Evidence from the ‘Frontline’? An Ethnographic Problematisation of Welfare-to-Work Administrator Opinions

Introduction

Researchers both supportive and critical of welfare schemes regularly explore the influence, legitimacy and effects of welfare administrator opinions. However, the ‘origins’ of those opinions are generally significantly less well considered. This article explores and problematises the use of welfare-to-work administrator testimony in social science and social policy research ‘on’ the unemployed and/or the effectiveness of unemployment welfare schemes.

Welfare-to-work administrators’ daily interaction with unemployment claimants and services grant them a malleable legitimacy as informants, broadly divisible into two approaches currently popular in welfare-to-work studies: either citing them as reasonably solid sources of in-the-field reportage, including on their own behaviours and use of discretion (Bartholomew, 2006; Dunn, 2014; Shildrick et al., 2013) or, conversely, as prosecutors of punitive neoliberal ‘micro-regulation’ (Goldberg, 2001; Peck, 2001; Schram et al., 2010). In the former case, administrator knowledge is assumed to stem from daily observation of schemes and participants; an idea captured by the phrase ‘frontline workers’. In the latter, from the more abstract Foucauldian concept of diffuse, pathological ‘governmentalities’, meaning, administrative techniques that transubstantiate service providers and users into organisational/neoliberal ‘subjectivities’ (Danaher et al., 2000). While it is not clear precisely what this means, the implication is that administrators have elite neoliberal ideology somehow ‘downloaded’ into them.

Drawing upon field research conducted in 2014 in two UK ‘Work Programme’ (henceforth ‘WP’) welfare-to-work centres, this article highlights an alternative possibility: that welfare-to-work centres are active sites of opinion construction in their own right. Second, that this often hidden level of opinion construction must be considered when assessing the validity of staff interview data. This article therefore contributes to sociological debates surrounding the day-to-day functioning of welfare-to-work programmes, first, by highlighting the importance of direct ethnographic enquiry (as opposed to interviews alone) in contextualising administrators’ statements. Second, by arguing that the fairly straightforward economic realities of day-to-day ‘service delivery’ can be at least as important in forming staff perceptions as any other factors; such as, for example, ‘elite discourse’, ‘street-level bureaucracy’, media influence or Foucauldian ‘governmentalities’.

The remainder of this article is divided into five sections. Section two provides a brief overview of the WP. Section three discusses the research and field methods used in this study. Section four provides a critical literature review. Section five presents the data, plus the outlines of an ethnography of the centres researched. Section six offers a brief conclusion.

The Work Programme

The WP is a welfare-to-work scheme introduced into the UK in 2011, which is set to run until the end of 2017, when it will be replaced by the ‘Work and Health Programme’. The latter will shift most WP emphasis onto sickness benefits claimants (Department for Work and Pensions (DWP), 2015). Currently, according to the UK government, the WP’s purpose is: ‘to increase employment [for the long-term unemployed] compared with previous schemes, decrease time spent on benefits, increase time employed for those coming off benefits, and narrow the performance gap between easier and harder to help claimants’ (National Audit Office (NAO), 2012: 5).

The UK government divides WP administration into 18 ‘contract package areas’ covering the entire British mainland (Newton et al., 2012). In 2011, 18 major organisations (two charities, 16 private companies) known as ‘primes’, were awarded WP service delivery contracts (NAO, 2012). The matching figures of 18 contract areas and 18 primes is a coincidence: 40 prime contracts were awarded nationally (Centre for Economic and Social Inclusion (CESI), 2013b). Each contract area has at least two primes operating within it. Five have three (Newton et al., 2012). Primes often subcontract to other service providers. Approximately 900 WP providers subcontract to primes nationally (CESI, 2013a: 4).

WP providers enjoy ‘freedom to develop [their] own processes to support participants, plan activity and manage their experience whilst on the Work Programme in line with [their] delivery model and contract terms and conditions’ (DWP, 2014d: 1). In pursuance of this aim, the UK government initially permitted providers to set their own minimum service delivery standards. By 2013, this decision had been revised. With 212 minimum service standards set by 18 primes, assessing compliance had proven impossible. The government’s contract expectations and performance monitoring processes were subsequently tightened (NAO, 2012: 34). While primes retain broad autonomy in specific service design, each prime’s promised minimum service provision is now listed in public contracts and approved by the government. All are generally similar (DWP, 2013a). For Igneus, ¹ for instance (not the prime involved in this study), minimum service provision is listed as:

1) A flexible service that is convenient and accessible.

2) A personalised package of support that is tailored to individual needs.

3) A professional In Work Support service which will help customers develop and progress in the workplace.

4) Access to job vacancies and labour market information.

5) Treating customers with respect at all times and enabling them to be active participants in setting their own goals. (Quoted directly from DWP, 2013a: 1)

Primes earn income by placing ‘referred’ (i.e. sent from the public Job Centre Plus [henceforth ‘JCP’]) participants into ‘sustainable employment’. Participants are divided into Jobseeker’s Allowance claimants (JSA) who, for the most part, have been unemployed for a year or more, and Employment and Support Allowance (ESA) claimants, who claim sickness benefits but who have been assessed as capable of performing some form of training or ‘work-related’ activity (Litchfield, 2013). However, providers receive nothing for just getting a claimant into work. An ‘outcome payment’ is only triggered when claimants remain in work for six months, or, in the case of some customers (e.g. ex-prisoners), three months (DWP, 2012). This six- or three-month period is classed as ‘sustainable employment’ by the government (DWP, 2013b). Payments are graduated to provide higher returns for placing more ‘difficult to employ’ categories, such as ESA claimants and ex-prisoners, into work. By March 2014, the outcome payments paid out nationally since the commencement of the WP amounted to £332 million, with £177 million paid out for the 12 months 2013–2014 (DWP, 2014c: 3), the year in which the present research was conducted. The total potential contract value of the WP has been estimated at £3–5 billion (NAO, 2012: 4).

Around 1.5 million people attended the WP between 2011–2014 (DWP, 2014b). The ‘success’ rate of the WP in its initial 12 months has been estimated at around 3 per cent (CESI, 2012: 1), and by 2014, at around 11.1 per cent (CESI, 2014: 2). JSA referrals were initially much higher than expected, followed by an increasing ratio of ESA to JSA participants, with ESA customers estimated to be more difficult to place into work. This is likely to have negatively affected outcome rates (House of Commons Library, 2016: 22).

WP attendance lasts for a maximum of two years, during which time participants continue to ‘sign on’ at the JCP. Failure to conform to the mandatory requirements of either the WP or the JCP can trigger a ‘sanction’, that is, loss of benefits. In total, 918,600 sanctions were levied in the fiscal year 2013–2014, including those levied against claimants mandated to the WP, but also those only ‘signing on’ at the JCP (DWP, 2014a).

Literature

A vast literature exists exploring the operation of welfare systems. Given the context of this article, this section presents only a sample review of influential, usefully instructive and/or contemporary treatments of welfare administrator testimony.

For critics of social democratic welfare regimes, and/or supporters of modern welfare-to-work, administrators generally ‘meet’ claimants, and experience scheme realities, at a social and economic ‘frontline’. In the simplest iteration of this assumption, claimant behaviours and scheme realities impress themselves upon the beleaguered administrators, who then report these realities to social researchers, either directly, or as secondary data. Classically, Moynihan’s (1965) The Negro Family and Murray’s (1984) Losing Ground each treat US administrators as generally unproblematic sources of welfare reality, in turn supporting influential claims of benefits-induced pathological welfare-dependency and putative ‘underclass’ behaviours. Following this tradition, but in modern British context, Dunn (2014) draws on interviews with 40 welfare ‘activation workers’ to support claims of intergenerational ‘worklessness’, claimant social dysfunction and intransigent ‘choosiness’ in refusing low-paid work. However, Marston (2013) argues that Dunn’s work is overly lenient in assuming that administrators’ opinions are unproblematic reflections of reality, despite being contradicted by empirical facts. A study by the Joseph Rowntree Foundation (2012: 1), for example, found that ‘There [is] no evidence of “a culture of worklessness”’ or ‘intergenerational unemployment’. Perhaps a more worrying approach in this vein is the use of administrator hearsay/anecdote, or a sense of ‘everyone knows that this is true’, as ersatz empirical evidence. Signally, the influential Centre for Social Justice (2013: 2) claims that ‘Intergenerational worklessness does exist […] Many of our Alliance members see it regularly; second, third or even fourth generation.’ Marston challenges such approaches for their failure to interrogate and contextualise vague, sometimes absurd, claims and contestable interpretations.

Mead’s (1986) influential and highly controversial Beyond Entitlement, which draws on in-the-field observations and discussions with administrators in the UK and the USA, offers the beginnings of an arguably more dynamic, ‘interactive’ model. Mead presents administrator views as influentially partisan, with successful welfare, or ‘workfare’, occurring when administrators are hard-nosed and focused on making claimants work. Conversely, ‘liberal-minded’ and overly sympathetic administrators encourage passivity and ‘benefit-seeking’. Bryner (1998) and Quaid (2002) make similar claims based on updated North American data, while Bane and Ellwood (1994) use a comparable methodological approach, but to argue the reverse: that is, that work-focused, hard-nosed administrators are correlated to cycles of welfare re-claiming, as opposed to more sympathetic, training-based approaches. Perkins (2016), meanwhile, extends Mead’s argument into neuro-biology, proposing that ‘passive’ administrators and welfare systems facilitate the emergence of a partially genetically inherited ‘welfare trait’, compounding the problem of intergenerational unemployment.

Theorists critical of welfare-to-work vary in their treatment of administrator testimony. A common theme, particularly in the USA, is to highlight the underlying racism and/or misogyny of welfare workers as precursory belief structures influencing later scheme operation (e.g. Handler, 1995; Piven and Cloward, 1971). In the UK, it has been more common to link administrator views to class chauvinism (e.g. Jones and Novak, 1999; Spicker, 1984; Westergaard, 1995). More recently, a ‘neoliberal conduit’ approach has tended to predominate in several critical welfare studies. This characterises administrator views as the terminus point in a process of elite conceptual download, regularly drawing on the paradigm of an overarching ‘workfare state’ (Goldberg, 2001; Peck, 2001) or ‘neoliberal state’ (Schram et al., 2010; Wacquant, 2010). The real-life presence of administrators within these conceptualisations varies. Boland (2016), for example, focuses entirely on UK welfare documentation, finding that while such written output is often presented by the government as conducive to the improvement of claimants’ self-esteem and self-efficacy, in reality, the implied discourses and related practical implementation of such materials prompt administrators to construct and reinforce the notion of unemployment as a wholly personal failure. Boland (2016: 348) concludes that: ‘Far from being empowering, discourses positing job-seeking as intense self-control which can generate convincing performances may be subjectively and psychologically injurious; the consequences of repeated failure or exhaustion through constant self re-invention have yet to be seen.’

Following a more field-based approach, Peck (2001) backgrounds real administrators as generalised prosecutors of ‘micro-regulatory’ regimes administering bullying and exploitation in the service of neoliberal economics, leading to a ‘structural congruence’ between government and business imperatives, justified by the intertwined ideologies of the ‘pathological underclass’, ‘competitiveness’, ‘flexibility’ and ‘personal responsibility’ (Peck, 2001, 2002). However, while Peck sees this process as embodying pathological neoliberal subjectivities within benefit claimants, he is less clear as to how administrators come to be active vessels of neoliberal governmentalities. Goldberg’s (2001) analysis of New York’s ‘Work Experience Programme’ arguably presents a more interactive and fleshed-out version of a similar approach. According to Goldberg, welfare-to-work is not simply a matter of power, domination or control; it is an active semiotic process of creating ‘welfare claimant’ as a symbolically polluted social category. Crucially, this involves interaction, and a flow of mutual influence, passing dialectically between welfare administrators and the sources of elite discourse, such as media, academics and government. However, administrators themselves remain notably absent from Goldberg’s study.

On the extreme end of the ‘absentee’ scale, Wacquant (2010) erases administrators altogether, transforming them into faceless malevolent spirits in a neoliberal welfare demonology. Theory is foregrounded, and welfare praxis all but erased.

Where real administrators do appear in Foucault influenced analyses, linking their testimony to the theory can remain problematic. For example, Schram et al. (2010: 746) argue that: ‘The new tools of governance […] such as performance systems, are instruments for a deeper project of governmentality […] designed to produce self-disciplining governing authorities by cultivating appropriate “governing mentalities”.’ Yet, the interview data used to evidence this highly theoreticist and specific claim are arguably sometimes unconvincing. For example:

When asked whether they have time to ‘work on skill development,’ case managers frequently give responses similar to this one: ‘No, you do not. […] You just have to refer ’em out, and, well, call this number, and see what you can find. (Schram et al., 2010: 750)

It is not clear how this represents a Foucauldian dissemination of diffuse ‘governmentalities’, rather than just administrators doing a minimal job with poor resources.

In the latter vein, an arguably more reality-based approach has been presented by some proponents of ‘street-level bureaucracy’ theory (Fletcher, 2011). Shildrick et al. (2013), for instance, agree that UK ‘welfare reform’ constitutes a deliberate New Right policy intended to re-valorise low-wage exploitation. However, they also find that administrator practices and attitudes are sometimes at odds with elite policy discourse, creating local frictions between government ambition and localised service realities. MacDonald and Marston (2005) similarly embrace street-level bureaucracy theory, but argue that interactions between administrators and service users can form sites of radical contestation in and of themselves. A recent variation on this approach has been to revert to the ‘direct data’ methodology, attempting straightforward empirical evidence gathering, but for purposes of critical and functional assessment. Grant’s (2013) study of the WP, for instance, finds advisers describing their role as ‘tailoring’ services over long periods to meet difficult individual customer needs; an approach often at odds with the generic and intensive employment outcomes targets imposed by management. Newton et al.’s (2012) assessment of the WP similarly found services ‘tailored’ to some degree, but severely limited by cost restraints, while Fuertes and Lindsay (2015) uncovered a minimal and generic service focused primarily on CV writing and ‘job search’. The authors describe their central conclusion as ‘the potential relevance of organizational and systemic factors in accounting for frontline practices and street-level interactions that have become increasingly […] repetitive, standardized jobsearch activities – speculative applications, “to whom it may concern” letter writing, and CV polishing’ (2015: 539). Contrary to the usual ‘street-level bureaucracy’ sense of administrators modifying elite policy goals, the authors argue that their findings indicate an institutional simplification of services that welfare workers merely administer, leaving little scope for localised alterations of schemes. Finally, Rees et al. (2014) conducted eight key informant interviews with workers in a range of roles associated with WP delivery. They conclude that the WP is not likely to provide a useful service given the limited tools and skills available to staff, as indicated by their own testimony.

Methods

This research was undertaken for a PhD (completed in 2016). The data are drawn from 18 days’ full-time fieldwork conducted inside two WP service delivery centres in two different north-western English towns in the summer of 2014. Permission to conduct the research was obtained from a senior gatekeeper following protracted negotiations.

Both centres were operated by the same provider. This provider was a ‘for-surplus’ social enterprise, subcontracted to a major for-profit prime. There was a slight difference in the unemployment rates for each town. However, service provision across both centres seemed virtually identical (as did average staff target achievement), with some staff working, or having worked, at both centres. The provider had been operating the WP since its inception in 2011, and had a significant previous history in employment and training services.

Interviews and ongoing discussions were conducted with 23 staff members, including prime and provider managers, trainers, ‘complex families’ workers, placement officers, customer advisers, an in-house self-employment adviser and administration/reception workers. All staff in both centres were white British. Fifteen were female, eight were male. Age ranges were as follows: 18–19: one; 20–29: six; 30–39: nine; 40–49: five; 50–59: one; 60–69: one. All of the 30–59 age group had some experience of working on previous welfare, or other social enterprise, schemes, as did four in the 20–29 category. Three staff members noted that they had previously been welfare claimants.

Research interviews were conducted in multiple stages over several days. As Spradley (1979) notes, a process of ‘incremental discussions’ helps to build rapport, leading to more detailed revelations. For example, initially, several staff members thought that the research was a management ‘spying exercise’. Continuous field presence over several days radically altered this perspective, leading to deeper revelations about workplace practices, and several participant ‘modifications’ of previous statements; for example, through participants adding significant further detail or nuance after giving thought to their previous statements. In several cases there was a noticeable change in staff responses over several days, from fairly formal repetitions of generic welfare-to-work policy claims, through to more detailed and confidential revelations regarding in-house practices, the working environment and personal feelings, later on.

When respondents provide their opinions over longer periods (as opposed to one-off interviews) they sometimes give contradictory statements (Thomas, 1993). A reasonably common occurrence, for example, was for an interviewee to say something off-hand, such as ‘They’re all skivers in here’, then, at a later point, to state that most WP participants would work if there were jobs available for them. The question is, which statement ‘represents’ that interviewee’s ‘real view’? The answer is neither, and both. Human beings often express contradictory views depending on context; something that a formal single-interview method does not always capture. Good ethnography should at least attempt to explore this complexity of field narrative (Dubois, 2009; Thomas, 1993).

Incremental discussions also permitted interview content to form organically around specific events that often occurred immediately prior to the discussions. ‘Events’ observed included multiple staff/participant interviews (well over 100 observed directly, albeit sometimes more than one at a time, and sometimes only partially), team meetings, placement officer meetings, staff and participant training, ‘job search’ sessions and the majority of general day-to-day centre life. (Some private meetings were not observed.) This led to a more in-depth and ethnographically grounded interaction, and a deeper understanding of the field, contextualised and triangulated by exposure to actual service delivery. As Brady (2011: 278) notes: ‘while it is self-evident that social policies are not homogeneous entities it is only field investigation of these programs that makes this visible’. Interviews were also conducted with programme participants, but these data are not included in this study.

Nevertheless, analysing ethnographic data is an art, not a science (Spradley, 1979). The data presented here represent what appeared to me to be the most genuine examples of staff views according to my ethnographic experience of these participants over multiple days. The participants might also reasonably be expected to have been ‘on their guard’ to some degree. As Robertson and Dearling (2004) note, this should be taken into account when considering the data; as should the fact that the data have been selected for emphasis by the researcher.

Staff decided, for the most part, what they thought was relevant to disclose. Nevertheless, as Davies (1999) points out, even in fairly unstructured interviews the researcher is often guiding the process through subtle, or not so subtle, prompts. The purpose of the research was to assess overall delivery practice, hence, information was continually elicited that related to key theoretical and policy claims regarding welfare-to-work, mostly drawn from academic literature and wider public policy discourse. While the specific phrasing of research questions varied, the following information was elicited from all staff:

The data presented here reflect, primarily, the opinions of staff regarding the WP and its participants. However, this forms only one subset of the wider data gathered. The present data have been selected specifically for their relevance to wider debates surrounding the role of administrator views and opinions in welfare-to-work delivery; although the surrounding discussion draws these data into an ethnographic consideration of how these views were formed. The data and discussion presented here cannot necessarily be extrapolated to the WP generally. All names are pseudonyms.

Findings

While Foucauldian analyses of administrator views generally emphasise inculcation within elite discourse, interviews and discussions indicated that staff did not read academic ‘neoliberal’ literature regarding welfare. No other direct ‘download’ mechanism was identifiable within the centres researched. There were no intensive training sessions, reorientation days, re-education or other forms of neoliberal indoctrination. All staff training was ‘on the job’, that is, shadowing other employees for two to three weeks before commencing their roles.

Some ‘diffuse’ influence, such as the general public might be subject to via media output and immersion within general social life, might be plausibly assumed. Staff were also unquestionably sensitised to media and political output due to their employment roles, with several making comments such as ‘I see this on the news…’ However, this was invariably followed by a ‘but…’ and a questioning of the media or political claim, usually based on the idea that only ‘frontline experience’ can give someone the ‘real picture’. The data indicate that this sensitisation may have had some influence on staff views, most significantly through the use of key terms such as ‘underclass’, ‘intergenerational worklessness’, ‘troubled families’, ‘skivers’ and ‘welfare reform’. Several key welfare-to-work ideas and terms were also used in the formal documentation issued by the contracting prime, provider management and the DWP. Hence, some ‘elite conceptual download’ is plausible, as explored by Boland (2016) and Goldberg (2001).

Mead’s (1986) argument that staff generally come to the role with prior ideological leanings and opinions is also plausible, being a fairly unobjectionable truism. Yet, Mead claims that staff practices ‘loop’ into a self-fulfilling reinforcement of beliefs. As might be expected, this proved an unhelpfully one-dimensional proposition. Notably, it failed to account for the potential power of groups and circumstances to influence individual opinions towards a consensus (Berger, 1963).

The positivistic ‘unproblematic reportage’ model, that is, that staff views broadly recount ‘frontline’ reality (Bartholomew, 2006; Dunn, 2014; Shildrick et al., 2013), could explain why administrator opinions often strongly cohere around central themes (see, for example, Dunn, 2014; Goldberg, 2001). In such cases, a group consensus might be expected, with or without external influences. For example, in the present research, most staff strongly articulated that they were on the ‘frontline of welfare reform’ – a common term used by politicians, media and academics. As one, ‘Cormac’, put it:

We’ve got an incredibly difficult ask put on us here. We’re the frontline in coping with, er… what we’re coping with is sometimes third or fourth generation unemployment. You must know that with all your research. Who picks up the pieces? When does that get dealt with? Who intervenes because you’ve got whole families living like that, when your dad and your grandad never bothered to get up and go to work, then what kind of environment are you getting your values from?

Similarly, most staff (21 out of 23) strongly articulated the idea that a culture of ‘welfare dependency’ blighted the UK, and that this was precipitated by an overly generous benefits system.

Reflecting another key media and political claim, all staff also argued that there were at least some, if not many, ‘skivers’ on the programme. As ‘Trisha’ put it:

I think we’ve really dug ourselves into a ditch with the welfare system by paying people to do nothing. I think we’ve encouraged a whole generation of people to think just going on the dole is a real option. It’s so ingrained now as a culture, it’s an uphill battle to change it.

Coding such data is relatively easy in both the elite conceptual download and the unproblematic frontline reportage models. In the latter case, it is taken for granted as fact (cf. Dunn, 2014). In the former, staff views are checked against factual data, such as statistics (e.g. Shildrick et al., 2013), or else, seemingly, sifted to discover where they match ‘neoliberal’ ideology (e.g. Schram et al., 2010). Comparing interview data to neoliberal ideology is a particularly useful methodological approach in that it solves the problem of how to interpret and contextualise interview data when the research lacks a broader ethnographic setting (see Brady, 2011). Comparison to neoliberal ideology becomes the context, and this approach can be circularly justified by reference to the Foucauldian model of neoliberal governmentality that the data might be being carefully selected to fit. While this might be justified in documentary analysis (e.g. Boland, 2016), the present research found significantly more texture, contradiction and variation in administrator views, and even often in a single respondent’s comments. Two examples will help to unpack this complexity: the key ‘neoliberal’ claims of widespread ‘workshyness’ and of ‘intergenerational worklessness’.

The issue of putative workshyness was far more complex than staff internalising neoliberal pathologisations of all benefits claimants as ‘skivers’. Numerous WP participants, for example, thought that there were many people on-programme who were lazy and workshy. Still, ‘skiver’ – and that word in particular – was a nuanced category of meaning for staff. As ‘Holly’ put it:

Some [participants] could skive for England. Look, there’s all sorts. It’s common sense, isn’t it? […]. If you want to show people that there are people here who obviously shouldn’t be here, well yeah, there are, but there are people who really should be here. They should have been here years ago.

Being described as a skiver within the staff’s group discourse was generally associated with not applying for enough jobs. ‘Enough’ was a variable figure, however, dependent on a caseworker’s discretion (cf. Fletcher, 2011), but usually meaning five to 10 applications per week. In addition to inadequate job-seeking, primary evidence for ‘skiving’ was established from a combination of a participant’s length of time on benefits, a related belief that the past welfare system had been too passive and, significantly, participants’ alleged ‘attitude’ in review meetings. As ‘Trisha’ put it of the latter: ‘Some of them just aren’t real… you know the attitude… like, aggressive, like saying we can’t tell them what to do and we can’t take their money off them. We can.’

‘Intergenerational unemployment’ was cited by six staff as being responsible for a small but noticeable minority of the participant base. According to ‘Trisha’, for example: ‘A certain culture set in [in the past] and people think, well… why work? And then they see that their dads and grandads, well they don’t go out and work either. So why should they?’ (‘Dads’ and ‘grandads’ were mentioned a handful of times, but never mothers or grandmothers in this context.) All staff recognised that a range of other factors mitigated against the employment of certain categories, particularly ESA claimants. As ‘Xandra’ put it: ‘There are people who really are in health trouble, you know, serious back problems, knees that don’t work, diseases, right through to people with mental health issues.’ A poor job market for older people was similarly recognised by all staff as, to some degree, a key ‘employment barrier’, with one noting that: ‘[T]hey [management] think that you can get a bunch of ex-brickies or long-term unemployed on a basic computer course and somehow they’re going to be transformed into wonderful, work-ready individuals. Never going to happen’ (Charles).

Reflecting the way in which the same staff members sometimes articulated contradictory opinions, ‘Trisha’, a broad supporter of the WP, noted that: ‘I don’t know what the government is doing with this [the WP]. It’s a cruel programme. […] No one wants duffers [old people]. But it’s not a moral economic policy to just force the old folks into pointless schemes.’

Similarly, and in harmony with the findings of Grant (2013), all ‘caseworkers’ partially distanced themselves from sanctioning by arguing that sanctions were mostly automated (e.g. for being 15 minutes late), while also arguing, on separate occasions, that they themselves exerted significant discretion in not sanctioning ‘vulnerable’ participants. While this reality broadly supports ‘street-level bureaucracy’ theory (cf. Fletcher, 2011), the interview data are clearly messy, and arguably less solid empirical reportage, and possibly more a reflection of psychological self-evaluation and distancing-from-blame behaviours.

There was recognition across virtually all staff that structural unemployment was real. However, most staff articulated that this was often aggravated by a claimant’s behaviour. As one ‘complex families’ adviser stated of her caseload:

If I was to describe an average customer I’d say older male, white, debt and alcohol problems, depression. What I’m seeing, it’s like I know it. I’ve been seeing it since I was a kid back where I live. If you’ve got whole industries disappearing and even the men who’ve worked all their lives find it really hard to get into work.

Poor mental health and low self-esteem were likewise regularly cited as key ‘employment barriers’, often linked to being unemployed itself.

Fieldwork also highlighted that staff did not simply ‘hold’ views, ‘neoliberal’ or otherwise; they held many views with real emotionality (cf. Quaid, 2002). The most commonly articulated view in this regard was that many participants, in so many words, ‘take the piss’ (a common staff phrase in the centres): ‘It’s taking the piss at the moment, with the benefits system. Like it pays everything. People just living on benefits doing nothing at all. Staying in bed. I don’t agree with it’ (‘Xandra’). While many people might share such views, even those who are not welfare administrators, this sentiment had a strongly personal element to it in this research. As ‘Jacqui’ put it: ‘they spend so much time on the welfare system. They know how to play it. And they know how to push your buttons.’ This provides a conceptual window onto an arguably more robust model for assessing the origins of administrator opinions: a materialist field ethnographic approach.

Staff were not on any ‘frontline’; an arguably over-the-top, militaristic metaphor that replicates administrator self-perceptions more than it reflects reality. Most centre WP participants were unremarkable, save for being long-term unemployed (one year or more), mostly male (around 70%) and mostly former manual or skilled trade workers. However, staff were a group who regularly met another group. This interaction took place within two key contexts. First, staff spent all day together. Second, the centres were run, as one senior prime manager stated, on a minimal budget. Full-time customer advisers had caseloads of between 120–180 people. Each ‘participant’ attended the centre, on average, only once per month for a 15–30-minute interview. It is important to emphasise that staff were not JCP employees or public sector workers; they were privately hired, and working in a target-driven role (cf. Grant, 2013). ESA advisers were expected to place two participants per month into work, and JSA advisers six. This was high-pressure work. As ‘Jacqui’ put it: ‘I don’t think anyone could work in this industry long-term now. The stress is too high. The targeting is too high compared to the customer base and the job market.’ Several staff described, in so many words, the treatment by management of those who failed to achieve their targets as outright bullying. Moreover, new additions to the caseloads were gradually falling due to most eligible long-term unemployed people already being on the programme. According to staff, this meant an increasing ratio of unemployable to employable participants in their caseloads; ‘employable’ meaning that they were highly likely to find work on their own. Given this managerial pressure (cf. Grant, 2013), it is understandable that staff might come to feel, and articulate, negative feelings towards ‘customers’ who they felt could be doing much more to get back into work. Personal interest, in other words, might be a more plausible cause of emotionally held views than ‘diffuse governmentalities’. This could also partially problematise conclusions such as Dunn’s (2014) claim that WP attendees were too ‘choosy’ in refusing employment, which is based on administrator testimony, and also Perkins’ (2016) claim that some welfare claimants have a scientifically detectable, partially genetically encoded, ‘disagreeable personality’. A great deal depends on who does the ‘detecting’, and what might be influencing their perceptions and emotions.

However, this is still not sufficient to explain why most staff strongly supported notions of a widespread, benefit-dependent underclass. A key insight here is that staff were not one-dimensional in their understanding of this concept. Belief in a dependency culture did not imply any de facto staff belief that every claimant was idle or anti-social. Rather, the ‘dependency culture’ was generally conceived of as the overall ‘customer base’, which was ‘dysfunctional’ by definition because its members were on the WP, but by no means as necessarily idle by definition:

I mean, basically, it’s dysfunctional individuals, isn’t it? It’s not just not having a job. That’s the situation but there’s reasons why a lot of them are unemployed and some of the reasons are like… I think wow, how have they managed to get this far without someone holding their hand? But of course that’s exactly why they’ve ended up here, isn’t it? (‘Jacqui’)

Another important factor may be the rate and type of interactive flow between staff and participants (cf. Newton et al., 2012). While individual participants were rarely in a centre, the flow of participants into the centres was relentless for each caseworker. Many participants had genuine problems with health, poverty, feeling down or debt, while a minority had mental health or addiction problems. Participants became ‘them’ to staff; a largely homogenous group entity, with some individuals coming to represent the whole; and conversely, one or two ‘exceptional’ individuals who represented ‘gumption’ and self-efficacy, who were rare by definition. Moments of group-belief reinforcement in this regard were observed regularly, usually following a ‘review’ interview with a particularly ‘negative’ (a staff keyword) WP participant. In such cases, a staff member would discuss a participant’s alleged idleness, intransigence or ‘bad attitude’ using words that pluralised individuals into broader generalisations, such as ‘typical skiver’ or ‘more of the usual excuses.’

A second key issue might have been the inability of staff to provide any real help. Most caseworker interviews were ‘chats’ about the participants’ lives, problems, health and so on. In the vast majority of cases, beyond CV writing and basic skills training (e.g. setting up an email or advice on interview techniques) staff could do nothing to help their caseloads (cf. Fuertes and Lindsay, 2015). Staff knew this, and articulated it. Even for staff who did reach their targets, these were so small compared to the size of the customer base that any sense of realistically altering long-term unemployment in the local areas was palpably absent from office discussions. As one noted: ‘you’re getting it dented, you know, but that’s it’. It was useful for staff to view their ‘customers’ as part of a pathological underclass; it gave meaning to their own role and made their failure as ‘job coaches’ seem inevitable. The defining feature of a pathological underclass, at least as a theoretical entity, is that it is pathological: that is, intransigent and chronic. As ‘Cormac’ put it: ‘If I went around to all the providers around here I can guarantee you I’ll see the same faces I saw 10 years ago, or over the years. It’s just the nature of the beast’, adding on another day:

realistically speaking maybe there is no answer for this generation of claimants. It’s their way of life. But I think it’s important to send the message to the next generation that this isn’t an acceptable way of life for anyone.

Belief in a pathological underclass made poor employment outcomes seem inevitable, and more importantly, to be the fault of the participants and not the staff. Moreover, given that staff could only offer basic social support and signposting advice, it transformed this ‘soft outcome’ work into a positive occupation, despite it having little effect on employment outcomes. This perception had particular value in a stressful, managerialist and counter-productively target-driven environment (cf. Fuertes and Lindsay, 2015; Grant, 2013). As ‘Pauline’ explained:

They [management] just want hard outcomes, jobs, but you’ve seen, you can’t do it. Like Philip, what can you do? He comes in, he’s got mental health problems. Half the year he’s homeless. I can’t just put him into a job even if there was one for him.

The WP itself was also broadly decried by staff as financially and organisationally incapable of tackling long-term worklessness. Lack of funding for better participant training and job creation were the primary complaints. As ‘Anna’ put it: ‘I don’t think it’s [WP] ideal by any means. And I think intervention is only one part of the puzzle. There’s got to be opportunities and money available to get people into training to fill them.’ The deflection of blame (not necessarily illegitimately) for minimal service outcomes onto the programme and/or the government, while simultaneously emphasising their own hard work despite such lacks and participant dysfunctions, was a noticeable element of the staff’s perspective (cf. Rees et al., 2014).

The final, and perhaps most significant, indicator that day-to-day realities, plus administrator roles and psychological immersion within these, influenced strongly staff opinion formation, was the notion of ‘complexity’. Phrases such as ‘complex problems’, ‘complex issues’ and ‘it’s complex’ occurred regularly, both in interviews and in general office conversations. As one staff member noted:

But when you’ve got people falling through the net, you get… it’s a complex problem, yeah? They end up falling through every hole in the net. So you’ve got no work, but your relationship with your wife and kids is a train wreck. You’re in debt, you drink, you fight, you wake up in a cell, your health collapses. I mean that is one whole category of customer… very complex but no one goes long-term unemployed without that bringing with it a lot of other problems.

‘Complexity’ is, potentially, a powerful ideological tool permitting the genuinely interconnected nature of numerous social problems to be unfairly ‘evened out’, making each element seem of equal importance. The causal power of structural factors in prompting wide-scale long-term unemployment can be downplayed by this means, or even erased, as the arrow of causality spins to all points simultaneously. That might serve ‘elite’ interests, but far more significantly to administrators, by the same means all factors, structural and personal, can be equally discursively ‘weighted’ in terms of importance. This might make a minor ‘intervention’ into a single person’s ‘dysfunctional’ lifestyle appear, and actually emotionally feel, as palliatively significant as addressing real, mass structural defects. There was strong interview evidence that staff felt this. As ‘Annemarie’ put it:

If there’s one thing I take away from here it’s that you can’t judge it on the statistics. Soft outcomes [i.e. personal help not leading to employment] are a vital part of the work we do and often the most rewarding part because a job role could just be for a few weeks but if you help someone overcome a major life-barrier, you’ve made a significant intervention in their life.

‘Complexity’ was also regularly linked to the idea of participants’ problems being ‘long-term’. Staff could ‘signpost’ participants to debt counsellors, doctors or treatment clinics, or they could engage in what many of them saw as ‘morale raising’ chats, or ‘tough love’; all things that might, in theory, ‘help towards’ (a vague concept, as one ‘customer’ noted) gaining employment, but only as a ‘long-term goal’. By this means, a poor WP outcome rate, and an employment role that staff fully accepted barely dinted the real problem of mass long-term unemployment, were transformed into an apparently hugely positive ‘social service’ (cf. Fletcher, 2011). This self-perception of role did not match the target-driven, managerialist approach expected by the government or local managers.

Finally, managerialism had itself become absorbed into the staff views, as one more complex ‘barrier’ to achieving what really mattered, that is, doing at least something to fight back against the tide of long-term, complex problem-bearing members of the underclass swamping ‘the system’, by ‘intervening’ in ‘their’ attitudes and lifestyles. As one staff member put it, explaining how they had sanctioned someone for ‘non-compliance’:

To cut a long story short they ended up going under financially, rent wasn’t paid and they were out, homeless. That all comes from a sanction. But you see that on the news, you don’t see that I then worked with that customer to find them a new home. And then you start re-building. It’s a shock. But from that you re-build, start building up your relationship with them and start getting them to take their responsibilities for themselves seriously. If you didn’t have sanctions, what else could you do?

This ‘customer’ did not gain employment from the WP; but her caseworker was clearly proud of this ‘soft outcome’ intervention.

Conclusion

Welfare administrator views form vitally important components of much welfare-to-work research. Recent debates, particularly Marston versus Dunn, and contentions, such as that between the Centre for Social Justice and the Joseph Rowntree Foundation regarding ‘intergenerational worklessness’, have highlighted the need to problematise these views with reference to empirical field contexts (see Brady, 2011; Marston, 2013). Without ethnographic fieldwork, welfare-to-work interview data contextualisation has tended to fall, broadly speaking, somewhere between acceptance as ‘frontline reportage’, and a Foucauldian governmentality/elite conceptual download model. Exploring data drawn from both interviews and fieldwork, this study explored a third option: that the day-to-day realities of welfare-to-work staff might plausibly be shown to provide both a strong – and also a strongly skewing – influence on their views. This includes reflections on their own day-to-day activities, such as ‘discretion’, sanctioning and providing ‘help and support’. A pressurised, low-budget and target-driven environment, sharing the same working space, plus a constant stream of problem-bearing, demoralised ‘customers’ and a need to give meaning to an otherwise largely pointless employment role, each contributed to the construction of the majority of WP participants as a singular underclass cohort. The alleged chronic and intransigently pathological nature of this group excused and explained away staff failures to provide anything beyond the most rudimentary forms of help. Moreover, while staff recognised the existence of structural causes for unemployment, the ideology of a ‘complex’ socio-economic pathology ‘weighted’ structural and personal factors equally. This perception used key ‘neoliberal’ terms, that is, ‘worklessness’ and ‘welfare reform’, and it drew on real people with real problems for its empirical basis. However, it also radically inflated staff self-perceptions of their roles as ‘frontline’ social medics providing vital, if doomed-to-fail, ‘help and support’. Non-employment-related ‘soft outcomes’, often meaning just having a chat and signposting, became transformed into positive, meaningful interventions, which required, by definition, a strong belief in the existence of a pathological, welfare-dependent underclass.