van Oorschot, Roosma, Meuleman and Reeskens2017). On the opposing side of the argument, discussants simply reject the premise that future labour will be automated. Opponents here lament the lack of ‘factual support’ Footnote 104 , and note that ‘the end of work has been predicted for centuries’ Footnote 105 . Although underdeveloped, the argument suggests that new work will be created in the process of automation, simply shifting the type of available jobs.
In sum, automation and the threat of structural unemployment are recent catalysts of the UBI debate. Newspapers in multiple countries have used the threat of automation to make the case for UBI policy (Perkiö et al., Reference Perkiö, Rincon and Van Draanen2019). At first glance, this accords with studies showing higher deservingness in times of high unemployment (e.g. Buß et al., Reference Buß, Ebbinghaus, Naumann, van Oorschot, Roosma, Meuleman and Reeskens2017; Laenen and Meuleman, Reference Laenen, Meuleman, van Oorschot, Roosma, Meuleman and Reeskens2017). However, this apparent catalyst of debate has so far not proven to drive popular support for UBI policy. Existing work finds that only a very small proportion of the public is concerned that their jobs are at risk of being automated (Kurer and Häusermann, Reference Kurer and Häusermann2021). Moreover, even though those at risk are found to show a greater preference for passive income support, at-risk workers are not more likely to support UBI policy (Dermont and Weisstanner, Reference Dermont and Weisstanner2020). Thus, while relevant to the public debate and potentially influential on public opinion, the automation narrative seems to be especially driven by journalist elites.
Conclusion
This article set out to identify key controversies in the UBI debate through a content analysis of Dutch tweets. We identify four central controversies – different aspects along which the discussion takes place. Participants discuss UBI in terms of (a) economic redistribution, (b) welfare conditionality, (c) welfare state efficiency, and (d) structural unemployment. These controversies each cover multiple aspects of welfare legitimacy, particularly redistribution principles (Mau, Reference Mau2003; Jaeger, Reference Jaeger2006), welfare critiques (Roosma et al., Reference Roosma, van Oorschot and Gelissen2016), and recipient deservingness (van Oorschot et al., Reference van Oorschot, Roosma, Meuleman and Reeskens2017). In broad strokes, the observed discussion of UBI mainly regards economic redistribution and welfare conditionality. This is unsurprising since these are the major controversies of welfare politics: who pays and receives social support and what can be expected in return. The prominence of redistributive questions aligns with popular opinion across Europe, where UBI support is also chiefly motivated by redistributive concerns (e.g. Roosma and van Oorschot, Reference Roosma and van Oorschot2020). The notion of ‘welfare without work’ naturally sparks controversy, given the deeply embedded moral value of paid employment in society (Mau, Reference Mau2003). Thus, to a large extent, existing theories of welfare legitimacy seem to fit well into the UBI debate.
Still, we recognise elements that existing welfare research does not yet capture. First, critiques of targeted activation policy show that, even while supporting the (deservingness) principles behind targeted activation policy, discussants are highly sceptical of the efficiency and effectiveness of its current implementation. The ‘performance-critical’ view on activation policy (Roosma et al., Reference Roosma, van Oorschot and Gelissen2014, Reference Roosma, van Oorschot and Gelissen2016) seems crucial to understand the resonance of UBI policy. Second, the prospect of automated labour has the potential to challenge the legitimacy of work-based social security. Even though prior studies found no strong link between automation risk and support for UBI (Dermont and Weisstanner, Reference Dermont and Weisstanner2020; Kurer and Häusermann, Reference Kurer and Häusermann2021), the idea of a ‘world without work’ does form an integral part of the UBI discussion. More in-depth research into the public attitudes towards ‘post-productivism’ (e.g. Goodin, Reference Goodin2001; van der Veen and Groot, Reference van der Veen and Groot2006) and its relation to welfare legitimacy is therefore warranted. Finally, the analysis of a ‘live’ discussion also shows that the controversies of UBI support are not equally important to everyone: proponents are more focused on conditionality and efficiency, while opponents tend to emphasise redistributive aspects. The variety of motivations people have to support or oppose UBI point to a multidimensional policy, wherein not all policy aspects are equally supported (e.g. Dermont and Weisstanner, Reference Dermont and Weisstanner2020; Rincón García, Reference Rincón García2021).
Similar to Rossetti et al. (Reference Rossetti, Roosma, Laenen and Abts2020), we took a qualitative and exploratory approach in an attempt to complement the increasing number of studies with a more quantitative angle. We analysed the discussion taking place on mostly purposely selected days on a single social media platform. Although much of the discussion under scrutiny was triggered and influenced by documentaries, the discussion is not a mere rehashing of arguments. A comparison of arguments in Appendix B shows that while all arguments in the documentaries also appear on Twitter, some arguments – especially counterarguments – are only found on Twitter. Concrete retrenchment-related arguments in favour of UBI such as abolishing the minimum wage and introducing flat-rate taxes are also not cued by the documentaries. Thus, while the documentaries emphasise the ‘egalitarian’ version of UBI, Twitter also features proponents (and opponents) of a ‘liberal’ version (De Wispelaere and Stirton, Reference De Wispelaere and Stirton2004). Moreover, while the documentaries lean heavily on the automation narrative, the Twitter debate is much more focused on redistributive questions.
This study has several implications for existing research. First, support for UBI is a multidimensional construct, that should be measured by incorporating redistributive, conditional, and reform aspects. We find that these aspects consistently confound the discussion: while proponents focus on removing benefit obligations, opponents emphasise the distributional outcomes of UBI policy. Prior work has used a single indicator variable for UBI support in an attempt to explain its legitimacy (e.g. Roosma and van Oorschot, Reference Roosma and van Oorschot2020; Vlandas, Reference Vlandas2021). Stadelmann-Steffen and Dermont (Reference Stadelmann-Steffen and Dermont2020) have made some strides in this area with an innovative survey experiment, although they were unable to measure support for the unconditional aspect of UBI policy. Future studies should at least separate support for the redistributive aspect (the level of the benefit and manner of taxation) from support for its universal and unconditional character (e.g. no means test, no job application requirement) and the degree to which UBI replaces existing benefits.
Secondly, we find that critiques of activation policy motivate support for unconditional benefits on both sides of the political spectrum. Early welfare critiques are directed towards benefit generosity, arguing that welfare policies depress economic growth, undermine work ethic, and erode social solidarity (van Oorschot et al., Reference van Oorschot, Reeskens and Meuleman2012). Similarly, populist welfare critiques employ distrust towards corrupt and costly welfare institutions to undermine support for welfare policies (De Koster et al., Reference De Koster, Achterberg and van der Waal2013). In the UBI discussion, welfare critiques are prominently employed to argue against benefit obligations. Means-tested benefits are considered ineffective in part because they prevent work from being sufficiently rewarding, and activation policies are thought to discourage a cooperative and trusting relationship between welfare claimants and government officials. These critiques problematise the targeted activation policy paradigm, and in doing so may provide a basis of legitimacy for a ‘social investment’ approach to social assistance (Hemerijck, Reference Hemerijck2018), and the legitimacy of UBI itself.
Finally, the prevalence of the redistribution and conditionality controversies in part reflects a certain conceptual confusion, i.e. whether UBI redistributes income or not, and whether it encourages or discourages paid employment (De Wispelaere and Stirton, Reference De Wispelaere and Stirton2004). The co-existing conceptualisations of UBI policy have led some to argue that UBI is ‘beyond left and right’ (Reed and Lansley, Reference Reed and Lansley2016): its ‘unique potential (…) as the basis for a coalition of supporters from left and right’ (Murray, Reference Murray2008) yields ‘support across the political spectrum, from right and left, from pro-marketeers as well as social democratic interventionists’ (Reed and Lansley, Reference Reed and Lansley2016). This view has been challenged by others who find ‘persistent political division’ underneath the leftist and rightist approaches to UBI (De Wispelaere, Reference De Wispelaere2016), due to ‘intractable policy design trade-offs’ (Chrisp and Martinelli, Reference Chrisp, Martinelli and Torry2019: 477). Unfortunately, our focus has been on identifying the arguments in favour of and against UBI: so this has come at the cost of paying attention to the political actors behind arguments. One of the open questions in this area is therefore a thorough analysis of which partisans endorse which frames, to further develop the political feasibility of UBI (e.g. De Wispelaere and Noguera, Reference De Wispelaere, Noguera and Caputo2012).
Acknowledgements
This publication is part of the project ‘unravelling the basic income debate’ with project number 406.18.531 of the research talent programme which is financed by the Dutch Research Council (NWO). The project was subjected to ethical review by the Ethics Review Board at Tilburg University and approved under reference RP583. Special thanks go out to Johan Gielens for his early assistance in the data gathering process.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/S1474746422000422
Open access
