Applying Neo-Republican Theory to Welfare to Work Practices: Rules and Arbitrary Power in Mandatory Work Programmes in the Netherlands

Drawing on the neo-republican theory of non-domination and a qualitative case study conducted in three Dutch municipalities, this article explores the extent to which external rules are able to prevent arbitrary power in relationships between welfare officers and work supervisors, on the one hand, and welfare recipients participating in mandatory work programmes, on the other hand. It concludes that external rules were insufficiently implemented in the three municipalities in question. In addition, it found that rules cease to be capable of constraining arbitrary power where institutional contexts themselves are unpredictable and insecure. Under these conditions, welfare recipients may seek to avoid risks and act in accordance with the preferences (or their expectation of the preferences) of the welfare officer or work supervisor by playing the role of the ‘good recipient’ instead of relying on available rules of a protective nature or rules that enable them to have a say in their participation in mandatory work programmes.


Introduction
Many jurisdictions subject the right to social assistance or welfare benefits to rules that specify the behaviour desired of recipients. Although welfare officers usually have some discretionary freedom to decide whether a recipient is complying with sanction-backed obligations, discretionary spaces tend to expand and cease to be effectively subjected to supervisory jurisdiction (Watts andFitzpatrick, 2018: 28, referring to Donnison, 1977), particularly where recipients are subjected to conduct conditionality (such as a requirement to look for a job or participate in work programmes). Moreover, where welfare recipients participate in mandatory work programmes (MWPs), responsibility for supervising behavioural conditions is often (literally) removed from the welfare office. Under these conditions, welfare recipients may become subjected to the whims of the welfare officer or work supervisor as discretionary power evolves into sheer arbitrary power (Handler, 1979;Goodin, 1986;Dubois, 2010;Van Berkel, 2011;Molander et al., 2012;Eleveld, 2020).
Based on the neo-republican theory of non-domination developed by Philip Pettit and Frank Lovett (Pettit, 1997(Pettit, , 2012Lovett, 2010), this article investigates the extent to which effective external rules are able to prevent arbitrary power in the welfare to work (WTW) relationship between, on the one hand, welfare recipients who are obliged to participate in an MWP and, on the other hand, their welfare officers and work supervisors ('WTW relationships'). Drawing on empirical research conducted in three Dutch municipalities, it aims to contribute to the emerging scholarly literature that is using neo-republican theory of non-domination for a deeper understanding of power relations governing WTW relationships (Eleveld et al., 2020b).
In order to understand the neo-republican meaning of arbitrary power, we first need to have a look at the neo-republican conception of freedom. As a political philosophy, neo-republican theory seeks the 'promotion of freedom as non-domination on the basis of an equal concern for each citizen' (Pettit, 2012: 297). Pettit uses the example of the relationship between the slaveholder and the slave to illustrate the difference between the neo-republican conception of freedom and that of a liberal thinker, such as Isaiah Berlin (1969), for whom the only condition for freedom is that a person does not interfere, in practice, with the life of another person. This is contra-intuitive, according to Pettit, as this conception of freedom implies that the relationship between a benevolent slaveholder (who abstains from interfering in slaves' lives) and the slave does not constitute a relationship of domination. According to Pettit, what matters is that the slaveholder is still able to interfere in the slaves' lives at will and that slaves will therefore seek to avoid harm and adapt their behaviour to the master's expectations. As such, therefore, the slaveholder still controls the will of the slaves, without actually interfering in their lives. The neo-republican conception of freedom should also be distinguished from Hayek's liberal definition of freedom as the absence of coercion (Hayek, 1994), because this conception insufficiently addresses practices of domination between citizens. As Pettit (1997) argues, in Hayek's preferred weak state and a largely unregulated market economy the weakest and less respected members will not be free from domination (see further Spector, 2010;Lovett, 2013).
The neo-republican conception of arbitrary power is extremely useful for understanding the behaviour of welfare recipients who are subjected to domination (cf. Eleveld, 2020). Just like the slaves of the benevolent slaveholder, welfare recipients may be inclined to act in accordance with the preferences (or their expectation of the preferences) of the welfare officer or work supervisor in order not to lose the benefits of last resort, even when they have a legal right to act differently (cf. Lovett, 2010: 46;Pettit, 2012: 64-66). In other words, the freedom of these recipients is constrained even though actual interference is absent.
The neo-republican theory of non-domination distinguishes several ways in which the exercising of arbitrary power can be minimised in order to achieve neo-republican justice. For Pettit (1997Pettit ( , 2012, people should be able to control interference in their lives, given that the threat of domination arises only in the area of discretion enjoyed by the state. If applied to WTW relationships, this means that discretionary spaces of welfare officers and, by inference, a private employer acting as a welfare recipient's work supervisor should be subjected to democratic oversight. 'Workplace republicans' have also argued that workplace democracy and regulations that function as constraints on the scope of discretion afforded to managers or employers have the effect of minimising the exercising of arbitrary power on the work floor (Hsieh, 2008;Gonzales-Ricoy, 2014;Breen, 2019). In their analysis, therefore, arbitrary power can also be exercised (and controlled) outside the realm of the state. Lovett (2010), who, like the workplace republicans, applies the neo-republican theory to private relations, has systemised a non-normative republican concept of freedom as non-domination. While the first part of Lovett's book on justice and domination contains a descriptive conception of domination, the second part elaborates his (neo-republican) normative account of domination. Lovett's analysis is helpful for understanding how arbitrary power operates at a micro level of analysis. He argues that external effective rules, goals and procedures are essential for preventing the exercising of arbitrary power in dependent and unequal relationships, such as a relationship between a husband and wife or between an employer and an employee. Although Lovett does not apply the descriptive part of his theory to WTW relationships, many of these relationships can also be characterised as dependent and unequal. This is because, assuming that welfare recipients are not able to obtain work in the regular labour market, they are likely to face high 'costs' if they choose to exit the WTW relationship (Eleveld, 2020).
This article builds on Lovett's analysis of how rules operate in unequal and dependent relationships such as WTW relationships. However, his argument for implementing effective external rules is not restricted to the neo-republican theory of non-domination, given that a growing body of social policy literature is arguing in favour of implementing a system of constraining rules in the context of WTW relationships (Van Berkel, 2011;Ervik et al., 2015b;Eriksen, 2019). But while scholars in the street level bureaucracy literature tend to focus on how street-level bureaucrats use their discretionary spaces, the neorepublican theory of non-domination puts the emphasis on the domination of people, such as welfare recipients, at the receiving end. The research focus then shifts from policymakers or street level bureaucrats 'interpreting and making sense of the discretion that is granted to them' (Van Berkel, 2017: 25) to the ways in which rules prevent welfare recipients from being subjected to abuse and manipulation without being able to act on their own will.
This article starts by critically examining how rules are able to prevent the exercising of arbitrary power in WTW relationships. After explaining the context, data and methods, it turns to the research findings. The final section discusses the results of the research against the neo-republican theory of non-domination.

Rules a nd the prevention of a rbitrary power
The neo-republican theory of non-domination, particularly in Lovett's approach, provides a strong argument for the implementation of rules to prevent people from being subjected to arbitrary power in dependent and unequal relationships. This article defines rules as 'all written legal and non-legal rules, regulations and procedures that provide external, desireindependent reason for behaviour'. To a large extent this definition derives from that formulated by Lovett (2010: 108;2016: 36-41). This section critically considers how rules are able to prevent the exercising of arbitrary power in WTW relationships.
Professionals' use of discretionary freedom is an important topic in social policy and socio-legal literature. According to this body of literature, discretionary freedom is formally stipulated in legal regulations, such as 'competence rules' which govern the decisions and actions of policy officers. As a result, however, of ambiguity in rules and the absence of effective control of discretionary spaces, professionals additionally enjoy a space of uncontrolled discretion (see, for example, Evans, 2010;Jessen and Tufte, 2014;Van Berkel and Van der Aa, 2015;Van Berkel et al., 2017). While to some extent these discretionary spaces are needed to genuinely involve welfare recipients in decisions made by street-level workers (Ervik et al., 2015a), uncontrolled discretion has been found also to Applying Neo-Republican Theory to Welfare to Work Practices be at risk of evolving into arbitrary power, especially where recipients are subjected to conduct conditionality (Watts and Fitzpatrick, 2018). The socio-legal literature on discretion has proposed some solutions for preventing this risk (Goodin, 1986;Molander et al., 2012;Molander, 2016;). According to Goodin, the problem of 'manipulation and exploitation' inherent in discretion demands a set of additional rules and rights to protect welfare recipients from welfare officers 'abusing their discretion to control [recipients'] conduct' (Goodin, 1986: 240). From the viewpoint of neo-republican theory, we are particularly interested in Goodin's proposal for the implementation of rights constituting control from below and that allows welfare recipients to exercise control over welfare officers in the form, for example, of the right to receive social assistance benefits and the right to appeal decisions in an independent court (Goodin, 1986: 254). Molander (2016; see also Molander et al., 2012) offers an additional solution; in his view, one way to achieve control from below can be to implement 'epistemic measures', which aim to improve the reasoning process within the discretionary scope. An important example of such measures are participatory measures that introduce 'a demand for mutual justification into the interaction between co-deciders' so as to include all affected or concerned parties (such as the welfare officer and the recipient) as decision-makers (Molander et al., 2012: 226; see also Rosholm, 2014;Danneris and Caswell, 2019). If applied to WTW relationships, this would mean welfare recipients being genuinely involved in the decision-making on their participation in an MPW.
In summary, we can discern from the social policy and socio-legal literature that the following external rules prevent welfare recipients from being (in the neo-republican sense) subjected to arbitrary power: 1) competence rules, or rules governing the decisions or actions that a welfare officer or work supervisor may take and which as such constrain their discretionary freedom; 2) rights constituting control of discretionary spaces from below; and 3) rules requiring welfare officers to involve welfare recipients in decisions regarding their participation in, for example, MWPs. Particularly the second and third type of rules respond to Pettit's emphasis on the need for democratic oversight. The role of all three types of rules in preventing exercises of arbitrary power is the main topic of the empirical sections.
Another issue to be discussed regarding the role that these rules play in preventing arbitrary power is how they work in practice. According to Lovett's theory of domination, the implementation of certain rules suffices to prevent welfare recipients from being subjected to arbitrary power. It can be questioned, however, especially in the case of WTW relationships, whether people will always invoke such rules. As Goodin (1986) argues, 'The central plank in [the] argument [that legal rights prevent exploitation and manipulation in WTW relationships] is, obviously, an empirical one' (1986: 255, italics added). Indeed, numerous empirical legal studies have shown the interpretation and the mobilisation of specific rules and rights to depend on the availability of legal assistance and other social and cultural factors (Galanter, 1974;Nelken, 1981;Cotterell, 1992: 250-253;Ewick and Silbey, 1998).
In order to account for the way in which individual recipients' decisions to invoke rules or demand rights evolve against the background of limiting structuring conditions, it is worthwhile considering Goffman's work on social interaction and role-playing (see, for example, Goffman, 1990). The latter's work is particularly interesting in the context of this research as the notion that people may be inclined to act in accordance with the preferences (or their expectation of the preferences) of the other (Dubois, 2010;

Anja Eleveld
Mik-Meyer and Villadsen, 2013) aligns with the neo-republican insight. For Goffman, role-playing is a process in which individuals 'will try to decode "the others'" expectations and adjust their own actions' (Mik-Meyer and Villadsen, 2013: 32). On the one hand, roleplaying is determined by 'the situation', which comprises the institutional context, including formal regulations and their implementation (Goffman, 1974). This provides a solid framework for guiding how to behave in specific situations. On the other hand, Goffman (1970) recognises that people have the ability to act and play with the institutional context. That is, people exercise agency when their behaviour is not fully determined by the institutional setting, but where they act strategically or with the intention of achieving specific goals. The main relevance of Goffman's analysis for our study is that where institutions do not offer security, people will be less able or inclined to rely on formal rules and rights when acting strategically. Instead, they will be preoccupied by seeking to 'read' the other's expectations. Hence, Goffman shows that an analysis of the extent to which rules prevent welfare recipients from being subjected to arbitrary power requires us to pay particular attention to the interpretation of rules in different institutional settings, such asin the case of the Netherlandsin different municipalities. In addition, and based on Goffman's work, it can be hypothesised that, under institutional conditions of security and certainty, welfare recipients will be more inclined to rely on external rules that prevent them from being subjected to arbitrary power.

Context, data and met hods
The Participation Act [Participatiewet], enacted by the Netherlands in the 1990s, obliges all welfare recipients, except those who are fully incapacitated for work, to seek to re-integrate into regular paid employment. However, municipalities are free to choose which re-integration instruments to deploy, including the possibility of mandatory participation in a work programme. This legislation also stipulates rights for welfare recipients, with the most important rights in this context being the right to social assistance benefits, and the right to municipal help in transitioning from welfare to work. In addition, the Dutch Administrative Act provides for the right to lodge an appeal against a municipal decision and to lodge a complaint against, for example, maltreatment by municipal officers.
This article draws on empirical research conducted between 2017 and 2018 in three Dutch municipalities, each of which imposed an obligation to participate (under municipal regulations) in an MWP for up to twelve months. One of these municipalities was one of the four largest cities in the Netherlands, with over 300,000 inhabitants ('municipality A'), while the other two were medium-sized cities with between 50,000 and 100,000 inhabitants. ('municipality B and C'). All three municipalities offered very basic training programmes (such as courses on applying for jobs) and could require welfare recipients to participate in MWPs; these programmes generally involved unskilled manual work, such as gardening, serving coffee in a nursing home, simple production work or delivering mail.
Since municipalities enjoy some discretion in implementing the Participation Act, this research enables us to consider the interpretation of rules in different institutional contexts. The main institutional differences include the policies on sanctions and the policies regarding assignment to MWPs. To start with policies on sanctions, it should be noted that the Participation Act does not give municipalities the freedom to determine the level of sanctions imposed on individuals who fail to comply with work-related obligations.
Applying Neo-Republican Theory to Welfare to Work Practices However, the ways municipalities apply the national sanctions regulations in practice were found to differ substantially (Inspection Service SZW, 2017; SCP, 2019). As Table 1 shows, municipality A (16.2 per cent) was clearly stricter than municipality B (11.2 per cent) and municipality C (10.4 per cent) in terms of the number of sanctions (mostly pertaining to work-related obligations) imposed in 2015 and 2016 as a percentage of the number of recipients. In addition, the average monetary sanction imposed in municipality A amounted to 51.3 per cent of monthly benefits, whereas in municipality B this was only 35.9 per cent. Secondly, the municipalities differed with respect to their referral policies. Municipalities A and C allowed some leeway in referring participants to an MWP. Welfare officers were therefore able to decide whether participation in an MWP was necessary for enhancing a recipient's chances of re-integrating into regular work. In municipality B, by contrast, all able-bodied welfare recipients were in principle referred to an MWP after eight weeks.
While these differences reflect municipal discretion, this research focuses on how this discretion is applied in WTW relationships and views municipal discretion as one of the factors constituting the institutional context, and which may affect the effectivity of rules.
The research data consist of internal and external municipal documents on municipal rules and policies; semi-structured interviews with twenty-one municipal policymakers and staff, thirty-one welfare officers (including work coaches), thirty work supervisors, seven members of client councils and forty-seven welfare recipients (mostly MWP participants), and forty-five observations of interactions between recipients and welfare officers (see Table 2). The interviews lasted an average of one hour, while the observations lasted between fifteen minutes and three hours. In each municipality, the researcher investigated five to seven MWPs. Most welfare recipients were interviewed while on the

Anja Eleveld
MWP. Most of the work supervisors of these MWPs had asked the participants whether they would cooperate with the research. The empirical data were analysed using methods developed by Ritchie and Spencer (1994) (see also Ritchie et al., 2003). Transcripts of the interviews and observation notes were coded in Atlas TI. After the initial codingbased on the interview topics such as 'referral to an MWP', 'learning goals', 'evaluations', 'experience of working on an MWP' and 'lodging appeals and complaints', or based on descriptive codes (in the case of observation notes) such as 'relaxed atmosphere' or 'tense atmosphere'new (descriptive) codes such as 'feeling maltreated' and 'anxiety' and (analytic) categories such as 'feelings resulting from being subjected to domination' were then determined, all of which could be linked to the different municipalities in specific ways.

Rules c onstraining discret ionary freedom
This section investigates the competence rules governing the assignment of welfare recipients to MWPs and the supervision of welfare recipients in MWPs (i.e. the first type of rules that prevent welfare recipients from being subjected to arbitrary power).
With regard to the referral to an MWP, the empirical findings revealed that none of the municipal (or internal) regulations provided instructions on the conditions for referring recipients to an MWP or on whether, and to what extent, these MWPs needed to be tailored to the recipient's work experience, education or preferences. This implies that welfare officers enjoy considerable discretionary freedom to decide whether a recipient should be referred to an MWP and, then, to which particular MWP. In addition, none of the municipalities was found to have implemented mechanisms to monitor these decisions. Some welfare officers even admitted that they had occasionally assigned welfare recipients to particular MWPs specifically in order to punish them (cf. Hatton, 2018).
Regarding participation in MWPs, employment law is an important source of rules constraining work supervisors' discretionary spaces. Aside, however, from some basic national rules (such as regulations on working times and health and safety) (Eleveld et al., 2020a), Dutch law does not explicitly stipulate employment rights for welfare recipients participating in MWPs, and the research found that municipal regulations did not fill this 'gap' left in national regulations. Nevertheless, all three municipalities had stipulated a few alternative measures that could potentially impose constraints on the discretionary powers of welfare officers and work supervisors. These included the requirement for written contracts (which also specified a few employee rights), the requirement for learning goals to be included in MWPs and the requirement to systematically and regularly evaluate recipients' progress on the MWPs. However, many of the welfare officers and work supervisors interviewed were not aware of the contents of the written contracts, and not all welfare officers formulated learning goals or evaluated welfare recipients' participation in MWPs. Moreover, learning goals were sometimes thwarted by the need to 'keep production running': in, for example, all the municipalities we observed companies putting pressure on recipients to work harder in order to increase turnover or to have external orders ready in time, even where the programme's only goal was to reintegrate people 'at a disadvantage in the labour market' into the regular world of work. Welfare recipients also participated in MWPs where they had to work hard during the peak season (such as when doing maintenance work in municipal gardens) or at peak Applying Neo-Republican Theory to Welfare to Work Practices hours (such as in a high school canteen) and were sometimes responsible for supervising other recipients and trainees and occasionally even regular paid workers.

Cont rol from below
This section investigates the extent to which rules and rights enabled welfare recipients to control welfare officers' (and work supervisors') discretionary freedom, including rules requiring welfare officers to involve welfare recipients in decisions, firstly regarding the referral of welfare recipients to MWPs and, then, regarding the supervision of their participation in MWPs (i.e. the second and the third types of rules that prevent welfare recipients from being subjected to arbitrary power).
At the outset, none of the municipalities had implemented participatory measures that granted welfare recipients a formal say in their referral to an MWP. Moreover, even though all three municipalities had implemented instruments, such as training programmes enabling recipients to discover 'who they are' (Arts, 2020a) and what they wanted to learn (i.e. formulating 'learning goals'), these instruments were in practice rarely used as the basis for mutually exploring recipients' needs and preferences in depth in order to select the most suitable MWPs for them. In addition to these similarities in formal regulations, some important differences were observed in practice with respect to recipients' participation in the decision-making processes. Whereas, for example, welfare officers in municipalities A and B usually held short discussions (of less than fifteen minutes), welfare officers in municipality C spent more time on these discussions and allowed more scope for discussing recipients' views than welfare officers in municipalities A and B. In addition, assignment discussions and counselling in municipality C took place in a relatively relaxed atmosphere, and recipients did not necessarily view being referred to an MWP as an obligation being imposed on them. Instead, they often regarded it as both a steppingstone to paid work and an opportunity for self-development. In municipalities A and B, by contrast, the assignment discussions were much more tense. Recipients felt pressured, and their main motive for participating in an MWP was often simply to avoid a financial sanction. In municipality B in particular, the obligation to participate in an MWP fuelled feelings of insecurity because it was imposed on all recipients expected to be able to transition to regular paid work within two years. Some recipients also felt they were subjected to the whims of the welfare officer, who left them no choice but to participate in an MWP that they saw as being insufficiently tailored to their individual circumstances. As Carla, a welfare recipient in municipality B, said: I had to work in the second-hand shop and I've been crying ever since : : : and then I feel like I am not being heard ( : : : .) that they can stop your benefits, that's something you have to deal with. I feel like a puppet; they're playing with you ( : : : ) they decide for you.
(Carla, municipality B) These municipalities' referral policies also sparked feelings of insecurity on a more fundamental level, with some recipients doubting whether they could really rely on the right to sufficient means of subsistence. As, for example, Irma, a highly educated and law-abiding MWP participant, said:

Anja Eleveld
I've been afraid they're going to cut my benefits as every letter they sent me said so ( : : : ) in these situations I tell myself I don't have to be scared my benefits will be cut or withdrawn. If that happens, I'll be living on the street with a bottle and a paper bag. (Irma, municipality B) On the one hand, Irma told herself that, rationally, she had to trust the law, which stipulates a right to welfare benefits, providing she complied with the work-related obligations. On the other hand, however, she feared ending up homeless and living on the streets, despite complying with all the work-related obligations. Like many other recipients in municipalities A and B, Irma felt she had no rights, only obligations. We rarely found this view expressed by the MWP participants interviewed in municipality C. Regarding participation in MWPs, the requirement for written contracts and learning goals enabled MWP participantsat least in theoryto influence the terms and nature of their work on the programme and the supervision they required. In practice, however, these effects were found to be very limited. This was firstly because the municipality drew up the contracts unilaterally (cf. Freedland and King, 2003) and secondly because welfare officers used the same unilaterally determined and very basic learning goals for each recipient, such as 'learning basic worker skills', irrespective of the person's years of work experience or educational level. In addition, all municipalities had implemented the requirement to evaluate the progress made at MWP. Theoretically, this requirement provided recipients with the opportunity to: 1) inform their work coach and supervisor about their experience on the MWP; 2) share their views on how these activities contributed to their personal development; and 3) possibly propose changes in their supervision or even to request an alternative MWP. However, like contracts and learning goals, the requirement to evaluate participation in MWPs was not necessarily interpreted to mean measures enabling participants to have a say in WTW activities. This was true especially for MWP participants in municipalities A and B. This can be illustrated by the way recipients in municipalities A and B responded to the lack of clear medical guidance on sick or incapacitated MWP participants. In municipality B, where half of the MWP participants interviewed were partially physically incapacitated for work, many of them felt that the work coaches did not take sufficient account of their medical conditions. As Lena, who for medical reasons needed to work sitting down, said: I don't think that my work coach knows that I'm not doing work sitting down, which is what I was supposed to do. Instead, I'm lugging crates around all day long.
(Lena, municipality B) Yet instead of asking for more guidance and support from the work coach and work supervisor, Lena mainly blamed herself for not being able to set boundaries. It did not occur to her that she could ask for more guidance in this respect by, for example, bringing this point up in evaluations. Likewise, Natascha, who participated in an MWP in municipality A, with its strict sanctions, did not raise the issue of her back problems during the evaluation discussions, even though the pain forced her to lie down for hours after returning home from the MWP. In addition, and while the under-use of learning goals seemed to be largely attributable to the fact that most of these goals entailed 'learning basic skills', the findings in Applying Neo-Republican Theory to Welfare to Work Practices municipality A also point to another reason for the failure to ask for guidance or support: contracts, learning goals and evaluations were frequently regarded as instruments of control in the hands of 'agents of domination'. In other words, participants in MWPs in municipality A often considered 'the contract' as proof of their having fulfilled their obligation, while 'learning goals' were seen as instruments for welfare officers to assess whether the recipients were fulfilling their obligations, and 'evaluations' as opportunities for welfare officers and supervisors to tell them what they had done wrong. Indeed, some MWP participants in municipality A believed that they were being closely watched during their work, that supervisors would be 'snitching on them' and that they would be rewarded with more fulfilling work if they demonstrated that they were working hard. As Sandra, an MWP participant in this municipality, remarked: I think the work supervisors will tell the welfare office if you're not doing your best. I guess this is one of the reasons why the supervisors pay close attention. They know right away who's doing her best and who isn't and [if you do your best] they'll let you do more interesting work, like sorting out clothes for a care home.
(Sandra, municipality A) Moreover, because they regarded the requirement to use contracts, learning goals and evaluations as measures of control in the hands of 'agents of domination', recipients in municipality A tended to adapt their behaviour to fit their expectations of the welfare officer's behaviour instead of participating as co-decision-makers. This is further illustrated by the story of fifty-three-year-old Maria in municipality A, who had participated in various MWPs over a period of several years and who had recently been told by the work coach that she was no longer physically able to work. Although Maria did not understand why she then still had to attend the MWP to perform low-intensity production work, she did her best ('as the work supervisor was watching'), without asking any questions. When asked whether she would discuss the possibility of stopping her work on the MWP with her work coach, she became upset and said: I'm not going to say that I don't want to work there. Because the welfare officer will say 'Why don't you want to come here?' You're not supposed to say that. (Maria, municipality A) During the research Maria was invited to a meeting at the welfare office to share details of her experience with fellow welfare recipients. At that time, she was spending thirty-two hours a week on an MWP and the meeting was planned on her only day off in the week. Although her welfare officer explained that she was not obliged to come to the meeting (and so non-attendance would not result in a sanction), she did not dare to risk rejecting the invitation. In summary, for Maria, measures such as contracts, evaluations and learning goals did not offer any security. She was also uncertain about the welfare officer's expectations. Playing the role of a 'good recipient' was thus her way of coping with this difficult situation. The interview excerpts with Irma and Maria exemplify a significant number of the participants interviewed in municipalities B and A, respectively. These participants did not trust the rules or measures that enabled them to participate in decision-making processes. The only rules they trusted were rules that stipulated a sanction-backed obligation.

Anja Eleveld
This focus on obligations completely overshadowed recipients' right to social assistance benefits and to receive support in re-integrating into regular work.
In addition, like many of the other MWP participants interviewed in municipality A, Maria would not consider lodging an appeal or complaint against the obligation to continue working on the MWP as she did not want to disrupt her relationship with the work officer. Other MWP participants in this municipality also chose not to appeal, not only because this would disrupt their relationship with their welfare officer, but also because they felt that, given the huge power differences, there would not be much point in doing so.
More in general, appealing against a referral to an MWP was not easy because municipalities often failed to confirm participation in an MWP in writing, let alone send a separate letter explaining the opportunity to appeal against being assigned to a specific MWP. In practice, therefore, the only legal remedy for recipients wanting to refuse referral to an MWP was to appeal against a sanction imposed for not participating in the MWP to which they had been assigned. These appeals forced welfare officers to document recipients' non-compliant behaviour very precisely. This applied particularly in municipality A, which implemented a strict sanctioning regime. These practices also had contrary effects as they incentivised work officers to intensify the control on welfare recipients.
It also turned out to be very difficult to lodge an appeal or complaint against a work supervisor acting on behalf of a private employer, given that public law provisions only regulate appeal and complaint procedures against public officers. According to welfare officers, this was not a problem as welfare recipients could report their welfare officers in the event, for example, of maltreatment. However, this was not a satisfactory solution for MWP participants, especially for those who refrained from lodging an appeal or complaint because they were afraid of repercussions (from welfare officers). Sometimes recipients were more willing to lodge a complaint anonymously. In municipality B, for example, the client council produced a 'black book' containing dozens of anonymous complaints about how recipients were treated by welfare officers assigning them to MWPs. During that same year the official complaints committee had received only a handful of (non-anonymous) complaints, all of which were rejected.

Discussion and conclusion
Neo-republican theory provides a strong moral argument for minimising arbitrary power in dependent and unequal relationships, such as WTW relationships, by means of externally effective rules. Our investigation of the extent to which rules constraining discretionary freedom of welfare officers and work supervisors and rules requiring welfare officers to involve welfare recipients in decisions regarding their participation in MWPs reduce arbitrary power in WTW contexts found that these kinds of rules tended to be scarce and insufficiently implemented, and sometimes even had contrary effects because they intensified the control on welfare recipients. As such, they sometimes rather seemed to represent the 'myth and ceremony' of the bureaucratic model (Handler and Hasenfeld, 2005).
The research also found that welfare recipients interpreted similar rules differently. While recipients in municipalities A and B often complained about having only obligations and no rights, these criticisms were rarely heard in municipality C. For reasons relating to these different interpretations, welfare recipients' motives for participating in Applying Neo-Republican Theory to Welfare to Work Practices the MWPs were also strikingly different. In other words, whereas recipients in municipalities A and B participated in MWPs mainly to avoid a sanction, recipients in municipality C were more likely to believe that participating in an MWP would help them to develop and transition from welfare to work.
These findings seem to confirm the hypothesis, based on Goffman's work on social interaction and role-playing, that, in institutional conditions of security and certainty, welfare recipients will be more inclined to rely on external rules that prevent them from being subjected to arbitrary power. It could be argued that the more relaxed institutional context prevailing in municipality C provided a comparatively solid background for strategic actions and that recipients in this municipality either 'played their role according to the rules of the game' or sought to make positive use of the rules to their own advantage (such as using learning goals as a means of self-development). Under the strict sanctioning regime applying in municipality A and the referral policies in municipality B, by contrast, recipients felt less confident and, instead of using the rules positively to their own advantage, they mainly sought to avoid risks and (particularly in municipality A) played the role of the 'good recipient'. Indeed, under the institutional conditions applying in municipalities A and B, many recipients did not seem confident that they could rely on formal protective rules, such as the right to appeal against sanctions. In addition, under the harsh sanctioning regime in municipality A, instead of the requirement for contracts, learning goals and evaluations being used to influence and control WTW practices, these measures tended to be seen as instruments of control in the hands of 'agents of domination'.
The findings thus imply that, even under the rule of law, welfare recipients are at risk of being subjected to arbitrary power. As a result, and especially where the institutional context itself becomes a source of domination, welfare recipients may seek to avoid risks and play the role of the 'good recipient' instead of relying on available (or potentially available) rules of a protective nature. This behaviour affects the effectivity of rules to prevent arbitrary power. It should be noted, though, that the interpretation of rules is also affected by the way they are implemented. In municipality A, for example, rules requiring welfare officers to involve welfare recipients in decisions regarding their participation in MWPs also seemed to have been implementedat least to some extentas instruments of control in the hands of 'agents of domination'. In order, therefore, to understand these measures' lack of effectivity in WTW relationships, the triangular relationship between rules, implementation and interpretation needs to be investigated in more depth through ethnographic research at WTW offices (see, for example, Arts 2020a; 2020b).
The findings of this research support studies that demonstrate that hard neo-liberal and paternalist WTW sanctioning systems crowd out intrinsic motivation (Soss et al., 2011;Watts and Fitzpatrick, 2018). The findings additionally suggest that, in solid and predictable WTW relationships, recipients will be more inclined to rely on rights, such as the right to support for re-integrating into paid employment, and on external rules, such as those requiring welfare officers to formulate learning goals and to evaluate participation in MWPs. These latter rules further minimise arbitrary power as they simultaneously constrain agents of domination's discretion and genuinely involve recipients in decision-making procedures. These policies may also be capable of reconciling enabling and disciplinary characteristics of WTW policies (cf. Caswell et al., 2017) as they seem to encourage welfare recipients to be 'positive entrepreneurs', using the opportunities offered to develop themselves and to make meaningful transitions to the labour market, Anja Eleveld rather than being harm-avoiding 'negative entrepreneurs' (Whitworth, 2016) who 'self-govern and critique their own behaviour' (Patrick, 2017: 301). As such, WTW policies inspired by neo-republican theory may be desirable not only for moral reasons, but also because they are more efficient than, for example, hard neo-liberal and paternalist policies. Nevertheless, more researchspecifically focusing on the triangular relationship between rules, implementation and interpretationis needed in order to establish the conditions in which rules: 1) encourage self-development and WTW transitions of MWP participants and, conversely, 2) discourage welfare officers from assigning specific MWPs to welfare recipients where this cannot be justified.