Making welfare conditional: A street-level perspective

This article looks at how welfare conditionality is delivered at the street level. It argues that the street-level delivery of welfare conditionality is structured by policies, the governance context in which workers deliver welfare conditionality, the organization in which they work, and the occupation they are part of. Characteristics of these contexts present street-level workers with a variety of signals and incentives that direct their decision making. The article elaborates on this proposition on the basis of a review of academic studies analysing the street-level delivery of various aspects of welfare conditionality: the use of sanctions, service personalization, and the treatment of vulnerable clients. The review shows that context characteristics together have a significant impact on the street-level transformation of welfare conditionality policies into practices. Street-level decision making concerning the use of sanctions is far more complex than can be captured by a perspective on street-level workers as merely policy implementers. Sanctioning practices are sometimes harsher, sometimes more lenient than policies lead us to expect. The “ soft ” side of welfare conditionality — represented by service personalization — is often under pressure at the street level, potentially strengthening welfare conditionality's tough side. This affects vulnerable jobseekers most: Street-level studies show that the balance between disciplining and enabling aspects of welfare-to-work is most at risk for more vulnerable groups. The article concludes that the contextual pressures street-level workers have to deal with in their daily work hardly reflect the “ delicate equilibrium ” that they need to deliver welfare conditionality in a professional, responsive, and responsible way.

context in which workers deliver welfare conditionality, the organization in which they work, and the occupation they are part of. Characteristics of these contexts present street-level workers with a variety of signals and incentives that direct their decision making. The article elaborates on this proposition on the basis of a review of academic studies analysing the street-level delivery of various aspects of welfare conditionality: the use of sanctions, service personalization, and the treatment of vulnerable clients. The review shows that context characteristics together have a significant impact on the street-level transformation of welfare conditionality policies into practices. Street-level decision making concerning the use of sanctions is far more complex than can be captured by a perspective on street-level workers as merely policy implementers. Sanctioning practices are sometimes harsher, sometimes more lenient than policies lead us to expect. The "soft" side of welfare conditionality-represented by service personalization-is often under pressure at the street level, potentially strengthening welfare conditionality's tough side.
This affects vulnerable jobseekers most: Street-level studies show that the balance between disciplining and enabling aspects of welfare-to-work is most at risk for more vulnerable groups. The article concludes that the contextual pressures street-level workers have to deal with in their daily work hardly reflect the "delicate equilibrium" that they need to deliver welfare conditionality in a professional, responsive, and responsible way.

K E Y W O R D S
conditionality, governance, sanctions, street-level work, welfareto-work

| INTRODUCTION
This article analyses the core theme of this special issue, welfare conditionality, from a street-level perspective. It looks at how core aspects of welfare conditionality policies are implemented and transformed into welfare conditionality practices. The article specifically focuses on welfare-to-work or activation policies 1 , as this is an exemplary type of policies where the introduction of welfare conditionality is concerned. Following the argument developed in a recent book on welfare conditionality (Watts & Fitzpatrick, 2018), the form of welfare conditionality this article is interested in concerns conditions of conduct (rather than conditions of status and of need), that is, behavioural conditionality. For it is behavioural conditionality that is most characteristic for the "activation turn" in social policies that made benefit entitlements increasingly conditional upon unemployed welfare claimants' willingness and readiness to behave in ways considered as contributing to their (re-)integration into the labour market.
The starting point of this article is that the practical meaning of welfare conditionality as well as its consequences for benefit recipients cannot be fully grasped without studying how welfare conditionality is produced at the street level of the agencies responsible for its implementation. Political scientists and social policy scholars do recognize that policymaking and policy implementation cannot be clearly separated and that policy implementation is, as Zacka (2017) puts it, suffused by moments of policymaking (also see Lipsky, 1980;Hupe et al., 2015;Pressman & Wildavsky, 1973). Nevertheless, implementation still receives modest attention in political science studies (Zacka, 2017), and studies of the street-level delivery of welfare-to-work occupy a modest position in the vast literature on active welfare-state reforms (Van Berkel, Caswell, Kupka, & Larsen, 2017). As a consequence, much of our knowledge of welfare-to-work policies is based on national or internationally comparative studies of "official," formal national policies regulating the substance of welfare-to-work. The lack of street-level studies is characteristic for welfare conditionality research as well. This can be considered problematic given the role of discretionary decision making at the street level which, as Watts and Fitzpatrick (2018) argue, the introduction of behavioural conditionality is likely to increase. Introducing conditions of conduct implies that street-level workers need to monitor and assess whether or not clients behave according to the required standards of conduct. According to Watts and Fitzpatrick, this also goes for sanctions that, although often defined in national policies, need to be interpreted and applied in local and individual contexts. In addition, it could be argued that the increase of discretionary decision making not only results from the need to police clients in the context of behavioural conditionality. The personalization of the process of setting the behavioural standards clients have to meet is another important source of street-level discretion.
Against the background of the proposition that street-level policy delivery matters in terms of making welfare conditional, the question arises how we can make sense of the street-level decisions and actions through which street-level workers transform welfare conditionality policies into welfare conditionality practices. A core point arising from the literature is that street-level practices are not only shaped and structured by the policies street-level workers are expected to implement. Lipsky (1980) pointed at the important role that conditions of work-including resources, goals, and performance measures-play in street-level decision making. Vinzant and Crothers (1998) discussed the complex environment in which street-level workers work, positioning them in the centre of "a matrix of influences." The need to deal with this environmental complexity is why these authors prefer to talk about streetlevel leaders rather than bureaucrats. Zacka (2017) argues that street-level workers operate in a "moral ecosystem" that confronts them with a "competing array of normative pulls."  analyse street-level workfare practices in terms of various "contexts" structuring these practises, considering policies as one of these contexts.
This literature makes clear that conceptualizing street-level workers as merely "policy implementers" provides us with an analytical framework that is too narrow to understand street-level work and its role in transforming policies into practices (Brodkin, 2013). It points at the importance of studying the impact of various environmental or context characteristics (policies as well as others) on street-level practices more systematically. And finally, it raises questions concerning Vinzant and Crothers' street-level leadership: How do street-level workers make sense of and deal with the competing array of normative pulls that policy delivery processes in a complex environment confront them with?
These insights and research issues concerning the street-level production of policy practices provide an interesting avenue for welfare conditionality research as well. This article contributes to this type of research by exploring and mapping the current state of the art in studies of the street-level production of welfare conditionality in welfareto-work policies. Most basically, it shows that these studies provide solid empirical ground that street-level practices do indeed matter in shaping what welfare conditionality practically means. In addition, the article contributes to welfare conditionality studies by exploring how various characteristics of the environment and context structure and affect the street-level delivery of welfare conditionality. As we will see, this is not a straightforward process: Similar context characteristics may result in divergent street-level practices. So although street-level practices matter, how they matter remains an empirical issue. This is related to what constitutes the final contribution of the article that concerns a core aspect and maybe even the core aspect of street-level practice. Environmental characteristics do not operate in isolation but together shape context "configurations" (cf. Zacka's moral ecosystems) in which street-level workers do their daily work. This article's exploration of the available literature aims to shed light on this core feature of the reality of street-level work. More specifically, it shows that context configurations and the moral ecosystems they reflect result in street-level welfare conditionality practices that redefine the balance between the soft and harsh sides of welfare conditionality and between the disciplining and enabling elements of welfare-to-work.
The article is structured as follows. In the next section, the theoretical approach that is used and the literature review that forms the basis of this article are discussed. Then, the literature review itself is presented. The final section concludes.

| THEORETICAL APPROACH AND METHODS
In the above, it was argued that understanding street-level welfare conditionality practices requires sensitivity to the complex contexts (plural) in which street-level workers deliver policies. This article follows the approach used by Van  that distinguishes four broad contexts: the policy context, the governance context, the organizational context, and the occupational context. Policy context not only refers to laws and regulations defining the substance of welfare conditionality. In addition, it is important to take into account the processes of policy decentralization, which devolve policymaking competences to regions and municipalities in many countries. This implies that in many cases, street-level workers deliver (national, regional, and local) policies rather than (national) policy. The governance context refers to the service provision models through which welfare conditionality is implemented (such as quasi-markets and service networks) and to the management of street-level organizations responsible for policy delivery through, for example, new public management strategies such as performance targets 2 . The organizational context refers to a range of organizational (human resource or personnel) management decisions that shape street-level workers' work and work conditions, such as job design and task descriptions, autonomy, caseloads, and performance management. The occupational context concerns the professional (if any) education and socialization of street-level staff, the dominance (or lack of it) of any specific profession in street-level work, the degree to which workers organize and act collectively through professional or occupational associations, and so on. Of course, this list of contexts is not exhaustive. However, they have received considerable attention in street-level studies and are therefore most suitable to substantiate this article's proposition that context characteristics and configurations matter in making welfare conditional at the street level. The omission of clients as part of street-level workers' environment is probably most striking. Of course, there are many studies focusing on clients' experiences with welfare conditionality and its impact on their lives. But most of these studies investigate welfare conditionality outcomes from clients' point of view rather than analysing clients as part of the context in which street-level workers operate (for examples of the latter, see Eskelinen, Olesen, & Caswell, 2010;Prior & Barnes, 2011).
The mechanism through which these contexts structure street-level practice is by providing street-level workers with meanings, interpretive frameworks, norms, and incentives in doing their work and in working with their clients.
Context characteristics contain information, signals, and clues regarding how street-level workers are expected to look at their clients and their work roles, how to perceive clients' situations and the problems they are confronted with, how to change their behaviours, and so on. Context characteristics contribute to the shaping of what Zacka (2017) calls workers' dispositions or what Eikenaar, De Rijk, and Meershoek (2016) denote as frames of reference.
Many studies have analysed how street-level workers view their work and their clients. For example, Kallio and Karvo (2015) studied street-level workers' perceptions of the deservingness of clients and argued that these have an impact on the treatment of clients. This article does not primarily focus on street-level workers' attitudes and perceptions as individual worker characteristics but rather on how context characteristics structure these attitudes and perceptions.
As was mentioned before, context characteristics structure the work of street-level workers in configurations rather than in isolation. This calls attention to the issue of how context characteristics interrelate in shaping streetlevel practices. This does not render studies of individual context characteristics useless: They help to explore whether or not context characteristics matter in structuring street-level practices in the first place and if so how. In addition, methodologically, the study of context configurations and their impact is challenging, at the very least. Nevertheless, eventually, interrelations between context characteristics matter most: for street-level workers, for policymakers, and for researchers. Context configurations send diverse and often conflicting signals to workers on what to do and how to do it. This could be interpreted as a lack of alignment in shaping the contexts in which policies are delivered and in managing street-level work. However, this interpretation seems to be too simplistic. Even when looking at the policy context only, policies themselves are often "not aligned." As Watts and Fitzgerald (2018) argue for the case of welfare conditionality, policies themselves are full of tensions and contradictions concerning the goals of welfare conditionality, how they should be prioritized in street-level delivery and what instruments should be used. In that sense, it could be argued that an alignment approach amounts to an attempt to reduce street-level discretion-which, as many scholars have argued, is both impossible and undesirable (see, e.g., Lipsky, 1980). Against this background, Zacka (2017) offers an alternative approach. In his approach, the focus is not on putting an end to the competitive array of normative pulls that characterizes street-level work conditions. Instead, Zacka argues that the main challenge for policymakers and managers is to create a "delicate equilibrium" between these pulls. Formulated in terms of discretion, this implies that policymakers and managers should not reduce discretion but create conditions in which street-level workers can use discretion in a sensible, responsible, and professional way.

| Methods
This article is based on a review of literature reporting on studies of the street-level delivery of welfare conditionality. It focuses on three aspects of welfare conditionality. First, it looks at sanctioning practices. In the policy literature and in street-level research, sanctioning has by far received most attention in welfare conditionality studies. Second, it looks at individual action plans. In policy rhetoric, these plans are core instruments in personalizing welfare-towork and welfare conditionality, both in terms of the conditions of conduct expected from clients and in terms of their entitlements to support. Third, the review looks at street-level practices focused on vulnerable, hard to employ groups of clients that are increasingly being objected to welfare conditionality requirements.
The review presented here is not a systematic review. The main reason for this is that the aims of this article are explorative (what do we know about welfare conditionality practices and how these practices are structured by various context characteristics?) rather than an attempt to "test" the impact of specific (configurations of) context characteristics in a more or less robust way, based on results of available studies. The latter would probably be impossible anyway. The number of available welfare conditionality street-level studies is limited, many are themselves explorative and using qualitative and ethnographic rather than quantitative research methods, and the studies come from a variety of welfare states and took place in various periods of welfare conditionality's development, making a meta-analysis of impacts very difficult. For this article, a selection of publications was used that was made for a literature review in the context of a study of street-level welfare-to-work delivery in European countries . The original selection focused on publications on the street-level delivery of welfare-to-work in English academic journals. Relevant publications were found in a large variety of journals, including (but not limited to) social policy, public administration, and social work journals. For this article, a selection was made from these publications, including only articles that paid attention to sanctions, individual action plans, and the hard to employ. This selection was supplemented by recent studies published in the journals from which the original selection was made.

| Sanctions
Sanctions are no doubt the core instrument through which welfare conditionality is imposed upon unemployed people receiving benefits. As such, they are a central feature of the policy context in which street-level workers operate. It is widely recognized that the implementation of sanction policies and their transformation into sanction practices is not an automated process (despite attempts to organize it as an automated process, see below). Here, we analyse sanction practices as practices structured by characteristics of the contexts in which they are produced.
As far as the policy context is concerned, sanction policies introduced in the context of welfare-to-work policies confront street-level workers with the challenge to combine enabling and disciplining elements in service provision.
Specific policy settings affect how this challenge presents itself to workers. For example, in Poland, sanctioning is an option but not allowed in cases where sanctioning would deteriorate the situation of those dependent on the benefit recipient. Because of this, street-level workers often refrain from sanctioning, even in cases where they consider it useful or justified (Kaźmierczak & Rymsza, 2017). U.S. studies reveal a completely different situation. Several streetlevel work studies report that many workers experience that enabling elements are under pressure of sanctioning policies. The study by Soss, Fording, and Schram (2011) showed, for example, that street-level workers sanction their clients not because they consider them effective in terms of the targets they are expected to realize but rather despite the fact that they consider them ineffective. This and other studies show that street-level workers often make an assessment of the effectiveness of sanctions in terms of issues that they consider important for realizing clients' labour-market integration, such as clients' motivation, workers' relationship with clients, and the vulnerability of clients' situations . In doing so, street-level workers look at what is often considered a bureaucratic or administrative issue (the application of regulations concerning sanctioning) through a professional treatment lens (Sainsbury, 2008). Whether or not this evaluation of the effectiveness of sanctioning is decisive in actual sanctioning practices depends, among others, on the degree to which street-level workers feel under pressure to sanction (by policies but also by other context characteristics). Against this background, a comparative study of Australia and Denmark (Marston, Larsen, & McDonald, 2005) is interesting. The authors argue that forms of authority that street-level workers develop in their interactions with clients depend on the policy context in which workers work. They distinguish between empathetic, pedagogical, and punitive authority and found that in Australia, punitive authority is much more widespread than in Denmark. Interestingly, differences between sanctioning practices have also been found in studies comparing municipalities within countries, pointing at the combined impact of national policies, local policies, and organizational conditions in structuring sanctioning practices (Caswell & Larsen, 2017;Nybom, 2011;Van Berkel, 2017).
Performance management, which we see as part of the governance context in which street-level workers deliver services (see footnote 2), is considered another important factor in structuring street-level sanctioning practices (Brodkin, 2011). The U.S. study by Soss et al. (2011) referred to above elaborates how performance management explains why street-level workers resort to sanctioning clients even though they hardly expect sanctions to improve welfare-to-work results. The harshness of the performance management system in combination with a lack of alternative instruments and interventions besides sanctions contributes to this "perverse" effect of performance management. A study in a very different welfare state-Denmark-where alternative forms of client support are readily available and performance management does not include specific organizational or individual performance targets but takes the form of publicly benchmarking sanctioning rates of municipalities, showed that even under these completely different conditions, performance management may have significant consequences for local sanctioning practices as well (Caswell & Larsen, 2017). At the same time, the variety in municipal sanctioning rates in Denmark makes clear that other factors mediate the impact of national performance management systems on actual performance at the street level. Another perspective of studying the governance context in relation to sanctioning was used in an Australian study that compared sanctioning rates by public providers, for-profit providers, and non-profit providers. Surprisingly and counterintuitively, this study showed higher sanctioning rates for non-for-profit than forprofit providers (Considine, 1999). As a possible explanation, Considine points at the reputation of both types of providers: Whereas for-profit providers may be inclined to mitigate their reputation of being tough on clients, non-forprofit providers may be inclined to show that they can be tough on clients. Taken together, these studies show that performance management's impact on street-level sanctioning practices can take different forms, partly depending on other mediating factors, sometimes resulting in stricter and sometimes in more lenient sanctioning practices.
The organizational context is relevant for understanding street-level sanctioning practices, too. Several studies show that lack of resources may have an impact on sanctioning rates. Hasenfeld (2010), for example, argued that lack of time and energy may promote the use of sanctions as a way of ensuring client compliance, as the alternative strategy-building relationships with clients-takes more time than available. The opposite impact of resource scarcity on sanctioning has been observed as well. Sanctioning procedures are laborious, and this may be a reason for street-level workers whose time is scarce to avoid them (Fletcher, 2011;Riccucci, 2002). In a Dutch study, the high caseloads of street-level workers working with people very remote from the labour market compared with their colleagues working with easier to place clients was mentioned as a possible explanation for lower sanctioning rates among the first group of workers who simply have less client contact to monitor client behaviour (Van Berkel, 2017).
Another interesting example of the importance of the organizational context comes from Denmark (Caswell & Larsen, 2017) where one of the organizational measures taken to increase sanctioning rates was to strictly divide the tasks of providing activation support and sanctioning. Because workers responsible for sanctioning do not have direct client contacts (they act upon information they receive from workers providing support), the treatment of sanctioning as a purely administrative issue is encouraged and adopting a professional treatment lens avoided. This points at the impact of core tasks and job design on decision making and the use of discretion. Attempts to reduce workers' role and discretion in sanctioning also took place in France (Lavitry, 2017), by automating imposing sanctions when clients do not attend their monthly meeting with workers. However, this study shows how workers managed to circumvent the automated sanctioning process when their professional treatment perspective told them that imposing a sanction would be unfair. As a final example of how organizational context characteristics structure street-level sanctioning practices, a Belgian study revealed that under conditions of high participation in decision making and support from supervisors and colleagues, street-level workers are more likely to adopt an empowerment approach in their interactions with clients. A more disciplinary approach was found in contexts where workers experience low participation and support (Raeymaeckers & Dierckx, 2013).
Finally, with regard to the occupational context, research has paid some attention to the issue of how the educational background of street-level workers is related to their attitudes towards conditionality and sanctioning. A study in Norway, a country with a strong social work tradition in municipal activation, showed that social workers are somewhat more critical about conditionality than non-social-workers, but differences were quite small (Sadeghi & Fekjaer, 2019). A Dutch study compared several professional groups (rather than social workers with non-socialworkers: Van Berkel & Knies, 2017). This study revealed some differences between these groups, but no clear differences in dealing with sanctions were found. In this context, it is interesting to note that a study of social workers' perceptions of causes of poverty in the Nordic countries showed quite some variation between and within these countries (Blomberg, Kroll, Kallio, & Erola, 2013). These studies raise interesting questions for further research, such as whether it is useful to speak of a social-work perspective on conditionality in the first place. Finally, a Danish study looked at the role of professional work experience rather than educational background in sanctioning practices (Pedersen, Stritch, & Thuesen, 2018). This study showed that professional work experience reduces the role of ethnic stereotyping in street-level decision making on sanctions.

| Individual action plans and personalization
Individual action plans (or similar instruments, such as the claimant commitments in the United Kingdom) and service personalization could to a certain degree be considered as representing the "soft" side of welfare conditionality, at least when we look at the policy rhetoric that accompanies it. Individual action plans are considered tools for provid- The policy context with its increasing emphasis on the obligations of welfare claimants in welfare conditionality is one of the factors that play a role here. It strengthens the street-level focus on the disciplining aspects of welfare-towork and activating clients. In France, for example, monthly review meetings of street-level workers and jobseekers transformed from an instrument of service personalization into a monitoring instrument: a core welfare conditionality technique in assessing whether clients comply with the conduct conditions required from them (Watts & Fitzpatrick, 2018). No-shows at these meetings have become an important reason for sanctioning (Lavitry, 2017).
Other policy developments may shift the balance between supportive and monitoring elements in worker-client interactions as well. Like several other countries, France introduced changes in the definition of suitable employment, requiring more flexibility from jobseekers and reducing their options to refuse job offers. These policy reforms have an impact on the division of power in the "negotiations" between street-level worker and jobseeker, the definition of conduct conditions, and, thus, the nature of the individual action plans (Lavitry, 2017). In a similar way, Toerien, Sainsbury, Drew, and Irvine (2013) argue that policy changes and financial incentives to get people into work are likely to shape what personalization means in practice in the context of welfare-to-work. Besides, more substantial policy reforms countries sometimes introduce more operational rules and regulations that have an impact on personalization. The frequency of meetings provides one example here. Street-level discretion to determine the frequency of meetings between street-level workers and their clients can itself be considered an important aspect of personalization. In Denmark, however, national regulations set standards for the frequency of client meetings, limiting the opportunities for workers to schedule meetings based on clients' needs (Caswell & Larsen, 2017). This tension between personalization and standardization is also visible when we look at other context characteristics.
Apart from the policy emphasis on (quick) labour-market integration mentioned above, labour-market integration and benefit exit are also leading in characteristics of the governance context, especially performance management systems and the performance targets public and private providers and their workers are confronted with. Already in 2006, a U.S. study showed how street-level workers in private provider agencies experienced tensions between personalized case management on the one hand and pressures to provide clients a standard and quick solution on the other (Johnson Dias & Maynard-Moody, 2006). With the increasing importance of performance management in continental European countries, we see growing evidence that performance management shapes personalized service provision in these countries as well. Rice, Fuerta, and Monticelli (2018), for example, found that performance management focuses street-level personalization efforts on clients considered able to contribute to performance targets-which, given the emphasis in performance management on labour-market entry, disadvantages vulnerable jobseekers. A Dutch study reached a quite similar conclusion (Van Berkel & Knies, 2016). Hollertz, Jacobsson, and Seing's (2018) study concluded that performance targets divert street-level workers' attention away from clients' individual needs. At the same time, a German study pointed out that performance targets may motivate street-level workers to invest in relations with clients, in as far as they consider realizing these targets as a process of coproduction of workers and clients (Senghaas, Freier, & Kupka, 2018). Finally, tensions between street-level service personalization and standardization have also been discussed in the context of service provision models. In an Australian study, Considine et al. (2011) showed that quasi-markets did not meet the expectations of more flexible and personalized service provision, arguing that this is the result of both purchaser and provider behaviour in a market context. In the same vein, Lindsay, Pearson, Batty, Cullen, and Eadson (2018) argue that service provision models built on coproduction rather than contractualization and marketization may provide more opportunities and better conditions for the provision of tailor-made services and for a more active involvement and participation of jobseekers.
Organizational context characteristics matter as well in how personalization is enacted in practice. In the late 1990s, Meyers, Glaser, and Donald (1998)

studied the implementation of welfare-to-work reforms in the United
States and concluded that both the traditional focus on eligibility in the implementing agencies and high caseloads rendered the shift from a people processing towards a people changing client approach difficult. It could be argued that recent welfare conditionality reforms run the risk of (again) strengthening the organizational focus on eligibility rather than activation issues, as they potentially direct provider organizations' attention towards monitoring conditions for eligibility. The issues of lack of resources and high caseloads are also major ones when it comes to streetlevel workers' opportunities to personalize services and the functioning of individual action plans (e.g., Kupka & Osiander, 2017;Rice et al., 2018;Van Berkel & Knies, 2016). Finally, there are indications that street-level workers may experience individual action plans as an obligatory administrative duty imposed by their organization rather than as of added value to the service provision process. In these cases, the plans risk being used in a bureaucratic rather than professional way (Kaźmierczak & Rymsza, 2017).
Studies focusing on the role of the occupational context in the street-level practices concerning personalizing welfare conditionality are scarce. Hagelund (2016) argues that street-level workers often have competences regarding how to work with people rather than regarding what to do with them, promoting an inward-looking psychologization of personalized service provision processes. There are also some studies that point out that street-level workers, especially those with a social work education background, may resist elements of personalizing welfare conditionality through individual action plans that they experience as in conflict with their professional ethos. The elements that are specifically criticized are the focus on obligations and disciplinary measures, and the standardization processes jeopardizing needs-oriented personalization (e.g., Kaźmierczak & Rymsza, 2017;Nothdurfter, 2016). Nevertheless, there are also studies that revealed that street-level workers, including social workers, adapt their views and priorities to the dominant policy and managerial discourses (Larsen, 2013;McDonald & Marston, 2008).

| The hard to employ/vulnerable jobseekers
Welfare conditionality policies have seen a gradual expansion of the target groups of these policies to more vulnerable groups, especially the long-term unemployed, people with physical, mental, or psychological work impairments; single parents with young children; and people confronted with more complex problems than "merely" being unemployed. This has not only increased the target group of welfare conditionality significantly but has also made the target group far more diverse. A core issue is how welfare is made conditional for these vulnerable groups at the street level, given their more complex situations, their different needs compared with jobseekers closer to the labour market, and the increasing focus of welfare conditionality on conditions of conduct related to labour-market entry.
This tension in policies between targeting welfare conditionality increasingly at vulnerable groups on the one hand and the stronger orientation on job-finding on the other has presented a major challenge in many countries. As street-level studies looking at these vulnerable groups indicate, other than policy context characteristics often exacerbate rather than mitigate this tension, raising issues regarding the adequacy and successfulness of support for these groups and the fairness of subjecting them to welfare conditionality.
As part of the governance context, performance management systems play a key role in street-level practices regarding vulnerable groups. Studies have shown that (quick) labour-market reintegration and getting unemployed people off benefits are the core performance targets that providers and workers are confronted with. We hardly found studies that report performance targets specifically tailored to more vulnerable groups of jobseekers. An interesting case in this context concerns the United Kingdom, where the Work Programme was accompanied by the introduction of a rather elaborated differential payment system that intended to counteract processes of creaming and parking more vulnerable groups of clients. A study revealed, however, that this differential payment system was unable to direct providers' attention towards vulnerable jobseekers (Rees, Whitworth, & Carter, 2014). U.S. studies in particular have addressed the perverse consequences that performance management may have. These consequences include an organizational and street-level focus on "making the numbers" (Brodkin, 2008) that stimulates selecting clients that are considered most easy to reintegrate into the labour market (Johnson Dias & Maynard-Moody, 2006), and a shift from a clientcentred and needs-oriented towards a target-centred approach in street-level work (Ridzi, 2004). Both consequences can be regarded as potentially jeopardizing adequate service provision to vulnerable groups of clients. Nevertheless, contextualizing analyses of the impact of performance management is important. A Dutch study, for example, showed that a new national funding system for local social assistance expenses-providing financial incentives to municipalities to reduce numbers of social assistance recipients and in that sense a form of performance management-may have contributed to a stronger focus on clients closer to the labour market. At the same time, considerable differences between municipalities were found in results realized for vulnerable clients. This indicates that local or organizational decisions play a mediating role in "translating" national reforms into street-level priorities and practices (Van Berkel, 2017). The role of local and decentral policymaking authority was also illustrated in a study of French "insertion" schemes aimed at vulnerable people. The study showed how local actors were able to resist pressures by central government to introduce principles based on workfare and welfare conditionality, pointing at limitations of central government's power in regulating and steering local practices (Schulte, Greer, Umney, Symon, & Iankova, 2018).
The organizational context is relevant as well in understanding how welfare conditionality practices for vulnerable groups take shape. Here, again, the street-level availability of resources play a key role. Evidently, having the opportunity to provide adequate support to vulnerable groups seems to be an important condition in making welfare conditionality meaningful for these groups. However, street-level workers often lack forms of support specifically tailored to these groups' needs (e.g., Fuertes & Lindsay, 2016). Caseloads are, yet again, another relevant issue.
A German study reported that many provider agencies set specific standards for the frequency of client meetings, using different standards for different client groups. In practice, this means that clients relatively close to the labour market have more frequent meetings than clients with serious problems (Kupka & Osiander, 2017). A Dutch study found a similar pattern: Caseloads of workers working for vulnerable clients were higher than caseloads of workers working for clients closer to the labour market. In addition, street-level workers working for vulnerable clients more often combine providing support with benefit administration, which adds to their workload. One of the consequences of this turns out to be that workers working for the most vulnerable provide services to smaller proportions of their total caseload than other workers do (Van Berkel, 2017). A final issue concerns the ways in which organizations deal with employer-oriented services in attempts to promote clients' labour-market participation. A Norwegian study showed that combining support for vulnerable groups with support for employers and their organizations might have a positive impact on the labour-market inclusion of these groups (Frøyland, 2018), which may mitigate the tension mentioned above between strengthening welfare conditionality for vulnerable groups and focusing conditions of conduct on labour-market reintegration. However, there is also the risk that employer-oriented services decrease forms of support meeting the needs of vulnerable groups. A German study (Sowa, Reims, & Theuer, 2015) and a U.K. study (Ingold, 2018) revealed that street-level workers responsible for employer-oriented services were inclined to prevent vulnerable clients from being introduced to employers as potential candidates for jobs. As Ingold (2018) writes, these workers are mainly concerned with their relationship with employers and are afraid that "bad products"-as vulnerable clients were referred to-jeopardize future opportunities for placing clients.
Finally, when we look at the occupational context, vulnerable groups confront street-level workers with more complex problems and situations compared with jobseekers closer to the labour market. Often, these complex problems and situations are exactly why these jobseekers are vulnerable and more remote from the labour market.
Street-level workers who are trained as social workers are likely to have more experience and expertise in dealing with complex problems than workers without an education in social work. At the same time, the position of social workers in the delivery of welfare conditionality and activation is under pressure in several countries. Policymakers and managers sometimes consider social workers badly equipped for the new realities of welfare support: welfare conditionality and the increasing focus on labour-market integration. This is, for example, the case in some Scandinavian countries (on Denmark, see Caswell & Larsen, 2017), where social workers used to have a strong position in the provision of local welfare services. Irrespective of whether or not social workers have more competences and skills to deal with vulnerable groups, it seems evident that supporting vulnerable groups does require different competences than supporting clients close to the labour market. A German study (discussed in Kupka & Osiander, 2017) revealed that street-level workers do not always have the expertise and communication skills to adequately cope with the multitude and complexity of problems vulnerable groups are confronted with. This has a negative impact on the quality of casework they are able to deliver.
Summarizing, the studies discussed above present a picture where the increasing behavioural requirements that vulnerable groups of jobseekers are confronted with are, due to a variety of contextual conditions, not accompanied by an increasing provision of adequate services for these jobseekers. In fact, often an opposite development takes place: Services are increasingly aimed at (quick) labour-market integration. This obviously challenges the balance between the conduct that is expected from vulnerable jobseekers and the nature and level of support they receive in order to meet these expectations.

| CONCLUSION
Although, as discussed before, the diversity and nature of available street-level studies makes drawing robust conclusion problematic, some tentative conclusions can be drawn on the basis of our more explorative effort. Studies of the street-level delivery of welfare conditionality provide sufficient evidence that we should indeed be careful in using formal, official policies as proxies for policy practices. Partly, this is because welfare conditionality policies and how they relate to welfare-to-work are themselves ambiguous and sometimes contradictory. For example, sanctioning practices at the street level will look different when the focus is on punishing rule violations by clients or when sanctions are one of the instruments that can be used to integrate jobseekers into the labour market-welfare conditionality policies often use both types of arguments for sanctioning (cf. Watts & Fitzpatrick, 2018). More importantly, for the argument in this article, other contexts matter as well. At times, they may even leave a stronger mark on street-level practices than policies themselves. Each of the various aspects of welfare conditionality that we looked at provides illustrations of this point. We saw that making welfare conditional for hard to employ clients may in practice have limited impact when performance management directs organizations' and workers' attention to those most easy to reintegrate, or when workers lack the resources (professional knowledge, services, time, etc.) to offer these clients the services they need. We also saw that performance management, performance targets, and lack of resources may effectively emphasize the "harsh" side of welfare conditionality and erode what we called the soft side of welfare conditionality: service personalization and the involvement of clients in developing individual action plans. In addition, the impact of context characteristics on street-level practices is far from unequivocal. We saw how sanctioning practices may become more lenient or stricter as a consequence of governance contexts (e.g., performance management systems), organizational contexts (e.g., task design or caseloads), and occupational contexts (e.g., the professional lens street-level workers use to evaluate the impact of sanctions). A similar point can be made regarding the impact of performance targets. Although there are numerous studies emphasizing the negative consequences of performance targets, there are also studies indicating positive effects, for example, in terms of promoting collaborating with clients in order to realize targets. These conclusions also lead to two broader observations concerning street-level research. On the one hand, they highlight an issue already discussed in Pressman and Wildavsky's (1973) famous book: the importance of avoiding the extremes of "policy nihilism" and "policy determinism" in implementation research. On the other hand, they show that street-level agency is not simply an issue of workers' individual preferences, characteristics, or policy loyalties. Stricter than prescribed conditionality and sanctioning practices do not necessarily reflect harsh attitudes of workers towards their clients. Lenient conditionality and sanctioning practices do not automatically point at workers' resistance towards sanctioning. Of course, streetlevel workers' preferences and attitudes play a role: denying that would amount to a denial of street-level workers' agency. But when we no longer view street-level workers as policy implementers only and broaden our focus to include the governance, organizational and occupational contexts in which they deliver policies, a far more nuanced picture of street-level work and the role of street-level workers arises.
Another conclusion of this article refers to the theoretical argument that in the end, context configurations rather than separate context factors matter in structuring street-level practices. Even though these configurations have hardly been studied in a systematic way, many of the studies discussed in this article implicitly or explicitly address a variety of context characteristics and their interrelationships in studying street-level practices. This makes the study of these configurations and their role in structuring street-level practices an important topic for future research. A relevant issue in this type of research will be the theoretical framework adopted in the analysis of context configurations. Zacka's (2017) approach to study context configurations in terms of the degree to which they provide street-level workers with a "delicate equilibrium" between normative pulls in their work promises to be a fruitful theoretical lens. Many of the findings of street-level studies discussed in this article can be interpreted in terms of the lack of equilibrium at the street level. When welfare conditionality policy that pays lip service to its soft aspects only is combined with performance management systems that emphasize quick benefit exit and labour-market integration and with a lack of resources for street-level workers to provide personalized services, street-level workers will experience a context configuration in which they feel little encouragement to use sanctions in a subtle way or provide tailor-made services to vulnerable groups. Theoretically, a more delicate equilibrium can be realized through interventions in all contexts relevant for street-level practices. Obviously, this has immediate practical relevance: successful and effective policies require attention to and investments in the governance structures, organizations, and occupations that structure the ways in which policies are transformed into practices. Zacka's approach provides an evaluation criterion for analysing these investments by focusing attention on how these investments influence the balance between normative pulls street-level workers experience in their work.
To conclude this article, it is appropriate to point out that the general argument developed here is not newcontemporary street-level work and policy implementation scholars owe much to the pioneering work of authors such as Pressman and Wildavsky (1973), Lipsky (1980), and others that were published decades ago. At the same time, the recent revival of this type of studies 3 makes clear that the topic of street-level policy delivery is still highly relevant and undervalued, by policy researchers but certainly by policymakers.

CONFLICT OF INTEREST
The author has no conflict of interest to declare.