Intensity, moderation, and the pressures of expectation: Calculation and coercion in the street-level practice of welfare conditionality

This article offers a street-level perspective on welfare conditionality as it was practiced in contracted-out UK activation programs between 2008 and 2015. Drawing on observation and in-depth interviews, the article illustrates the ways that behavioral conditionality provided street-level workers with the means to intensify or moderate activation for particular claimants. Responding to arguments about the curtailment of street-level discretion, the article argues that in the particular context of target-driven, work-first, and otherwise highly constrained services, discretion resided in the ability to intensify or moderate conditionality and its coercive potential — in decisions about how, on whom, and to what extent it would be applied. The article argues that attending to this form of discretion provides an alternative frame through which to view the differentiated treatment typically understood as “ creaming ” and “ parking. ” In so doing, the article problematizes accounts that draw clear lines between calculative, normative, and dispositional forms of street-level reason and practice. It shows how advisors' responses to the “ street-level calculus of choice ” were articulated in terms of expectation , where attempts at future-oriented calculation necessarily entailed making other forms of speculative and normative judgement about claimants and their situations. The article thus contributes to an understanding of both the causes and meaning of differentiated treatment in conditional welfare services.

the ways that behavioral conditionality provided streetlevel workers with the means to intensify or moderate activation for particular claimants. Responding to arguments about the curtailment of street-level discretion, the article argues that in the particular context of target-driven, workfirst, and otherwise highly constrained services, discretion resided in the ability to intensify or moderate conditionality and its coercive potential-in decisions about how, on whom, and to what extent it would be applied. The article argues that attending to this form of discretion provides an alternative frame through which to view the differentiated treatment typically understood as "creaming" and "parking." In so doing, the article problematizes accounts that draw clear lines between calculative, normative, and dispositional forms of street-level reason and practice. It shows how advisors' responses to the "street-level calculus of choice" were articulated in terms of expectation, where attempts at future-oriented calculation necessarily entailed making other forms of speculative and normative judgement about claimants and their situations. The article thus contributes
In the punitive turn of "the global workfare project" (Brodkin, 2013c), conditionality is but one spoke, alongside individualized activation (van Berkel & Valkenburg, 2007) and the extension of marketized governance (Brodkin & Larsen, 2013;Lødemel & Moreira, 2014). Street-level organizations, defined as those sites where the public meet and interact with low-level employees of state or state-funded services (Lipsky, 2010), are particularly important as locations where these different policies are enacted and produce their effects (Brodkin & Marston, 2013;Lødemel & Moreira, 2014). To date, street-level welfare-to-work research has tended to focus on individualized activation and the effects of new public management and marketized governance (Brodkin, 2011;Fuertes & Lindsay, 2016;Greer, Schulte, & Symon, 2018;Soss, Fording, & Schram, 2013). New public management and marketization are widely argued to have severely curtailed street-level discretion: competitive payment-by-results contracting arrangements and individual performance targets have enforced narrowly calculative forms of street-level reason and decision making (Brodkin, 2011;Fuertes & Lindsay, 2016). Despite the overtly moralistic tone of political discourse surrounding behavioral welfare reform (Garthwaite, 2011;Jensen & Tyler, 2015;Patrick, 2016;Wiggan, 2012), the meaning and import of behavioral conditionality for street-level practice have received less attention. Given that the individualized, moralized, and coercive character of welfare conditionality is often central to citizens' experiences of welfare reform (Flint, 2019;Friedli & Stearn, 2015;Garthwaite, 2014;Patrick, 2017;Reeve, 2017;Wright, 2016), it is important to understand the role of street-level practitioners in producing these relations and experiences.
Drawing on observation and in-depth qualitative interviews with street-level workers, this article explores the street-level practice of behavioral conditionality in contracted-out UK welfare-to-work programmes between 2008 and 2015. Although the research looked at the experiences of both street-level workers and benefit claimants, this article is more narrowly focused on the experiences and practices of street-level workers. This focus is justified by the difficulty of accessing such workers and the relative absence of their perspective from accounts of welfare conditionality. As this article will demonstrate, the role of street-level workers in the enactment of conditionality is complex and multifaceted and deserves sustained attention.
The article begins by outlining the related policies of conditionality and activation, situating these in the context of governance reform, contracting out, and quasi-marketization. The article summarizes the well-established argument that these developments have severely constrained street-level discretion. Findings are then presented in three sections dealing, in turn, with caseload management, sanctions, and expectation setting. On the basis of these findings, it is argued that tight managerial control and a narrow work-first emphasis notwithstanding, behavioral conditionality provided street-level advisors with a means to either intensify or moderate the level of activation applied to claimants. Rather than strictly curtailed, increased conditionality has shifted the locus of street-level discretion and altered its form. This has consequences for understanding both the causes and meaning of differentiated treatment in such services. Brodkin (2013aBrodkin ( , 2013b has argued that what she calls "the global workfare project" is orchestrated through the combination of formal policies (such as conditionality and activation) with governance and managerial reform encompassing the contracting out of services, quasi-marketization, and new public management. Following Brodkin, this article is interested in the confluence of these developments at street-level and in situating the practice of conditionality within this broader context. These different policies and reforms have their own distinct definitions and meanings, however, and are worth briefly outlining separately.

| CONDITIONALITY, ACTIVATION, AND MARKETIZATION
Conditionality, in the broadest sense, simply refers to the fact that eligibility and entitlement to social security and/or assistance is not automatic or universal but conditional upon a variety of factors such as personal circumstance, group membership, and behavioral compliance (Clasen & Clegg, 2007). Behavioral conditionality, sometimes simply referred to as "welfare conditionality" (Watts & Fitzpatrick, 2018), applies once other criteria have been met and has "the function of regulating ongoing benefit receipt" (Clasen & Clegg, 2007, p. 174).
Conditions of conduct are not new (Jones & Novak, 1999), but alongside a cumulative tightening of traditional eligibility and entitlement criteria recent decades have witnessed the proliferation of new forms conduct monitoring and surveillance in the regulation of ongoing benefit receipt (Dwyer, 2004;Dwyer & Wright, 2014;Fletcher & Wright, 2018).
Activation refers to the linking of behavioral conditionality to active labour market policy, otherwise known as "welfare-to-work" (Lødemel & Trickey, 2001;Sage, 2013). A "particularly ambiguous policy category" (Bonoli, 2010, p. 439), active labour market policy, encompasses a range of different policy measures through which the state takes an "active" rather than "passive" role with respect to labour supply. In the form of "welfare-to-work," activation in the UK has focused on incentive reinforcement-through conditionality and sanctions-with a short-term work-first emphasis (Lindsay, 2007), premised on ideological tropes about inactive and demoralized welfare subjects (Wright, 2012).
Through a series of flagship programmes and contracts-from New Deal to the Work Programme the development of activation in the UK has been associated with an increasing role for voluntary and private sector providers.
The advance of activation and conditionality has thus been achieved through a variety of developments in operational policy and governance reform. There has been a marked shift first toward managerialism in the running of public services, then to contracting out, and finally toward the construction of large-scale quasi-markets in welfare-to-work services (Wright, 2011). A key feature of this quasi-marketization is payment-by-results arrangements that reward providers for successful "job starts," and the proportion of funding provided in this way has increased over time. The street-level context for increasing conditionality has thus been marked by the intensification of competitive and calculative market rationalities. As the next section will explore, this-rather than the punitive and moralistic turn embodied by conditionality-has been the dominant frame though which street-level practice has been interpreted and analysed.

| DIMINISHING DISCRETION AND STREET-LEVEL AGENCY
Street-level organizations are held to be of consequence to the extent that they mediate a variety of frequently conflicting and contradictory policy aims and concretize the ambiguities of formal policy in forms of everyday practice (Lipsky, 2010). Typically, the freedom of street-level workers to exercise discretion in the course of their work is taken to be a crucial factor in how policies are implemented in practice. Originally conceptualized in terms of KAUFMAN 207 the coping mechanisms of street-level workers, as the street-level literature has developed, there has been move toward thinking more broadly in terms of street-level dispositions and situated agency (Dubois, 2012;Maynard-Moody & Musheno, 2003;Watkins-Hayes, 2009;Zacka, 2017). This conceptual development has reflected a need to understand the different role orientations (Maynard-Moody & Musheno, 2003), identities (Watkins-Hayes, 2009), and dispositions (Dubois, 2012;Zacka, 2017) that street-level workers bring to and develop through the course of their work. It has been recognized that not all street-level workers react to the demands and pressures of their role in the same way, and there is often a complex interplay between structural constraint, situational context, and individual agency. In different ways, these recent theorizations of street-level work have advanced relational accounts of discretion as a form of agency, foregrounding interactions between identities, dispositions, and broader organizational and social contexts. Although a description of the particular dispositions found among street-level workers is not the focus of this article, its argument is nevertheless grounded in these conceptualizations, and in the idea that workers react in different ways to the various pressures and incentives they face. It is also grounded in a conceptualization of agency as a set of interrelations and dependencies inseparable from its social and structural context-as something that is produced and shaped through processes of subjectification. In this account, agency does not stand in opposition to situational or structural constraint; both are enmeshed and mutually constitutive (Bourdieu & Wacquant, 1992;Dubois, 2012).
These conceptual developments notwithstanding, it is widely argued that street-level discretion in welfare-towork services has been significantly constrained and eroded, and this is largely attributed to managerial and governance reform. This argument has been made in different national contexts including the United States (Brodkin, 2011;Soss et al., 2013), Australia (McDonald & Marston, 2006), Germany (Greer et al., 2018), and the United Kingdom (Fuertes & Lindsay, 2016;Greer et al., 2018). What each of these different contexts share is a trend toward contracting out, marketization, and the intensified performance management of street-level workers, tied to the pursuit of short-term job outcomes in the service of payment-by-results contracting arrangements. These reforms have been observed to result in a tightly controlled labour process focused on hitting targets and characterized by job insecurity and punitive firing (Greer et al., 2018). In this way, the disciplinary turn of the global workfare project has been achieved, in part, through the disciplining of street-level workers (Soss et al., 2013). In this context, it is widely argued that street-level workers have minimal agency to shape the nature of the service they provide, either in terms of offering support addressed to the needs and circumstances of individuals (Brodkin, 2013a;Fuertes & Lindsay, 2016;McDonald & Marston, 2006;Soss et al., 2013) or indeed in terms of completing the administrative processes required to enact sanctions (Greer et al., 2018(Greer et al., , p. 1439. What discretion remains tend to be understoodimplicitly or otherwise-in terms of a rational and calculative response to street-level incentive structures, with little scope for advisors to exercise more substantive forms of agency or judgement in line with their identities, role conceptions, or dispositions. Examples where street-level workers attempt to exercise non-calculative forms of agency are taken to be "the exception that proves the rule" (Brodkin, 2011, p. i260).
The differentiated treatment of street-level clients, interpreted as "creaming" and "parking," is often given as an illustrative example of both diminished discretion and the prevalence of calculative reason in street-level activation services (Carter & Whitworth, 2015;Fuertes & Lindsay, 2016;Greer et al., 2018;Rees, Whitworth, & Carter, 2014;Wiggan, 2015). The need to chase outcome targets exacerbates inherent pressures to focus resources on those service users who are most likely to produce short-term job outcomes ("creaming") at the expense of people with more difficulties finding paid work, who are neglected ("parking"). This dynamic is widely associated with a loss of discretion and either assumes or is explicitly attributed to the rational choices of street-level workers in response to the situational incentives-or "street-level calculus" (Brodkin, 2011)-produced by narrowly defined and highly consequential outcome targets. Attempts to design this dynamic out of the system by altering the contractual incentives-through differential pricing of different claimant groups-are argued to have exacerbated the problem by simply reproducing it within each claimant category (Carter & Whitworth, 2015;Rees et al., 2014).
The narrative about the loss of street-level discretion exhibits a tendency to position the value judgements and dispositions of street-level workers in opposition to the structural constraints or "street-level calculus" that they face.
The potential for street-level agency is presented as having been overwhelmed by an enforced calculative logic dictated by managerial oversight and situational incentives. In this respect, recent street-level analysis stands in marked contrast to the rich history of work in economic anthropology and sociology, which have long foregrounded the moral significance of calculative economic practices, and the close relationship between calculation and judgement (Beckert, 2013;Boltanski & Thévenot, 2006;Graeber, 2014). This article similarly seeks to problematize the opposition of dispositional judgement and rational calculation and attends to the ways that agency is shaped by the structural constraints of the street-level situation. Like recent work by Hansen (2017), it foregrounds some of the moral and justificatory dimensions of street-level practice but with a particular focus on how these are articulated in the context of competitive market mechanisms. Although it has been argued that governance reforms have "changed where discretion is located and how it is exercised" (Brodkin, 2013c, p. 277), the new locations of discretion have received little empirical attention. This article thus focuses on the street-level practice of conditionality and asks whether discretion has been diminished or, instead, displaced; and if the latter, to where, and in what form?

| METHODS
The findings presented in this article are based on Economic and Social Research Council-funded doctoral research, the aim of which was to explore the everyday dynamics, practices, and experiences of "activation" and "behavioral conditionality" from a street-level perspective-that is, from the perspective of street-level workers and service users. The research comprised of 30 in-depth interviews with benefit claimants (n16), street-level employment advisors in contracted-out welfare-to-work providers (n13), and one (n1)  My own prior experience as a street-level worker-for a prime welfare-to-work provider, for 9 months during 2008-2009-played an important role in this research, not least in terms of facilitating access to advisors but also in terms of an embodied and subjective understanding of the context in which they worked, with its particular pressures and dilemmas. It might be argued that such prior involvement in the welfare-to-work field requires a certain form of reflexive writing. However, rather than offer an autoethnographic account, this article is instead more conventionally styled and structured around quotations from coded interviews conducted in 2014-2015, some years after my employment in the field. This reflects a desire, in this article, to focus on how the experiences of different advisors speak to dominant themes and arguments from the street-level welfare-to-work literature, rather than center my own experience or questions of methodology and knowledge production as such. These, and other issues, deserve fuller autoethnographic treatment elsewhere. This is not to wholly efface my involvement, however, and where appropriate, the article does include contextual observations drawn from my own experience and participation in the field.
Of the 13 advisor interviews, eight were with former colleagues. A further five advisor participants were recruited using snowball sampling through personal contacts. The duration of participants' experiences of working as street-level advisors ranged from 6 months to 10 years, and several had worked on more than one contract (e.g., New Deal, Flexible New Deal, Pathways to Work, and the Work Programme). Participants were drawn from three provider organizations: two private prime contractors (n11) and one voluntary sector subcontractor (n2). Five participants also had experience working for the same employer in managerial, policy, or other head office roles. The range of different experiences among participants is both a limitation and a strength of the sample. The differing durations of street-level employment meant that whereas some participants had extensive experience of the sector as it changed and developed, others' experiences related more narrowly to one contract during a specific period. On the other hand, interviewing current advisors (n5) alongside former advisors (n8) offered insights into why people either continued or left the role. The research was limited by an advisor sample almost exclusively composed of employees from contracted-out providers. Although it is true that claimants were able to offer and compare their experiences of conditionality from both public and contracted-out providers, the data from advisors are limited to the latter. Focusing on data from advisors, this article is thus concerned with practices of conditionality in contracted-out services.
The advisor interviews were semi-structured, lasted between 40 and 180 min, recorded, transcribed, and then coded using Nvivo software. Participants were asked to talk about their role, everyday working practices, experiences of management, and interactions with clients. The research was approved by the University of Glasgow Social Science

Ethics Committee and conducted in accordance with the Economic and Social Research Council Research Ethics
Framework and the British Sociological Association Statement of Ethical Practice.

| Calculation, coercion, and caseload management
At its foundation, behavioral conditionality is about attendance. Prior to and underpinning any specified actions that are set-to apply for this job, to go to that course, claimants must attend a defined minimum number of regular (weekly, fortnightly, or monthly) work-focused interviews. Advisors were responsible for ensuring these conditions were met, and that each client attended the minimum number of work-focused interviews at the correct frequency.
Although caseload sizes fluctuated, the advisors I spoke to typically handled in excess of 100 active clients. Caseload management as diary management-deciding who to see, when, and how frequently-was thus a pressing concern.
This was not the only caseload management consideration. Although ensuring compliance and fulfilling the basic administrative requirements of conditionality were important, the overriding objective of any provider, office, or advisor was to "move clients into work." Payments to providers were tied to such job outcomes, and advisors' own jobs were conditional on meeting defined monthly job outcome targets. Caseload management therefore involved the calculated distribution of time and attention in pursuit of this goal. As one advisor, Deborah, explained: [There is] lots of time pressure. You have to prioritise which clients are more likely to be job-ready.
You'd look through all your caseload and think potentially I need to get X amount of people to work, looking at all of them, which ones are most likely to get into work? So you almost like do a prediction and think, which ones can I focus more on, and how can I focus on them? (Deborah, advisor) It is widely recognized, in the academic literature, official evaluations, and among practitioners themselves, that welfare-to-work providers distribute their attention very unevenly among clients. This "differentiated activation" (Wiggan, 2015) and the prioritization of some clients over others are usually understood in terms of "creaming" and "parking" (Carter & Whitworth, 2015;Greer et al., 2018;NAO, 2014). Creaming refers to the calculated allocation of time and attention to those clients who, with fewer problems or "barriers to work," are more likely to enter employment and become a job outcome; "parking" entails the neglect of clients with more problems and greater difficulties, and who are less likely to help achieve short-term outcomes.
There was little secret about this practice among participants. This despite the fact that, in the words of one former manager, Nicola, "providers are meant to say they don't do that, but of course they do, and they have to." The financial realities of the payment-by-results contracting arrangements left the organizations-and by extension, advisors-with little choice but to focus on quick wins and easy outcomes. Indeed, early in my own time, as an advisor, I was explicitly instructed to classify the clients on my caseload according to a traffic light coding scheme (cf: Greer et al., 2018Greer et al., , p. 1438: green for those "job-ready" clients that I expected to enter work within the current or next target period, and on whom I should focus my attention; amber for those who were expected, with some help (or pressure), to move up the queue; and red for those with "multiple barriers to work" who were not expected to start work any time soon and who ought not occupy significant time or attention. Although "creaming" and "parking" are typically understood as unofficial street-level adaptations and coping mechanisms, for the advisors I spoke to, they represented an official and managerially endorsed approach to handling a caseload.
At first glance, this might seem yet more clear evidence of "creaming and parking" in need of no further interpretation. However, I would argue that it warrants further attention for three reasons. First, although "creaming and KAUFMAN 210

| FINDINGS
parking" represented an officially endorsed strategy, it was not always one that advisors found so easy to implement. This poses a challenge to the interpretation of "creaming and parking" as the street-level path of least resistancesimply a rational response to street-level incentives. Second, to the extent that "creaming and parking" has become the taken-for-granted descriptor for differentiated treatment-both in the literature and among practitioners, it risks obscuring other forms and causes of differentiation in the treatment of claimants. As this article will show conditionality, judgement, and coercion, each had important roles to play. Finally, critical interpretations of differentiated treatment in terms of "creaming and parking" are underpinned by normative assumptions about the nature and value of the "services" being provided or withheld-assumptions which are subject to doubt and contestation, even among those providing and withholding.
Initially, it might not seem to matter much whether "creaming and parking" names a bottom-up street-level practice or a managerially promoted strategy; in either case, it is supposed that claimants are treated differently according to their perceived "distance" from the labour market. However, this difference mattered a great deal to advisors as they tried to manage large caseloads meeting both contractual requirements and performance targets.
The managerial expectation that they maintain lists of "job ready" clients and predicted employment start dates was one that many advisors judged unrealistic and a source of unwelcome pressure: Here, Jess draws attention to a gap between managerial expectations (and the creaming and parking model) and her street-level experience. The problem with a planned and calculative approach to caseload management, for advisors, was that their caseloads were not always so predictable or calculable. As Katie explained: I mean, you don't really get control over how your caseload is made up. So first of all, we can't control who's mandated [to do what]. And I don't control who I get. So there could be a string of two weeks where I get customers that are so far removed from the job market that I'm starting to think, oh my dear goodness, what am I, like, that doesn't really stand me in good stead. Whereas there's other days where just through the way it's divided, that you can get somebody that's gone off sick because they've had an operation and they're going back to their job. So I can't control that. (Katie, advisor) The inherent uncertainty of caseload allocations made creaming an undependable strategy. Not only did caseload sizes vary over time, but so did the qualities of a caseload, assessed in terms of the overall mix of job-ready clients and those with greater difficulties. Like the most sophisticated conceptualizations of individual employability (cf: McQuaid & Lindsay, 2005), the "quality" of a caseload ought not to be seen as intrinsic, essential, or fixed but as a flexible and relational property determined by various factors including wider labour market conditions. The problem, for advisors and providers alike, was that their targets were not similarly flexible. This meant that providers had to proceed "as if" wider labour market conditions were of no consequence, managing advisors "as if" the particularities of an individual caseload did not matter (cf: Crawford & Flint, 2015). That activation and welfare-to-work have exhibited an excessive and ideological focus on the unemployed welfare subject, at the expense of policy addressing wider structural and labour market factors, has been much commented on as a problem for the subject of welfare policies, but it also presented problems for those tasked with enacting them.
Such problems were particularly evident for advisors in the immediate aftermath of the 2008 financial crisis who, amid rising unemployment and competition for jobs, found it relatively more difficult to meet their targets than before. In subsequent years, Work Programme advisors encountered a slightly different problem when a falling claimant count meant that their flow of relatively "job-ready" Jobseekers Allowance referrals dried up, and their caseloads became dominated by longer term Employment and Support Allowance claimants with recognized health problems and no compulsion to find work. In these different circumstances, advisors, faced with inflexible performance targets, found that they did not have enough obvious candidates for quick wins and calculative creaming. In this context of uncertainty, where outcomes were never guaranteed in advance, advisors needed to try and make outcomes happen, and in this, they were aided by the coercive potential of conditionality.

| Disposition, discretion, and the coercive calculus
Sanctions-and the threat of sanctions-are an integral and defining feature of behavioral conditionality. If job outcome targets exemplified new public management and marketization, then sanctions epitomized the coercive and punitive turn. Contracted-out welfare-to-work providers did not apply sanctions directly but "raised doubts" about compliance with behavioral requirements. When a doubt was raised, the case would then be referred to a Department for Work and Pensions decision-maker who was, in the last instance, responsible for applying the sanction. This provided advisors with a useful source of ambiguity, enabling them to both raise doubts that resulted in sanctions, but also to deny responsibility for the consequences. As one advisor, Deborah, explained to her clients, "It's not me that makes that decision. I'm just following a process." This distinction between raising doubts and issuing sanctions was not always invoked, however, and it was tacitly understood that if a doubt was raised, then a sanction would likely follow.
Advisors were differently disposed to the coercive and punitive possibilities of their position. Whereas some were enthusiastic arbiters of claimants' compliance with conditional requirements, others were less comfortable with this aspect of their role. As in other street-level studies, advisors had different moral dispositions and working identities (Watkins-Hayes, 2009;Zacka, 2017). Despite the constraints of their position, such differences were held to have important consequences, particularly with respect to sanctioning practice: I sanctioned less people because my personality came into it. I was afraid of confrontation, and I did not want to be the person that took people's money away from them. Other people did it a lot because they had a more sort of penal attitude to it, or they thought that sanctioning people would get them back into the office … it was phenomenally inconsistent from advisor to advisor. (Robert, advisor) Your "hardnosed" ones are very transactional, so they'll use sanctions, that's just a part of their technique, it's their toolkit. Your "counsellors" will never sanction and so they'll have peopletheir managers will get annoyed with them, probablybut [the client has] failed to attend ten appointments, mandatory appointments, and [the advisor has] done nothing. (Alexandra, advisor) Both Robert and Alexandra differentiate between their colleagues according to dispositions toward coercion, punishment, and control. Interestingly, this axis of difference was often perceived to fit closely with another, namely, orientations toward either prioritizing the perceived needs and wishes of claimants or prioritizing the attainment of job outcome targets. For Alexandra, "hard-nosed bitches" (as she elsewhere put it) were not only more likely to make use of sanctions, but they were also the advisors most focused on "the numbers"; people who, she thought, would be equally suited to a sales or recruitment role. "Counsellors," on the other hand, were not only reluctant to make full use of their coercive powers but were also less well adapted to the target-driven environment. For advisors, then, the street-level calculus was not (or not only) a rational calculus but a coercive calculus imbued with moral and normative significance.
Raising doubts and initiating sanctions represented perhaps the clearest manifestation of the latent coercive potential of behavioral conditionality. But although advisors were differently disposed to handling this potential, it was not something from which they could abdicate altogether. Indeed, to the extent that the attendance and participation of clients was a mandatory requirement, coercion and the threat of sanctions underwrote their role. If advisors were uncomfortable with this, then it was sometimes possible for them to ignore the coercive basis of their position, particularly in the case of claimants who were themselves keen to comply, participate, and quickly find work. However, in cases where compliance was questionable, or in the face of more overt resistance, the fiction of voluntary participation became more difficult to sustain. Again, whereas for some advisors this was relatively unproblematic, for others, it represented an uncomfortable ethical dilemma.
When conditionality became a dilemma for advisors, it was rarely expressed in terms of abstract principle but in rather in terms of applicability and appropriateness. Many advisors who professed support for the principle of conditionality nevertheless also thought it too harsh or too widely applied: Some of the people that come, that are on health related benefits, they're not going to move on, towards work, in that time, and they shouldn't be here, they should be left to deal with their illnesses and their health issues … There's also people that have worked all their life. Worked, they've contributed their whole life. They're six months, eighteen months away from retiring, and they're mandated to come here, they've never worked a computer, they've never, you know? And I think, that's a bloody sin. (Judy, advisor) Judy, like many advisors, was ambivalent about the principle of mandatory participation underpinning her role. In some cases, she did feel that people ought to be doing more to find work and that mandatory participation-in, for example, directed and supervised job search-was justified. She also thought that, in other cases, mandatory participation meant that people who might not otherwise seek help could be directed to other appropriate services.
However, there were also instances, as above, where she rejected the principle of mandatory participation in the strongest moral terms as a "bloody sin." For Judy, these were instances where she judged someone as unconditionally entitled owing to their age, status, and past contribution or where she perceived that someone was too unwell to be realistically expected to pursue paid work.
Ostensibly, in the policy rhetoric surrounding activation, such cases ought not to present any ethical or moral dilemma because advisors' role was to provide "flexible and tailored support," and moreover clients on health-related benefits were under no obligation to actively seek work. Judy's ambivalence, however, was not uncommon among advisors, and this reflected an understanding that, whatever the policy rhetoric, participation necessarily entailed placing a certain amount of pressure on claimants. The target regime inevitably produced a work-first ethos aimed at stimulating short-term labour market entry, with little scope for advisors to offer more substantive forms of support or advice. Another advisor, Jess, could thus describe welfare-to-work as a "threatening environment" that was "more stick than carrot at the end of the day. It was basically like, well, you have to find something, or your benefits will potentially be lost." Like Judy, Jess judged that in many cases, paid work-often low quality and insecure-was either an unlikely prospect or something that would be detrimental to a person's health and well-being. Jess' solution to this dilemma was to differentiate her treatment of people on her caseload according to how feasible or beneficial she though finding paid work would be: "I thought that I would try and get jobs for the people that could potentially get one." Critical interpretations of differentiated treatment in terms of "creaming and parking" are not only based on analytic assumptions regarding its "rational" and calculated motivations but also normative assumptions about the nature and value of the service provided. When differentiated treatment is criticized in terms of the inequitable distribution of resources, those resources are understood in abstract terms distinct from the actual form in which they are delivered and assumed to be universally benign or beneficial. From the perspective of advisors like Jess and Judy, however, an inflexible and coercive emphasis on short-term job outcomes meant that the value of the service varied according to the needs and circumstances of their clients; they sometimes judged their own involvement in clients' lives more harmful than helpful. In this sense, decisions to "park" clients could be a way of protecting them from unwanted and unhelpful interference; conversely, decisions to prioritize or "cream" others could result from a judgement that greater coercive pressure was warranted. It can be seen that differentiated treatment resulted from both calculative and normative concerns. Viewed in terms of the coercive calculus, then, differentiated treatment was less about the equitable distribution of resources than about intensity, moderation, and the application of pressure. Advisors' primary means of applying, intensifying, or moderating the pressure on claimants was through expectation setting, to which the next section turns.

| Conditionality and the pressures of expectation
When advisors made choices about who to target-who to "cream" or who to "park" they were also making decisions about who ought to be pressurized. In this sense, decisions about caseload management, although governed by the need to meet a defined monthly target, were never simply instrumental or calculative. Clearly, given their targets, advisors did need to prioritize those claimants who appeared very "close to work," but this appearance itself was also a matter of judgement. When discussing their clients and caseloads, advisors were thus inclined to speak in terms of expectation, a word whose polysemy captures this mix of probability and prediction, judgement and directive. Expectation played a key role in activation services in the following ways, some of which have already been introduced: advisors were expected (by their managers) to draw up lists of those claimants expected to enter work each month; advisors used claimants' expectations as a means of assessing their "job-readiness"; advisors worked on adjusting claimants' expectations in order to make them more "job-ready." This section explores the role of expectation and its relationship to conditionality in more detail.
When drawing up lists of those clients on whom they would focus their attention, and for whom activation would be intensified, advisors' foremost consideration was "job readiness." This was a vague and loosely defined concept but one which many advisors held to be self-evident and easy to intuit: Jim: Could you just explain what, like from your point of view, what does 'job-ready' mean?
Robert: [Laughs]. It's very difficult to know. The term job ready is something that was talked about a lot in our provider. And it was, I took it to mean, people who you felt you could get a job for in the next month.
"Job-readiness" was intrinsically linked to the target regime and its short-term demands and was a way of referring to the speed and ease with which advisors thought someone might be able to find a job. Although almost all advisors felt that it was easy to intuit quickly whether or not a particular individual was job ready, it was also the case that advisors had different perceptions of the job readiness of particular individuals. Perceptions of job readiness reflected some of the subjective perceptions and normative prejudices of advisors. Sometimes this was explicitly acknowledged, and advisors talked about their preference for working with particular types of individual that others preferred to avoid-older people, lone parents, and young people. Nevertheless, amid such differences, the most important characteristics of a job-ready client were widely held to be their motivations and expectations and, in particular, whether their expectations were judged realistic: So [a job-ready client] is someone who knows what job they want to do, and is realistic. That's always the key. Because someone might say they want to be a teacher, but they'd never had a job in their life, so there's a long way to get there. So if someone comes in and says they want a job that is realistic, and they're motivated, and they want to have got the job like yesterdaythat's a job ready client.
(Isabelle, advisor) -In practice, having "realistic expectations" meant being willing to accept the widest possible range of jobs with the fewest conditions of one's own (in terms of hours, working pattern, and pay). Given the short-term demands of the target regime, advisors often had very little patience for claimants who had their own requirements and conditions or who expected to be able to pursue longer term aspirations, and these claimants were a cause of frustration or the subject of exaggerated mockery. Similarly, as others have noted (cf: Wright, 2003), advisors also disliked working with more qualified or higher status clients precisely because they had higher expectations. Advisors preferred working with people with fewer and lower expectations who were predisposed to compliance with the conditionality regime-those who might be easily "persuaded" to apply for work even when this was not a condition of their claim (in the case of Employment and Support Allowance claimants) or to apply for a wider range of jobs (in the case of Jobseekers Allowance claimants). As Alexandra told me, perhaps with a hint of facetious exaggeration, she "would rather have a murderer [than a professional] on my caseload" because "it's easier to get them into work because of the expectations." Explaining her reasoning, she thought that "somebody just released from prison was the best person to have because they were structured, they wanted a job, they were used to a kind of routine, it was a conditionthey had to really get work as well." Here, the conditionality of probationary supervision is seen as a valuable source of additional pressure.
Professionals (assumed to be without convictions), on the other hand, had "expectations that they're going to go into a CEO position, and you had to kind of break their spirit and say, realistically … you're going to maybe have to lower your expectations and do something else." Not only were expectations held to be a key indicator of someone's job-readiness, but they were also seen as one of the most important objects of advisors' work. Restricted by the work-first emphasis of activation, and unable to offer more substantive forms of support in the form of training or qualifications, advisors saw their role in terms of adjusting expectations. Occasionally, this was seen in inspirational and aspirational terms-raising expectations, and advisors presented their relationship to clients as one of paternalist encouragement. Echoing the discourses of both "demoralization" but also "rights and responsibilities," Nicola described her "personal style" as "all about expectation, expectation, aspiration," which involved "putting it back on them … that there's an expectation on them and that they are capable and yeah, they've got more skills and capabilities than people have expected them to have.
And I expected them to be capable and to succeed." Whether advisors couched expectation setting in terms of inspiration and raising expectation or realism and lowering expectation, it was always clear that the power and authority to do so derived from conditionality and the threat of sanctions: I do think that they've [advisors] got to come in and say, well this is what we expect of you for these, these and these reasons, and there will be sanctions imposed if you don't meet them. (Jess, advisor) If it got to the stage where they weren't complying or weren't doing anything, I'd be like, remember you said the worst job you could ever do was to work in McDonalds? There's a McDonalds vacancy, you're expected to go for it, so unless you find something else next week, you're expected to apply for that. So I would just use that as the stick, rather than sanctions. (Alexandra, advisor) Expectations were thus closely associated with both the target regime but also conditionality and the threat of sanctions and were also one of the main ways that advisors could exercise discretion in the course of their work.
Expectations were important because they determined the level and intensity of activation experienced by particular claimants: compliance with and completion of tasks was an important way that compliance with the conditionality regime more broadly was determined. As Alexandra implies, specifying or increasing the number of such tasks also increased or specified the grounds on which doubts might be raised and sanctions applied. Conversely, reducing the number of tasks or leaving them vague and open to interpretation could decrease the available grounds for determining noncompliance. Advisors could reduce the expectations of conditionality by setting as action plan tasks only the minimum requirements of attendance, or activities which a client either wanted to do or was already doing.
But they could also increase the pressure on claimants by increasing the number of appointments, or by mandating tasks and actions which a client was reluctant to do, had not done before, or might find difficult. In determining which course to take, advisors employed both moral, normative, and calculative forms of reason.
This article has offered a street-level perspective on behavioral conditionality in the context of contracted-out, work-first, target-driven activation programs. By exploring advisors' accounts of caseload management, sanctions, and expectation setting, it has argued that the form of discretion in such services is closely associated with the coercive potential of conditionality. Although marketization, performance management, and narrowly defined outcome targets may have limited the ability of advisors to determine the aims and scope of their role, the increase in behavioral conditionality nevertheless opened up a new disciplinary axis of discretion. Rather than curtailed as such, intensified conditionality shifted the locus of discretion to decisions about how much pressure to apply and on whom.
This shift is congruent with a more general rightward tilting of the welfare state in pursuit of differentiated recommodification (Wacquant, 2012;Wiggan, 2015).
The other contribution of this article lies in its attention to the specific form of agency produced by the convergence, at street-level, of calculative market rationalities with the coercive potential of conditionality. Although the street-level calculus posited a single, clearly defined short-term goal for street-level work-paid employment-for advisors, there remained considerable uncertainty about how (and whether) this was going to be achieved. The calculations that advisors engaged in were thus akin to a form of speculation in which they were required to make judgements about who among their caseload (as a kind of portfolio) was worth an investment of time and attention.
Such decisions entailed judgements about who was likely to respond to intensified activation and increased pressure but also normative judgements about who ought to be the subjects of conditionality and activation. This mixture of moral and motivated reason was expressed through the idiom of expectation, where expectations were both futureoriented predictions but also a form of normative judgement passed on (and onto) claimants. Given that the drive toward competitive market mechanisms and intensified conditionality have developed in tandem, this relationship between calculation and coercion perhaps deserves greater scrutiny and specification and is one avenue for further inquiry.

ACKNOWLEDGEMENT
The author is grateful to all those who read and discussed earlier drafts of this article, and in particular he wishes to thank John Flint and Ryan Powell for their valuable comments and suggestions.

FUNDING INFORMATION
Research Councils UK, Economic and Social Research Council, grant ES/S01134X/1.