Welfare Conditionality and Blaming the Unemployed

Welfare Conditionality and the Moral Economy of Activation

Welfare conditionality stretches back to the Victorian Poor Laws (Pantazis, 2016), although its more recent history can be traced to the late 1980s when conditionality requirements began to be (re)introduced in response to rising caseloads. In Australia, following the recommendations of the 1986 Social Security Review, recipients of out-of-work payments began to have to register and report their job search efforts to the CES to continue receiving assistance (Deeming, 2016). This was soon followed by the introduction of a new case management model under Working Nation, which required jobseekers to take any offers of suitable employment while empowering caseworkers to recommend payment suspensions if clients failed to attend interviews, training courses, or “other activities deemed to be helpful” (Considine, 1999, p. 189). In exchange for increasing the intensity of recipients’ conditionality requirements, the government committed to a reciprocal obligation to provide paid work opportunities for those who had been unemployed for 18 months or more (Fowkes, 2011). However, following the election of the Howard Coalition government, reciprocal obligation became “mutual obligation” in the form of a requirement to undertake WfD—unpaid community work placements—for 6 months out of every year. This was legislated under the Social Security Legislation Amendment (Work for the Dole) Act 1997, applying initially only to those aged under 25 years before being extended to all those under 50 years in the early 2000s (Deeming, 2016). In 2009, a new JSA system was introduced by the Rudd Labor Government, in part to address wide-ranging concerns that JN had become too prescriptive and that difficult-to-place jobseekers were being given little or no assistance. The funding model was adjusted to increase the incentives for agencies to focus on supporting harder-to-help clients and the compliance framework was partially softened, as the focus turned to “connection and cooperation” via temporary payment suspensions rather than the “standardised loss of payment” that had previously applied to a wider range of circumstances (Ramia & Carney, 2010, p. 268). However, many of these changes were removed when JSA was replaced by Jobactive in 2015. There was a return to strongly enforcing WfD requirements, and agencies’ powers to suspend clients’ payments and to recommend penalties for persistent noncompliance were increased (DoE, 2016).

This work-related conditionality model is underpinned by a deficit model of welfare recipients that was controversially advanced by Laurence Mead and others in the mid-1980s to explain the rise in the number of young working-class people on welfare. Rather than seeing this as a consequence of weakening labor market conditions, Mead (1986) attributed much of it to the de-motivation of the poor and the inability of administrators “to obligate the recipients of programs, even for their own benefit” (p. x). He saw welfare as a kind of moral hazard in which the longer people remained on benefits the greater at risk they were of becoming de-motivated and crippled by the fatalist belief, fostered by being on benefits, “that their fate turns on forces outside themselves” (p. 145). He called for social security entitlements to be coupled “with serious work and other obligations” (pp. 3-4) as recipients, in his view, had “too many other sources of income . . . to work reliably unless programs require them to do so” (p. 13). Mead’s arguments paved the way for a transformation in the institutional character of the neoliberal welfare state from a “people processing” to a “people changing” logic (van Berkel & van Der Aa, 2012, p. 494). Frontline workers became pivotal to this project of “resubjectivation” (Whitworth & Carter, 2014, p. 110). Through regular appointments and targeted sanctions, they became responsible for inculcating in clients “the attitudes and behaviour representative of responsible self-government” (McDonald & Marston, 2005, pp. 390-391). This emphasis on shaping how claimants “think about and regulate themselves” (Soss et al., 2011a, p. 4) most distinguished what Mead (1991/2014) termed a “politics of conduct” (p. 93). It was not enough for governments to coerce people. Welfare recipients, Mead (1986) argued, must be persuaded “to blame themselves” (p. 10).

Activation Workers’ Perceptions of the Unemployed

This “conduct shaping” dimension of frontline work has been negatively evaluated by critics of conditionality, who point to its potentially corrosive effect on the caseworker–client relationship (van Berkel & van Der Aa, 2012). As McDonald and Marston argue, “strengths-based” approaches depend on caseworkers adopting “an optimistic attitude towards the unemployed” (2008, p. 316). However, analyzing the views of 183 JN staff expressed in open-ended survey questions, they found that the majority held predominantly negative attitudes toward their clients, indicating that they were “content on benefits” and “needed to be forced into the workforce” (McDonald & Marston, 2008, p. 320). In more recent interview-based research with JSA and emergency relief caseworkers, Agllias, Howard, Schubert, and Gray (2016) found that most appealed to “individualistic psychological explanations” of unemployment, although workers tended to adopt more structural explanations the more contact they had with clients (pp. 302-303). These findings are consistent with research by Dunn (2013) on agency workers’ perceptions of U.K. Work Programme participants, with most estimating that 30% or more of their clients did not want to work and were “too choosy in the jobs they were willing to undertake” (p. 799)—a finding replicated in Jordan’s (2018) ethnographic study of two Work Programme delivery sites.

While previous studies have highlighted the extent to which frontline workers endorse behavioral explanations of unemployment, whether this is also associated with an increased willingness to use sanctions and enforce “work-first” remains unknown. Although the literature on the discursive framing of policy discourses points in this direction, insofar as it is the presumption that welfare recipients are de-motivated to work that “explains why work-first content tends to focus on . . . compelling people to ‘do more’” (Fuertes & Lindsay, 2015, p. 528). It stands to reason that activation workers who endorse this framing will be more disposed to using behavioral policy instruments. Several studies within the street-level bureaucracy literature also point in this direction insofar as they highlight how caseworkers’ determinations about who to treat routinely, minimally, or even harshly, and whom to give extra attention are often guided by normative judgments about clients’ perceived moral worthiness. This is highlighted by Maynard-Moody and Musheno’s (2000, p. 348) ethnographic research with police officers, teachers, and counselors. The street-level workers they interviewed saw themselves as citizen-agents “who act in response to individual citizen clients in specific circumstances” rather than agents of the state responsible for administering bureaucratic rules. The type of decision making they described encompassed moral reasoning about clients as persons, which then shaped whether administrative rules were interpreted “to help those they consider[ed] worthy [or to] . . . punish those they deem[ed] unworthy” (Maynard-Moody & Musheno, 2000, p. 351). Soss, Fording, and Schram (2011a) draw a similar conclusion on the basis of their research with welfare-to-work caseworkers in Florida, finding that “few factors are as decisive as the case manager’s assessment of the client as a person”: whether they are “a straight shooter, trying hard, or playing games” (p. 249). Importantly, their research suggests that caseworkers’ evaluations of clients in these regards are also shaped by ascribed social identities “in far-reaching” (Soss et al., 2011a, p. 248) and racially coded ways that generate strong biases in program implementation. This is confirmed by their experimental research in which caseworkers were more likely to sanction clients with discrediting behavioral markers such as a history of noncompliance if they were depicted as from a minority (especially Black) group. This disparity in the racial targeting of sanctions, they argue, reflects how the images of the poor used to justify the disciplinary turn in social policy “did not emerge de novo” but drew on a powerful preexisting cultural template by which dysfunctional stereotypes about “the black poor were generalised to the poor as a whole” (2011, p. 82). These dysfunctional stereotypes play a critical performative role in the “cultural denial” of poverty as something “that happens only to a tiny minority of pathological people” (Eubanks, 2018, p. 175). Moreover, as Eubanks’s work on the spread of digital decision making in public assistance programs shows, these cultural and racial minorities are not just singled out for targeting by caseworkers’ biases. Their lives and choices are also increasingly monitored by more intensive forms of digital surveillance that disproportionately expose them to “the rational discrimination of high-tech tools” (Eubanks, 2018, p. 192).

Our data do not allow us to consider whether the implementation of welfare-to-work in Australia is similarly biased by implicit processes of racial and cultural coding (although see Klein, 2016). Nonetheless, we can reasonably hypothesize that activation workers’ perceptions of the unemployed more generally will influence how they enact work-related conditionality on the ground. To the extent that increasing numbers of caseworkers come to endorse behavioral explanations of unemployment, we would expect to see a convergence around work-first approaches that prioritize “encouraging increased effort in standardised job-search activities” (Fuertes & Lindsay, 2015, p. 535), although any such convergence might alternatively be accounted for by the more intensive forms of performance monitoring applying to activation workers due to New Public Management (NPM) implementation reforms.

Performance Measurement and the Erosion of Discretion

It is axiomatic within the street-level bureaucracy literature that discretion is inherent in frontline work (Brodkin, 2011; Sandfort, 2000). However, this characterization is challenged by studies of the impact of some NPM governance reforms, such as performance benchmarking, on decision making at the frontline (Considine et al., 2015; Bredgaard & Larsen, 2007; Brodkin, 2015; Larsen, 2013). These studies suggest that caseworkers’ discretion to flexibly apply program rules has become increasingly bounded by performance measurement systems that have been designed to discipline choices and “to raise the odds that preferred paths will be taken” (Soss et al., 2011b, p. i204). This pressure can be observed in the evolution of the Australian quasi-market for employment services.

The first JN contract operated largely as a “black box” model, with providers free to tailor their service offerings as they saw fit provided they achieved the requisite outcomes. However, this discretion was reigned in under subsequent contracts as more detailed regulations about minimum servicing standards were added in response to mounting concerns about widespread practices of “creaming” and “parking” within the system and evidence that providers were “churning” jobseekers through short-term placements and even “creating phantom jobs” to collect payments (O’Flynn, 2007, p. 4). From the second JN contract, the government introduced a Code of Conduct for providers and a requirement “to record in more detail the actual services delivered” (O’Flynn, 2007, p. 3). The number of guidelines regulating employment services staff dramatically increased over subsequent years, reaching over 3,000 pages under JSA (Australian National Audit Office [ANAO], 2014) and, at the time of writing, including 45 pages of guidelines on managing clients’ mutual obligation requirements (Department of Jobs and Small Business [DJSB], 2018a). This emphasis on prescribing the management of mutual obligations illustrates how the rebureaucratization of the Australian quasi-market (Bredgaard & Larsen, 2007) was driven not just by concerns about “creaming and parking” but also by the government’s desire to ensure conditionality was enforced (pp. 292-293). Over the early 2000s, an increasingly distrustful purchaser moved toward a stronger regulatory regime with greater powers to recover payments from providers deemed to have misused their discretion or breached contractual requirements. One way it did this was through introducing a Star Ratings system, which uses a complicated regression analysis to measure the relative performance of providers. The principal indicator has historically been the proportion of clients who sustain 13 and 26 weeks of employment, although under the Jobactive contract 20% of agencies’ Star Rating also depends on the proportion of clients they commence (and how quickly) in WfD (ANAO, 2017). Under this system, each agency site is awarded a rating from 1 to 5, which then feeds into providers’ overall performance rating at an employment region level. The system was introduced in 2001 following an official review, which found that jobseeker choice was playing only a minimal role in the selection of providers as they had little information about providers’ performance and so found it difficult to exercise choice “in a meaningful way” (O’Flynn, 2007, p. 5). Accordingly, Star Ratings were introduced to drive competition for clients although they quickly became the key mechanism for awarding contracts and periodically (e.g., every 18 months) reallocating business shares from “low” to “high” performing providers. For example, approximately 60% of the providers contracted to deliver the third JN contract were automatically awarded contracts on the basis of their performance, with only providers rated three stars or below required to tender (O’Flynn, 2007).

The elevation of Star Ratings into a major mechanism for awarding contracts and business shares increased the contestability of local business allocations. This had the effect of “making all local practices more important to business viability than before” (Considine et al., 2015, p. 55) such that local office practices became subject to monitoring by not only departmental contract managers but also their own head office. This scrutiny was facilitated by the spread of IT systems, such as the Employment Services Software (ESS) Web interface that frontline workers use to record multiple dimensions of their work, enabling departmental officers and agency managers to review the decisions of individual staff (Marston & McDonald, 2008). In a context “where every transaction is visible” (Fowkes, 2011, p. 8), a “fear of non-compliance” gripped agencies who responded by embracing standardization “as a way to minimise risks” (Considine et al., 2011, pp. 826, 827). This intensification of performance monitoring has not eliminated decision-making discretion entirely. But it has altered the choice-architecture under which frontline workers exercise discretion to reward rule-bound decisions while increasing the risks associated with flexibility and innovation (van Berkel & van Der Aa, 2012, p. 9). Although caseworkers’ discretion continues to be broad in the sense that they are authorized to make a wide number of decisions affecting clients, they make these decisions “as actors who know they are being observed and evaluated” (Soss et al., 2011b, p. i226). The intensity with which performance monitoring is being applied has affected what Brodkin (2011) describes as the “calculus of street-level choice” by motivating patterns of routinization and service rationing as a way of maximizing those aspects of performance to which rewards and penalties are attached (p. i260). In motivating workers to do different things, and not just measuring the efficiency of what workers otherwise already do, performance monitoring systems “reshape agency itself” (Soss et al., 2011b, p. i204) and in ways that risk “exacerbating workfare’s already considerable commodifying tendencies” (Brodkin, 2015, p. 14). For example, in Brodkin’s (2015) study of Chicago welfare offices, performance benchmarking on the basis of client participation in job search training motivated organizational practices of refusing exemptions even for reasons required under law.

Data and Method

The data concerning this relationship between activation workers’ evaluations of welfare recipients and their willingness to report clients for sanctioning and to recommend work-first approaches to job searching are drawn from four surveys of Australian frontline employment services staff (client-facing workers who interact directly with welfare recipients to either make them job-ready or support them to find and sustain employment). The first survey was developed and refined between 1996 and 1998 as part of a study of the impacts of marketization on service delivery. Subsequent surveys—adapted from the original instrument—were repeated in 2008, 2012, and 2016 to follow the long-term impacts of street-level welfare reform in the Australian quasi-market. While the first survey included several hundred respondents from the publicly owned CES as well as newly licensed contracted providers, the respondents to all subsequent surveys were contracted agency staff, reflecting the privatization of Australia’s system in July 2003 (see Table 1).

OPEN IN VIEWER

Table 1.

Survey Respondents.

Year	Government employees	For-profit provider staff	Nonprofit provider staff	Other/unknown	Total
1998	273	118	196	38	625
2008	NA	392	1,056	64	1,512
2012	NA	316	882	66	1,264
2016	NA	393	799	41	1,233
Total	273	1,219	2,933	209	4,634

The survey instrument comprised approximately 100 questions about how frontline staff carry out their work as well as their perceptions of the system and of the clients they work with (for details, see Considine et al., 2015, pp. 171-178). Previously reported findings have addressed aspects of frontline staff decision-making discretion, the practices of creaming and parking, and the workload pressures facing activation workers in Australia from rising caseloads and increasing compliance requirements. In this study, we focus on the responses to several items specifically addressing frontline workers’ attitudes toward welfare recipients and the extent to which they are willing to use sanctions and adopt a “work-first” approach to activation. While findings from the first three surveys on frontline workers’ use of sanctions have been previously reported, this is the first study to report the findings on caseworkers’ attitudes toward the unemployed. Significantly, the intertemporal nature of our data enables us to track how these have shifted since the commencement of privatization in the 1990s and whether this has corresponded with the adoption of a more regular approach to working with clients. The principal measure that we use to assess respondents’ problem figuration of unemployment is the following question: “In your experience, which is more often to blame if a person is on benefits: lack of effort on their part or circumstances beyond their control?” Respondents could answer on a 7-point Likert-type scale ranging from 1 (lack of effort) to 7 (circumstances beyond their control). The question responses were recoded for analysis such that lower scores indicate more negative attitudes toward welfare claimants and higher scores less negative attitudes. Where appropriate, we also report responses to other questions addressing caseworkers’ attitudes toward the unemployed, such as one question asking, “What percentage of people who apply for benefits or an allowance would rather be on benefits than work to support themselves and their families.” However, it is the question about whether respondents attribute being on benefits to recipients’ lack of motivation that best captures the behavioral problem figuration of unemployment underpinning work-related conditionality. After presenting the results about frontline workers’ attitudes toward welfare recipients, we report on the extent to which activation workers are increasingly adopting a work-first approach and using sanctions to “motivate” the unemployed. The question items used for assessing these aspects of the enforcement of work-related conditionality are described in the presentation of the analysis. This is followed by a consideration of the extent to which activation workers who internalize the behavioral problematization of unemployment display an increased willingness to enforce work-related conditionality via work-first job searching and sanctions. Chi-square and Kruskal–Wallis tests were used to detect differences between survey years where the responses were categorical, ordinal, and rank order. In cases of numeric responses or continuous variables, ANOVA was used.

Findings

The attitudes of frontline staff toward welfare recipients, and the extent to which they attribute being on benefits to recipients’ lack of effort, are presented in Table 2. The changes in response between surveys are statistically significant (p < .0001) and show a hardening of attitudes at the frontline. In the 1998 survey, around a quarter of frontline respondents blamed being on benefits on the recipients’ lack of effort, rather than on circumstances beyond their control. By 2008, however, this had risen to 43.5% of respondents, whereas less than 20% perceived that being on benefits was more likely a product of circumstances beyond claimants’ control. The proportion endorsing a voluntaristic rather than more structural problem figuration has declined marginally since but remains high at almost 40% of 2016 survey respondents. This hardening of attitudes is further reflected in respondents’ different estimates of the proportion of people applying for benefits “who would rather be on welfare than work to support themselves and their families.” As Table 3 shows, in 1998, the mean estimate reported was just below 29%, although the high standard deviation indicates that frontline workers’ estimates varied widely. In all subsequent surveys, however, the proportion of recipients who are estimated as preferring being on benefits to employment is closer to 40%.

OPEN IN VIEWER

Table 2.

Frontline Workers’ Problem Figurations of Unemployment.

Year	1Lack of effort	2	3	4	5	6	7Circumstances beyond control
Which is more often to blame if a person is on benefits? (% of respondents)
1998 (n = 247)	3.6	9.7	13.8	31.6	14.2	20.6	6.5
2008 (n = 1,095)	6.8	16.2	20.5	37.5	12.0	4.8	2.2
2012 (n = 897)	9.1	10.4	21.5	39.9	12.3	3.8	3.0
2016 (n = 820)	8.9	10.7	20.1	37.0	14.4	6.3	2.6
χ2(3) = 55.305, p < .0001

OPEN IN VIEWER

Table 3.

Proportion of Claimants Who Frontline Staff Perceive Would Rather Be on Benefits Than Work.

	1998 (n = 242)	2008 (n = 1,101)	2012 (n = 891)	2016 (n = 805)
Mean estimate (%)	28.88	40.95	41.41	38.79
SD	21.15	23.31	22.29	21.62
Welch’s ANOVA	F(3, 990.786) = 24.161		p < .0001

The Spread of Work-Related Conditionality

The enforcement of work-related conditionality via mandatory job-searching, WfD requirements, and sanctions for noncompliance with various forms of behavioral conditionality has become an increasingly central component of activation policy. This is reflected in the responses to several survey items addressing the sanctioning behaviors of frontline staff and their willingness to prioritize “work-first” in low-paid, low-skilled employment over remaining on benefits (see Table 4). The latter is examined via participants’ responses to the following scenario:

After a short time attending your service, an average jobseeker is offered a low-skill, low-paying job that would make him or her better off financially. Assume he or she has two choices: either to take the job and leave welfare or stay on benefits and wait for a better opportunity.

OPEN IN VIEWER

Table 4.

“Work-First” Orientation and Willingness to Report Clients for Sanctioning.

Year		1Take the job	2	3	4	5	6	7Stay on benefits
“After a short time attending your service, an average jobseeker is offered a low-skill, low-paying job that would make him or her better off financially . . . What would your personal advice to a client in this situation be: either to take the job and leave welfare or to stay on benefits and wait for a better opportunity.”(% of respondents)
1998	(n = 249)	49.8	27.3	10.4	4.0	4.4	2.0	2.0
2008	(n = 1,117)	53.9	19.9	11.6	8.3	3.2	2.3	0.7
2012	(n = 910)	63.6	16.3	9.9	6.7	2.2	0.9	0.4
2016	(n = 817)	62.2	18.4	7.6	8.0	2.6	0.9	0.5
“No. of circumstances under which respondents agreed they would typically report clients for noncompliance (% of respondents)”					1-2	3-4	5-6	7+
1998	(n = 575)				25	50.1	15.3	9.6
2008	(n = 985)				6.7	4.5	11.2	77.5
2012	(n = 839)				4.6	3.7	13.7	77.9
2016	(n = 827)				2.3	2.4	8.3	86.8

Respondents were asked what advice they would personally give clients in this situation, ranging from 1 (take the job) to 7 (stay on benefits and wait for a better opportunity). Over time, the proportion of respondents who indicate that they would adopt the strongest work-first orientation in this scenario has risen by 12.4 percentage points from 49.8% of respondents in 1998 to 62.2% in 2016. Conversely, the proportion indicating that they would be more likely to advise clients to remain on benefits has declined from 8.4% in 1998 to 4.0% in 2016. These changes are statistically significant, χ²(3) = 30.958, p < .0001.

Similarly, as Table 4 shows, there has been a sizable increase in respondents’ willingness to report clients for noncompliance under various circumstances, ranging from whether a client is dismissed from a job or training program, to whether a client fails to attend an appointment, to whether a client refuses to apply for a suitable vacancy. Over the survey period, there have been 10 different circumstances that respondents have consistently been asked whether (yes or no) they would typically report a client for noncompliance in such a situation. These circumstances closely relate to the conditions specified under Australia’s Social Security (Administration) Act under which welfare recipients may be penalized for so-called participation failures, including failing to apply for jobs, refusing to accept an offer of suitable employment, voluntarily leaving employment, being dismissed for misconduct, and failing to attend appointments with employment services providers. As such, the survey items on sanctioning are designed to consider the extent to which frontline employment services staff are prepared to enforce legislated conditionality requirements. The responses to these items have been recoded to give a numerical score of the total number of circumstances that frontline staff would report clients for sanctioning in. This score enables us to see how the reach of behavioral conditionality has either narrowed or expanded. For example, in the 1998 survey, most respondents were only willing to report clients for sanctioning in three to four out of 10 circumstances, and fewer than 10% indicated that they would report clients in seven or more circumstances. In all subsequent surveys, however, more than three quarters of respondents have said that they are willing to report clients for sanctioning in at least seven out of the 10 circumstances.

Comparing Responses by Agency Type

Although these data suggest a significant hardening of attitudes toward welfare recipients among activation workers overall, this may conceal important differences between the orientations of frontline staff working under different organizational conditions. According to commentators such as Bredgaard and Larsen (2007), one of the mechanisms through which NPM governance reforms help to substantially shift the content of welfare-to-work policy in the direction of “work-first” is through outsourcing delivery to agents motivated by economic incentives, who are expected to be “tougher in availability assessments, use of sanctions, demands, and other motivational initiatives” (p. 294). This is partially supported by Soss, Fording, and Schram’s (2011a) analysis of sanctioning rates by welfare-to-work program providers in Florida, which found that the odds of receiving a sanction were 25% higher for clients in regions with a for-profit provider “relative to similar clients served by non-profit providers” (p. 218). Accordingly, we considered whether there were any significant differences between the work-first orientation of for-profit, nonprofit, and government agency workers and their attitudes toward those on welfare. To consider this, respondents were grouped by agency type and their responses to the question items on “work-first” orientation and attribution of blame to jobseekers were recoded into rank scores (where a lower rank score indicates a higher propensity to blame welfare recipients and to encourage “work-first” activation). As the Australian quasi-market was entirely privatized in 2003, government agency workers only feature in the 1998 data when, as the data in Table 5 show, we found no significant differences between the degree to which government compared with contracted agency workers blamed jobseekers for a lack of motivation. There was a marginally significant difference in government agency workers’ willingness to recommend “work-first,” but it was in the opposite direction to what many would expect: Government workers were more inclined to recommend “work-first” than either for-profit or nonprofit agency workers. This is consistent with previously reported findings on sanctioning rates in the 1998 survey data, in which government workers were more inclined to sanction jobseekers than either for-profit or nonprofit agency workers (Considine, 1999, p. 194). Accordingly, it is unlikely that the hardening of frontline workers’ attitudes toward the unemployed since the late 1990s can be attributed to the privatization of the Australian quasi-market. Similarly, we found no significant pattern of differences between the attitudes and “work-first” orientations of for-profit compared with nonprofit agency workers. Although, in earlier survey years, nonprofit agency workers appeared marginally less “work-first” than their for-profit agency counterparts, this difference did not persist beyond the 2012 survey when, indeed, nonprofit agency workers were more likely to recommend “work-first.” Furthermore, other than the 2008 survey, there were no significant differences between nonprofit and for-profit agency workers’ perceptions about the blameworthiness of clients. This suggests that the orientations of activation workers in Australia toward the unemployed are relatively stable regardless of agency type, perhaps reflecting the deepening standardization that has occurred as the Australian quasi-market has matured and become more intensively reregulated (see Considine et al., 2015, p. 52).

OPEN IN VIEWER

Table 5.

Attitudes and Orientations of Respondents by Agency Type.

Year	Agency type	Mean rank
. . . Which is more often to blame if a person is on benefits: lack of effort on their part or circumstances beyond their control?
1998	FP agency staff (n = 24)	104.85
	NFP agency staff (n = 64)	112.2
	GOV staff (n = 132)	110.70
	χ2(2) = 0.247, p = .884
2008	FP staff (n = 291)	493.27
	NFP staff (n = 778)	550.61
	χ2(1) = 7.840, p = .005
2012	FP staff (n = 213)	451.42
	NFP staff (n = 647)	423.61
	χ2(1) = 2.176, p = .140
2016	FP staff (n = 246)	396.37
	NFP staff (n = 543)	394.38
	χ2(1) = 0.014, p = .907
After a short time attending your service, an average jobseeker is offered a low-skilled, low-paying job . . . Assume he or she has two choices: Either to take the job and leave welfare or to stay on benefits. What’s your advice?
1998	FP staff (n = 24)	108.06
	NFP staff (n = 63)	126.88
	GOV staff (n = 135)	104.93
	χ2(2) = 6.020, p = .049
2008	FP staff (n = 289)	500.95
	NFP staff (n = 976)	558.27
	χ2(1) = 8.506, p = .004
2012	FP staff (n = 219)	466.09
	NFP staff (n = 650)	424.54
	χ2(1) = 6.076, p = .014
2016	FP staff (n = 247)	392.78
	NFP staff (n = 538)	393.10
	χ2(1) = 0.000, p = .983

Note. FP = for-profit; NFP = nonprofit; GOV = government.

Blaming, Activation, and Discretion

Our study shows two patterns: (a) activation workers’ attitudes toward welfare recipients have hardened considerably since the introduction of marketization in the late 1990s, and (b) there has been a convergence toward the enforcement of “work-first,” particularly through a heightened emphasis on reporting clients for noncompliance. But the extent to which these two developments are linked remains unclear. For example, the circumstances under which sanctions can be applied for various forms of noncompliance are regulated under Section 42 of the Social Security (Administration) Act 1999. The increase in the number of circumstances under which frontline workers reported they were prepared to recommend sanctions may simply reflect legislative changes, although this would not explain the increase observed from 2008 to 2012 when the range of circumstances under which sanctions could be applied was narrowed (see Ramia & Carney, 2010). The association between respondents’ perceptions of welfare recipients and their willingness to enforce work-related conditionality was analyzed by calculating the correlation coefficients between the extent to which respondents blamed recipients for their “lack of effort,” on one hand, and the extent to which they would recommend “work-first” and “report clients for sanctions,” on the other. In addition, given the prominence of frontline workers’ decision-making discretion in the literature on street-level organizations, we further compared these correlations across respondents who reported having high versus low decision-making discretion. This was based on their responses to the following question: “When it comes to day-to-day work, I am free to decide for myself what I will do with each jobseeker.” This is one among several survey questions designed to track the extent to which frontline workers retain leeway to make decisions in their roles, or whether their use of discretion is increasingly disciplined by the spread of performance management systems and a more intensive focus on compliance auditing and outcome measurement by agency and contract managers. Findings to date point toward the latter, and that this narrowing of discretion is associated with “a fear of making wrong decisions” and being penalized by the purchaser (Considine et al., 2015, p. 55).

Calling to mind Bovens and Zouridis’s (2002) concerns about “screen-level bureaucracy” and the impact of “knowledge-management systems and digital decision trees” on workers’ scope for administrative discretion, approximately half of respondents in every survey since 2008 have agreed or strongly agreed that their computer system tells them “what steps to take with clients/jobseekers and when to take them” (p. 177). Over the surveys, the proportion of respondents who disagree or strongly disagree that they are free to decide what to do with jobseekers has nearly quadrupled from 7.1% in 1998 to 26.4% in 2016. Conversely, the proportion who strongly agree that they have administrative discretion in this way has declined from 27.1% in 1998 to 9.8% of in 2016. These findings resonate with Eubanks’s (2018) work on the ubiquity of computer-driven decision making in public assistance programs, and their legacy in redefining social work “as information processing” (p. 195). Insofar as Australian activation workers’ scope for administrative discretion has become more bounded, this suggests decreasing room for frontline workers’ decisions to be influenced by their perceptions of welfare recipients. Indeed, many commentators have interpreted efforts to standardize case management through the development of client classification tools and IT-driven knowledge-management systems as part of a “political struggle” to advance the enforcement of “work-first” through making it harder for activation workers to avoid adopting a “rule-bound and administrative” approach (van Berkel & van Der Aa, 2012, p. 501). In reducing frontline workers’ room for maneuver, Marston (2006) argues that the introduction of electronic assessment and profiling tools in Australia in the early 2000s “made it harder for staff not to apply harsh new financial sanctions and penalties to the unemployed” (p. 91). Larsen (2013) draws a similar conclusion about the introduction of standardized profiling tools in Denmark, arguing that they were essentially “designed to limit the discretion of frontline workers” who were distrusted by central government to advance a work-first approach (p. 114). In the United States, Soss, Fording, and Schram (2011b) argue that the threat of sanctions is one of the few remaining tools that caseworkers have left at their disposal, given the lack of resources in programs to achieve performance targets via alternative means. With little scope for individually working with clients to address their “real life” problems, frustrated caseworkers “turn to the most basic threat they can wield—the sanction—as a way to motivate client compliance” (Soss et al., 2011b, p. i224).

Underpinning this contestation over standardization is the assumption that reducing activation workers’ administrative discretion decreases their capacity to resist enacting work-related conditionality requirements on the ground. Accordingly, to the extent that there is any correlation between workers’ perceptions of welfare recipients and their willingness to enforce work-related conditionality via sanctions and a “work-first” approach, we would expect this correlation to be weaker among those workers who report reduced decision-making autonomy. Conversely, we would expect the correlation to be stronger among those activation workers who report high decision-making discretion. This is not to say that this latter group is either more or less likely to endorse voluntaristic rather than structural explanations of unemployment. Rather, whatever their perceptions of the unemployed, these perceptions are expected to more strongly mediate the decisions they make about sanctioning clients and recommending “work-first” than in the case of frontline workers who report little scope for administrative discretion and who can be expected to follow a more “rule bound” approach regardless of their personal views about clients. We consider this in the analysis reported in Table 6, which shows the strength and significance of these correlations across survey respondents overall and between those who strongly agree (high-discretion group) that they are free to decide what to do with each client and those who disagree or strongly disagree (low-discretion group).

OPEN IN VIEWER

Table 6.

Correlations Between “Blaming Welfare Recipients” and Work-Related Conditionality.

Year	Personal “work-first” orientation	Frequency	Coefficient (ρ)	p value
1998	Total sample	244	0.22	<.0001
	High-discretion group	72	0.15	.223
	Low-discretion group	12	0.24	.454
2008	Total sample	1,084	0.19	<.0001
	High-discretion group	159	0.20	.013
	Low-discretion group	198	0.20	.005
2012	Total sample	896	0.10	.004
	High-discretion group	90	0.27	.01
	Low-discretion group	175	0.08	.297
2016	Total sample	812	0.18	<.0001
	High-discretion group	73	0.25	.032
	Low-discretion group	198	0.15	.042
All years	Total sample	3,036	0.16	<.0001
	High-discretion group	394	0.22	<.0001
	Low-discretion group	583	0.14	.001
Year	Willingness to report clients for sanctioning	Frequency	Coefficient (ρ)	p value
1998	Total sample	222	−0.21	.002
	High-discretion group	69	−0.29	.017
	Low-discretion group	11	−0.09	.804
2008	Total sample	929	−0.12	<.0001
	High-discretion group	145	−0.26	.002
	Low-discretion group	167	−0.10	.220
2012	Total sample	801	−0.07	.04
	High-discretion group	80	−0.19	.09
	Low-discretion group	151	−0.14	.082
2016	Total sample	773	−0.35	.34
	High-discretion group	69	−0.31	.01
	Low-discretion group	190	0.05	.501
All years	Total sample	2,725	−0.10	<.0001
	High-discretion group	363	−0.30	<.001
	Low-discretion group	519	−0.06	.188

Note. Spearman’s rho is reported as the measure of effect size. A negative correlation coefficient is expected between “blaming welfare recipients” and “willingness to sanction” because the response scales are inverse (i.e., a lower value on the “blaming” question indicates a more negative attitude, whereas a higher value on the “willingness to sanction” measure indicates an increased willingness to report clients for noncompliance). The “high-discretion” group is the group of respondents in each survey who strongly agreed that “When it comes to day-to-day work, I am free to decide for myself what to do with each client/jobseeker.” To capture a sufficient number of responses for meaningful statistical comparison, the “low-discretion” group includes respondents who either disagreed or strongly disagreed with this statement.

Considering the overall sample, the data suggest that there is indeed an association between the propensity of frontline workers to blame being on benefits on recipients’ lack of motivation and their willingness to report clients for sanctioning and recommend “work-first.” However, the strength of the association is weaker than the criticisms of the behavioral problem figuration of unemployment suggest, and barely apparent at all (either not statistically significant or a coefficient of less than –.10) in relation to respondents’ willingness to report clients for sanctioning in more recent iterations of the survey. Interpreting the data through the lens of respondents’ decision-making discretion, however, reveals a different picture. In particular, the size of the coefficient between “blaming” and “willingness to report clients for sanctioning” triples from –.10 to –.30 across all survey years when only the responses of those reporting “high discretion” are considered. This correlation is statistically significant and its strength does not diminish over time. Similarly, the magnitude of the correlation coefficient between “blaming” recipients and the personal “work-first” orientation of activation workers increases when we consider only those respondents reporting high discretion—although not to the same extent. Conversely, when only the responses of those reporting “low” decision-making discretion are considered, the correlation between “blaming” and the “personal ‘work-first’ orientation” of respondents remains significant in only two out of the four surveys, while the correlation between “blaming” and “willingness to report clients for sanctioning” loses statistical significance. This is in line with our predictions as well as debates within the literature on the political struggle over street-level workers’ discretion and its implications for the enforcement of a more rule-bound approach to activation. Although, it must be noted that a limitation of our approach is that the survey question only addresses respondents’ self-reported perceptions of, and approaches to working with, clients: not how they actually implement and produce policy at the micro level of “everyday organisational life” (Brodkin, 2011, p. i273). Frontline workers may have minimal discretion over sanctioning decisions in practice. But this does not mean that their self-reported survey responses about the circumstances under which they would normally recommend sanctions should necessarily follow such a rule-bound logic. After all, their survey responses are not subject to managerial oversight in the way that their actual sanctioning decisions are. This makes the observed differences between the “high”- and “low”-discretion groups even more striking. Nonetheless, there are also reasons to believe that participants’ responses provide an accurate reflection of their behaviors. Numerous framing statements were included to prime participants to respond from the perspective of how government policies and contractual conditions affect how they do their jobs. At the start of the survey, participants were informed that “The questions ask you about your experience as an employment services professional.” They were subsequently reminded several times throughout that

Agencies must organise their work based on government policy and the nature of the labor market, among other considerations . . . We are interested in your opinion on how these factors impact on your agency and how you do your job.

Discussion and Conclusion

The data reported in this study on changes in the willingness of Australian frontline employment services staff to sanction jobseekers and to recommend “work-first” activation provide a demonstration of how the reach of work-related forms of welfare conditionality has expanded over recent decades. At a policy level, the intensification of work-related conditionality is frequently underpinned by negative assumptions about welfare recipients’ personal responsibility and motivation to work (Wright, 2012); a problem figuration of unemployment that the data reported in this study indicate is increasingly endorsed by a significant proportion of frontline activation workers in Australia. Although, the reasons why activation workers’ attitudes toward the unemployed have hardened remain somewhat unclear.

One possibility is that it is related to changing labor market conditions and broader social attitudes toward welfare recipients in Australia, which appear to have hardened since the 1990s. For example, in the 2009 Australian Survey of Social Attitudes, approximately two thirds of respondents endorsed the view that “most unemployed people could find a job if they really wanted one” (Schofield & Butterworth, 2015). Ten years previously, just 44% of respondents to a national survey agreed that “reluctance to work” was a cause of high unemployment (Eardley et al., 2000). Whether these attitudinal shifts are connected to changing labor market conditions is difficult to say. In 1998, the unemployment rate averaged around 7.8%, declining to below 4.5% in December 2008 and rising to 5.8% in December 2016 (Australian Bureau of Statistics, 2017). Whether the 3.3% change in the unemployment rate over the decade to 2008 sufficiently explains the 22 percentage point decline in the proportion of frontline respondents who endorsed a structural rather than voluntaristic explanation of unemployment over this same period is unclear. Moreover, the deterioration in economic conditions since 2008 appears to have had little impact on activation workers’ attitudes toward the unemployed, which have softened only marginally since. Substantial policy shifts over the early 2000s may have equally contributed to the hardening of attitudes at the Australian frontline, including the widening of WfD requirements to undertake up to 15 hr per week of compulsory training or unpaid work experience activities for 6 months in every year. Although introduced in 1997, these requirements only initially applied to young people but were gradually extended to all jobseekers aged under 50 years in the early 2000s along with “new prescriptions as to the frequency of meetings and intensity of activity” (Fowkes, 2011, p. 6). As McDonald and Chenoweth (2006) argue, changes in the institutional character of the welfare state “promote a set of norms, expectations and structures” that transform how caseworkers view the nature of client problems (p. 116).

Another important consideration is the organizational and professional context in which frontline workers carry out their jobs. Besides the privatization of Australia’s employments services system, there has been a persistent and deepening trend toward the standardization and de-professionalization of activation work. This is evidenced by declining levels of qualifications among Australian frontline staff over the early 2000s, when growth in the use of IT systems and client categorization instruments replaced “part of the skill set that a case manager might otherwise need” (Considine et al. 2011, p. 821). Caseloads within the Australian system are high, with caseworkers managing an average of almost 148 jobseekers, nearly two thirds of whom have been unemployed for more than 12 months and 20% for more than 5 years (DJSB, 2018b). Among jobseekers in the highest service stream, the average duration in employment services is more than 5 years. In the context of a system characterized by high caseloads, large proportions of very long-term unemployed clients, and limited administrative discretion for frontline workers to exercise professional judgment, blaming the unemployed for their lack of motivation may be a way of coping, or living with the system constraints for workers (Marston, 2013). As Dunn (2013) acknowledges, when considering the predominantly negative framing of client-motivation by Work Programme agency workers, “workers might be inclined to over-emphasise clients’ unwillingness to seek employment in order to deflect attention from their own possible failings in helping them into work” (p. 805). Similarly, as Jordan (2018) argues in relation to the programmatic constraints that hinder Work Programme staff from providing “any real help,” “belief in a pathological underclass” enables workers to accept poor outcomes as “inevitable, and more importantly, to be the fault of the participants and not the staff” (p. 69). Seale, Buck, and Parrotta (2012) similarly interpreted this blame-avoidance strategy as one reason why nearly half the North Carolina welfare-to-work program managers they interviewed “constructed clients as unmotivated or disinterested in doing the work necessary” (p. 51).

While the findings presented show the internalization of stigmatizing welfare discourses by increasing numbers of activation workers, our analysis suggests that these attitudinal shifts have contributed only marginally to the production and intensification of regulatory activation measures at the street level. A limitation of the study, as already acknowledged, is that our survey data do not capture the actual behaviors of frontline workers in their interactions with clients in the way that ethnographies of street-level work do (e.g., Brodkin, 2011; Fuertes & Lindsay, 2015; Maynard-Moody & Musheno, 2000). And it is likely to certainly remain the case, as predicted by many of these studies, that individual caseworkers may go out of their way to help particular clients or to penalize others depending on how they appraise those specific clients as persons or view themselves as (professional) workers. But what the data in this study capture in a way that those ethnographies do not is the relationship between activation workers’ wider evaluations of the unemployed as a social group and their willingness to utilize behavioral forms of conditionality in general. In this context, the study shows that the general turn toward utilizing regulatory activation measures at the frontline of welfare-to-work in Australia has been driven not by activation workers’ internationalization of stigmatizing welfare discourses but by the intensification of systems of performance monitoring and the (double) activation of frontline workers by the government purchaser. This is evidenced by the waning correlation between respondents’ perceptions of the unemployed and their enforcement of work-related conditionality over time, as the more intensive use of performance monitoring systems has standardized frontline practice, and by the absence of any significant correlation between frontline respondents’ views about the blameworthiness of recipients and their willingness to sanction clients among frontline workers reporting low decision-making discretion. While the personal beliefs and moral values of frontline workers have previously been considered important determinants of street-level decision making, Fuertes and Lindsay (2015) emphasize how “the organizational context is crucial in providing the setting for these norms to find expression” (p. 530). The policy settings in Australia, and internationally, have embedded work-related conditionality as a primary pathway from welfare-to-work, while NPM governance reforms have intensified this conditionality through the choice-architecture of contracted agencies and their staff via more intensive performance monitoring. Although these reforms have not eliminated all scope for administrative discretion, they have established the conditions for workers to embrace “routinised discretion” by making the choices “that maximise (or at least satisfice) the goal of meeting benchmarks or procedural requirements” (Brodkin, 2011, p. i259). Our findings expand the results of Fuertes and Lindsay’s detailed case study research into the implementation of the Work Programme in the United Kingdom. Although they found differences among caseworkers in their attitudes to the unemployed, these “appeared to have relatively limited impact on the support offered to clients, which was consistently rooted in a narrow, standardised form of work-first practice” (Fuertes and Lindsay, 2015, p. 535).