One Size Doesn’t Fit All: Diversifying research on active labor market policies

Sophie Danneris, Aalborg University, Department of Sociology and Social Work

Over several decades, there has accumulated a large body of research on active labor market policies (ALMP). In particular, numerous programs designed to encourage welfare recipients to move into work and off benefits have been evaluated in western countries such as Denmark. The results of the research have been selectively influential in shaping policy developments, but as this paper will demonstrate, there are several biases attached to this research. Based on an extensive mapping review of both Danish and international literature, this paper contributes with a systematic overview of the current research within the field. By focusing on the range of these studies (in design, populations covered and substantive issues addressed), the review looks beyond the specific results of each study, and illuminates substantial gaps in the knowledge produced. The analysis shows that the research is lacking sensitivity toward the complexity of the field, in particular regarding the diversity of the clients being served and the variety of challenges they confront.

1 The political frame – from welfare to workfare

Throughout the Western world labor market policies have undergone significant changes during the past two decades. These changes amount to a liquid transition, a silent change of tracks, from a social welfare perspective to a workfare perspective, where the primacy of work has been promoted while the provision of welfare has been increasingly limited and made conditional on making the transition from welfare to work (Torfing 2004; Bonoli 2010; Brodkin and Larsen 2013). This transition to a ‘workfare’ or ‘activating’ state has involved a combination of enforcing labor market participation, emphasizing the obligations of the individual, and increasing services in order to promote the employability of the individual (Dingeldey 2007). Despite the common goal of policies in the different countries in terms of reducing the number of people on welfare benefits, they have not evolved along a single path nor adopted a uniform approach when comparing ALMP (e.g., the US provides less employment services). Thus, the continuing policy changes within this field operate in various ways and under various labels (e.g., welfare-to-workfare, work-first, active labor market policies, activation, welfare reform) and the transition toward this plurality of policies has been bumpy and non-linear (Bonoli 2010; Brodkin and Larsen 2013). Although there is a common transition from welfare to workfare, there is wide variation in specific policies and programs across countries. Further, a number of countries have substantial variation in approaches across regionals and localities. Perhaps the most dramatic variation is between the U.S. and Scandinavian countries (recognizing their own internal variation). But other variations are worthy of note. Bonoli distinguish between four different orientations in ALMP based on the extent to which market-driven employment policies are emphasized and the extent to which the programs are based on investing in client’s human capital (Bonoli 2010). Further, due consideration should also be given to how the shift in these policies has changed over time. For instance, some European countries have undergone a movement from policies aiming at supporting and developing the well-being of an active workforce, toward more regulatory policies like the ones found in United States, where significant requirements have been imposed on the individual in order to receive benefits (Schram 2006; Brodkin and Larsen 2013).

Activation must also be understood as being part of a broader policy shift that changes the relationship between the citizen and the state. Now citizens’ entitlement rights are no longer un-conditional but are contingent upon their fulfilling responsibilities in exchange for rights. This basic philosophical change have profound implications and consequences for the unemployed workers as a whole, and especially for the individuals who are disadvantaged by market arrangements and thus confronted with increasing risks in an economic landscape that features worker flexibility, job instability, and diminished job-related benefits (Soss, Fording and Schram 2011; Brodkin and Larsen 2013). Activation has come to imply new forms of accountability and obligations as regards the individuals’ role in improving their employability and making due with what jobs offer them. Last, the New Public Management has popularized the use of Performance Measurement schemes, where, among other things, the documentation of program outcomes heightens pressure to move recipients into jobs (even those without full benefits) as quickly as possible (for further elaboration on this see for example de Bruijn 2007).

At the same time, ALMP approaches have led to development of more client-centered services designed to increase the chances that each client will get what he or she needs in order to achieve successful activation. Governments have increasingly focused on tailoring their treatment of individual clients so that they can be activated effectively. Personalization has become a buzz-word used by governments, service providers and international bodies in the current activation paradigm (van Berkel & Valkenburg 2007). Personalization is understood as the tailoring of services to account for individuals’ needs while applying a holistic approach that takes the whole person and his or her needs into account. Personalization creates the foundation for insisting some degree of user participation and co-production in their transition to work—given that the services have been tailored to the individual (Fuertes & McQuaid 2014). The implementation of the personalization principle appears as a growing necessity since the greater number of individuals often with multiple and complex problems are the targets of activation policies. Thus a great variety of instruments (for example training, education, personal contracts, sanctions and counseling) are deployed to treat a broader variety of unemployed. At the same time, there are increased linkages between formerly separated policy sectors such as health, social assistance, immigration, vocational training, and childcare (Heidenreich & Aurich-Beerheide 2014).

Both the increased number of individuals to be activated and the increased compulsion upon them increases the pressure to develop nuanced ALMP services that can be effective with diverse populations. But, as the analysis in this paper will show, the problems of the unemployed, the variety of ALMP approaches they are subject to, and the uneven outcomes of the policies studied, all indicate the ingredients for successful activation remain ambiguous and perhaps in far too many cases actually unlikely. A critical examination of the literature on activation suggests the need to reframe how we look at activation. Thus, the recurring and urgent questions to be answered within the research in this field address the effects of ALMP, the extent to which the workfare approach has in fact improved the employability of the individuals, what programs are most effective and so forth. As will be illuminated in this paper, numerous programs designed to encourage welfare recipients to move into work and off benefit have been evaluated across countries. I will give emphasis to whether personalization, as in the tailoring of activation-related services, has in fact gone far enough or whether the literature indicates that activation is still too much the “one size fits all” approach it should not be.

This review employs a mapping review technique specially constructed for grasping the extent, distribution and methodological quality of research within a specific area, the body of research relevant to answering these question, will be analyzed. In the first section, a detailed description of the methodological approach behind the mapping review will be presented, followed by an initial descriptive analysis of the body of research. Then, the paper is divided into different analytical sections each thematically addressing the patterns in the studies and thus identifying possible knowledge gaps within the research. Throughout the analysis, descriptions of the studies are included as examples to illustrate the analytical findings. Finally a case from the Danish ALMP will be presented with the purpose of demonstrating the consequences of the available knowledge, arguing, that what is known has been selectively influential in shaping policy developments but only without a thorough understanding of the biases and gaps present within the research.

2 Mapping review

The paper is based on a systematic mapping review in the field of social policy and with a specific focus on active labor market policy and unemployment studies. A mapping review allows an understanding of the extent, distribution and methodological quality of research within a specific area (Booth et al. 2012, Hetrick et al. 2010). The aim of the review was thus to identify what is currently know about the consequences of ALMP in the population of welfare recipients and what gaps are present in the literature that requires further research. The research question structuring the mapping process was; how is ALMP studied and to what extent does the research reflect the variety of approaches for treating the different challenges confronting the diverse population that make up the unemployed? On this basis, the mapping of evidence aimed at providing a clear picture of the nature of the available research, a reliable estimate of the reach of the studies and indicating the lack of needed research in the field. At this early point, it is important to stress, that the aim was not to provide a full picture of the results of the studies and thus striving to make recommendations as to which programs are most effective and deliver the best results. On the contrary, the purpose was to look behind and beyond the specific results of each study by shedding light on the subjects being investigated, the methodological and analytical approaches employed, and the philosophy of science underlying the research. By adapting this approach the mapping review illuminates the different types of knowledge present within this field, which is pivotal if the knowledge base is to be developed in a direction that takes into account the diversity and complex nature of the field when filling in the gaps.

The starting point of the search process was the identification of central key terms capturing both the population and the different types of programs and activities used in the ALMP. To ensure an exhaustive list of keywords a number of central studies were reviewed, which resulted in the identification of seven synonyms for the target group and twelve concepts capturing different employment services, programs and interventions. Regarding the target group, the synonyms included both unemployed in general and variations of specific types of unemployed – e.g. social benefit recipients and vulnerable clients. Regarding the different ALMP both content and the extent to which each activity is used vary across countries and also between regions or states within one country. Therefore a large variety of different outcome measures focusing on the employability of the population were included. This also being softer outcomes as improvement of work-life balance, social skills etc.

The scope of the review was on Danish as well as international research (primarily from European countries and United States) with a timeframe from 2004 onwards, although research prior to this was included in a smaller extent. Additionally, no hierarchy of evidence was established. On the contrary, a differentiation in methodology was strived. Thus, all relevant studies within the scope of the review were included; both qualitative studies (e.g. in-depth interviews, observations etc.), quantitative studies like statistical analysis of administrative data as well as mixed method studies.

Regarding the strategy of the search process, the review undertook several different approaches in identifying the evidence including comprehensive database searching, contact with experts, reference list checking and relevant sources of grey literature (Booth et al. 2012:83-84).

The database search was performed in eleven selected databases, including EbscoHost, Proquest and PubMed, which were assessed relevant to the research question and subject of the article. When performing the search, limitations regarding document type, language, source type and geography was applied – see table 1. Further limitations not included in the table were subject and category limitations, which were applied when relevant in the specific database. For instance, research within biochemistry genetics and biology were removed from the search string in the Proquest database.

Table 1: Constraints in database search

Type	Included	Excluded
Document type		News, commentary, conference paper, general information, editorial, interviews, speech/lecture and meeting abstracts
Language	English, Danish, Norwegian, Swedish
Source type		Newspapers, letter, short survey and magazines
Geography	All European countries, Australia, New Zealand, Canada and United States

Applying these constraints serves to narrow the scope and ensure that only relevant and qualified studies are included, but naturally it could potentially exclude some relevant research as well. In this case, literature not meeting standard academic criteria – for instance speeches or commentaries in magazines – would not reflect the purpose of the review and was therefore excluded. Regarding the exclusion of languages not English or Scandinavian, this was a pragmatic choice, which might have cut off research that met all other criteria. But, when looking at the review protocol, only a limited amount of studies were in fact in different languages. Further, as regards the geography, the constraints applied in this category are also related to a pragmatic question, but just as well, not many studies of ALMP has been made outside these countries, leaving only a small possible bias to be taken into account.

The studies remaining after applying the described conditions were qualitatively assessed by title, excluding the ones obviously not relevant. This resulted in a total of 427 references (when duplicates were removed). After exporting the 427 references to RefWorks, they were all reviewed on abstract level, which resulted in a further exclusion of 175 references due to irrelevant purpose, e.g., policy analysis of employment policies during the transition from welfare to workfare. On this basis a total of 252 references from the database search was included in the mapping review.

Of the supplementing search strategies, experts consulting, reference list checking and grey literature search added extra fourteen references. Additional grey literature including legislation documents and unpublished reports have been use as background knowledge and was not included in the mapping review. The final number of references of 266 was exported to the reference management software Mendeley from which the mapping review was carried out.

3 Descriptive analysis

By identifying, evaluating and synthesizing the existing body of research revealed in the literature search, the mapping process provided a systematic and explicit review of current research within the field of ALMP. As illustrated in the figure below, each study was coded into different categories and subcategories relevant to the scope of the article. The categories were established during the review process, this formed by the distinctions that initially seemed relevant when working through the material. Each category represents the assumptions behind the research question and following, the distinctions relevant to draw in order to illuminate these assumptions. Additionally, the categories aimed at covering the most relevant variations between the studies, so comparisons were made possible. The categories were not fixed and in several cases revised during the process. For example the outcome category initially consisted of the two categories ‘outcome focus’ or ‘no outcome focus’, but after reviewing the first 20 studies, it seemed relevant instead to distinguish between ‘hard outcomes’ operationalized as statistical measures of e.g. employment effects and ‘soft outcomes’ like improving social skills or self-perception (often measured qualitatively by interviewing the clients). Thus this category ended up focusing on the nature of the outcomes instead of whether there was an outcome focus or not. As a consequence, the first 20 studies were re-categorized into the two new sub categories.

Figure 1: Review strategy – coding categories

Intervention	· Clear defined · Unclear/undefined
Method	· Mixed Methods · Qualitative · Quantitative
Outcomes	· Hard Outcomes · Soft Outcomes
Target group	· Clients with problems besides unemployment · Clients with no problems besides unemployment · No clear target group
Welfare State Model	· Selective · Universal · Residual

Each of the final 266 studies were reviewed and coded into one subcategory in each of the five categories. The information relevant in order to do the coding was primarily found the methodology section (though not all studies did in fact have such a section) and in the analytical description of the types of outcomes found in the studies, thereby purposely ignoring the sections framing the theoretical perspectives and the recommendations as to ‘what works’ and would should be done in order to improve ALMP.

Across all studies included in the review, the overall picture is clear in terms of the methods being used, the type of outcomes addressed and the target group being studied. Thus, the majority of the studies use a variation of quantitative methods focusing on hard outcomes such as the employment success and for an unclear target group defined merely as ‘unemployed workers’. As regards the remaining two categories the pattern is muddier, revealing a greater variety in the type of interventions studied and the welfare state model under which the research has taken place. 

Looking closer at each of the above categories, the category coding each study as either having a clear and defined intervention (for example evaluations of a specific activation program) or an unclear and undefined intervention was fairly evenly distributed. Thus, no clear picture emerged in the body of research as to whether the studies primarily focus on specific programs/interventions or for instance implementation processes or life course studies. As goes for the methods applied in the studies, it is a different story. More than two-thirds have a quantitative methodology, leaving only a small amount of qualitative research and even fewer mixed methods studies. This tendency is closely connected to the next category and the question of hard or soft outcomes being in focus, as quantitative studies with a focus on soft outcomes are rarely the case. Therefore, it is no surprise, that the vast majority (more than two-thirds) of the research is concerned with hard outcome measures like the percentage being on benefits three months after completing a program. Regarding the target group, the distribution of the studies between the three subcategories is somewhat unclear. There are less than ten studies focusing directly on the ‘strong’ group of unemployed with little or no problems besides unemployment. On the other hand, there are several studies having vulnerable clients with extensive problems besides unemployment as their target group. But, close to 75 percent were coded in the last subcategory, where the target group is not clearly defined, besides being people without a job. The final category, differentiating three types of welfare state model, reveals an equal distribution between the selective model (Central Europe) and the residual model (United States and United Kingdom) whereas only around 20 percent originate from a universal model as found in Scandinavia. There is of course the obvious explanation of number and size of countries to be considered, as the Nordic countries are outnumber several times compared to the other two categories.

Looking beyond the five categories, there are also some more or less obvious patterns regarding the subject and cases being studied. The most consistent picture is the predominance of research studying the outcomes of specific, local or national activities like mentoring, job search assistance or counseling. Further, several studies look into threat effects, the outcomes of mandatory activation programs and sanctioning unemployed. Another pattern is the many studies concerned with health implications involved with being unemployed. Interestingly, much of this research shows how being unemployed affects individual health instead of the other way around. Finally, several studies are concerned with the economical dimension of unemployed living on benefits, for instance the significance of high or low benefits or the economic consequences of long-term unemployment.

Concluding on this initial descriptive analysis I find, that this field of research is an area with a great amount of evidence, but with substantial gaps regarding the target group of the research, the methods applied and outcomes being analyzed. In the following sections, these three subjects will be analyzed. Each section departs from the categorization of the research under which individual studies will be highlighted in order to exemplify the analytical findings. Thus, the highlighted studies are selected as they show the general tendencies, though some studies are also described because they are exceptions from the overall picture. Further examples to support the analysis are listed as references in the end of the paragraphs. A full search protocol is omitted from the paper but can be required by request.

4 The unemployed workers

Unemployed people are by no means a homogeneous group of individuals without a job. Besides the obvious variations in personal characteristics like gender, age, unemployment history and educational background, they also differ in less obvious, but nonetheless important ways when it comes to studying the consequences of ALMP for their employability. Just to mention a few other differences, clients vary widely in their mental and physical status, their social and personal skills as well as the extent of their job-related network. Thus, there is a large group of people being unemployed for reasons that goes beyond economic sluggishness that produces high unemployment rates. These additional factors need to be taken into account when addressing how to best serve the unemployed as a whole. During the review process three categories, reflecting the various target group definitions in the body of research, were developed. As illustrated in figure 1 the categories are; 1) clients with several and/or severe problems besides unemployment, 2) clients with little or no problems besides unemployment and 3) clients with no clear characteristics besides unemployment.

Predominantly in the studies within the first category of clients with several and/or severe problems besides unemployment is a focus on problems like alcohol dependency, psychiatric disabilities, homelessness and poverty. For example Bauld et al. (2013) examine alcohol misuse as a barrier to accessing and sustaining employment. The study involved a systematic literature review followed by qualitative research with 53 clients and 12 professionals from sites in England, Scotland and Wales. The findings focus on individual and organizational facilitators and barriers for returning to work. Some of the facilitators and barriers identified were similar to those found in research for other groups of unemployed adults while others were more specific to alcohol misusers (Bauld et al. 2013).

Besides the common tendency of focusing on the individual barriers for employment, it is hard to find many additional similarities between the studies in this category. Thus, some studies have a strong health perspective – looking for example at the health status among sick and disabled compared to employed workers (Herbig, Dragano, and Angerer 2013) while other studies focus on recidivism among former welfare recipients (Cheng 2005) and others again study the geographical mobility of unemployed workers (Arntz 2008). Nor are there any commonalities regarding methodology. In particular, one glaring omission of political significance is that very few studies adopt a bottom-up perspective that takes client perspectives as the point of departure.

As regards the second category of clients with little or no problems besides unemployment, there are less than 10 studies available using this definition and they are very similar in terms of the research questions and the applied methodology. The target group in these studies is primarily insurance benefit recipients – for example newly graduated people trying to find their first job or experienced workers who have been laid off because of cut downs. All of the studies have a clear outcome focus – looking into either the outcomes of a specific training program or the effects of reducing benefits, limiting the eligibility period or implementing mandatory programs. For example Graversen & van Ours (2009) focus on short-term effects of activation, showing among other things, that unemployed who were assigned to activation programs found a job more quickly than unemployed not assigned to an activation program. On this basis, Graversen & van Ours conclude that activation programs mainly work because they are compulsory and unemployed don’t like them (Graversen and van Ours 2009). Further, all of the studies in this category are based on a quantitative methodology, using primarily administrative data. This is the case with Graversen & van Ours, using an experimental setting to answer their research question, and for Gaure, Roed & Westlie as well, who examine the impacts of time-limited unemployment insurance on the duration and outcome of job search in Norway by using a comprehensive simultaneous equation model (Gaure, Roed, and Westlie 2012).

Comparing these first two categories with the third, there are more than twice as many studies in this last category of clients with no clear characteristics besides unemployment. The fact that the bulk of studies fall into this category is troubling for a number of reasons, two in particular deserve special note. First, this group is not the most difficult to serve given their problems are ostensibly limited to unemployment alone. The need for more research on more difficult cases would seem to be critical if ALMP programs are to live up their credo to tailor services to the client. Second, and perhaps more troubling is that this catch-all group is in all likelihood often more diverse than it is made out to be. In fact it may include many people with multiple challenges that are masked by being lumped into the simply unemployed category. Actually, the group of people being studied in this research category is very difficult to gain detailed knowledge about as they hide behind general and overall reaching categories as ‘welfare-to-work participants’, ‘unemployed workers’ or ‘benefit recipients’. In most of these studies objective criteria and randomized selection processes from administrative registers define the population.

For example, Crepon et.al. (2009) examine a nationally representative sample of all unemployed workers in France since 1990, drawing a random 2,5 pct. sample each quarter of 2007. From this sample the outcomes of training programs is studied by observing each individual’s unemployment duration in relation to their participation in a training program. The analysis includes a number of covariates including age, gender, former occupation, region, duration of affiliation to the unemployment system and wage in previous job, but does not take into account variables not directly observable in the national register. Thus, the population is all unemployed workers in France in the given time period, regardless of whatever problems the individual might have in addition to their unemployment. The study concludes that training has little impact on unemployment duration (Crépon et al. 2009).

Another example of studies in this category, where the inclusion of socio demographic variables apparently contributes with detailed information about the group of people being studies, is Lechner & Gerfin’s (2004) evaluation of ALMP in Switzerland. They include further covariates like language skills and type of work permit as well as the number of sanctions imposed by the case manager. The result of the evaluation is substantial positive effects for one particular program consisting of a wage subsidy for temporary jobs in the regular labor market that would otherwise not be taken up by the unemployed. Further, the evaluation finds large negative effects for traditional employment programs operated in sheltered labor markets while the results for training courses are mixed (Lechner and Gerfin 2004). As was the case for the Crépon et al. study, this socio-demographic information does not include knowledge about social problems, health barriers etc. that might have a great influence on the outcomes of these programs, depending on the client population in question.

Due to the use of general unemployment categories and limited inclusion of softer characteristics within the target group it is, from this body of research, impossible to conclude whether the people being studied are in fact especially vulnerable clients or simply unemployed workers with no problems beside unemployment. As a result, the analyses offered often fall into the trap of using a one-size-fits-all understanding of unemployed in terms of the way they are all addresses as part of the same homogeneous group of people. The resulting opacity is problematic for policy makers who remain unaware of the differentiation within the target group. Further, the lack of distinction between different types of unemployed makes it difficult to assess the validity of the research – if the research object was in fact what was intended to be studied. Thus, although some welfare to work programs may be appropriate for those who are more or less job ready, it is less clear whether they are suitable for hard-to-help groups with multiple disadvantages. Taken this notion into account, the next section will address the instruments and programs which this group ‘unemployed workers’ are subject to.

5 An empty shell: the understanding of instruments, measures and programs

When studying ALMP the term policies or programs intended to improve the employability of the unemployed, has little consistency and is defined in numerous ways. And, as the following will illuminate, often no definition is in fact included in the research, which makes the applicability difficult to assess. In the mapping review, two sub categories were developed (clear/defined interventions and unclear/undefined interventions), constituting the starting point of the analysis.

In the first of the two sub-categories where a little more than 60 percent of the studies are placed, is the research assessed to have a somewhat clear and/or defined intervention. A typical example is research on benefit sanctions and how effective sanctions are as a means of getting unemployed off benefits (Arni, Lalive, and van Ours 2009; Berg, Klaauw, and van Ours 2004; Bong, Slack, and Lewis 2004; Boone and van Ours 2006; Butler et al. 2012; Fording, Schram, and Soss 2013; Svarer 2011) Amongst others Arni et al. has done a comprehensive evaluation on temporary reductions in unemployment benefits as punishment for noncompliance with eligibility requirements. Besides looking at effects in terms of unemployment durations, the study also evaluates the effects on post-unemployment stability, on exits from the labor market and on earnings.

In the analysis Arni et al. uses administrative data from a Swiss register which makes it possible to distinguish between ex ante effects, the effects of warnings and the effects of enforcements of benefits. The results of the evaluation point in the direction that both warnings and enforcement increase the job finding rate and the exit rate out of the labor force. Further, the evaluation shows that warnings do not affect subsequent employment stability but do reduce post-unemployment earnings. Actual benefit reductions lower the quality of post-unemployment jobs both in terms of job duration as well as in terms of earnings. On this basis Arni et al. concludes that the net effect of a benefit sanction on post-unemployment income is negative (Arni, Lalive, and van Ours 2009).

This last finding points to the importance of moving beyond studies that rely on linear causality, where the intervention, in this case benefit sanctions, is studied in terms of their effects on reducing the number of unemployed without considering whether the intervention complicates the challenges the client is struggling to address. What is not included in evaluations like these are aspects like the consequences of sanctions on an individual level, for whom in particular the sanctions are effective and in what way. As result, the circular dynamic of sanctioning is missed where sanctions create additional hardships that make employability that much more difficult to achieve.

These sorts of complications are also apparent in the case in a study by Baumgartner & Caliendo, where the effectiveness of two start-up programs for unemployed are evaluated. This analysis focuses on the probability of being employed, and the results showed that at the end of the observation period, both start-up programs were effective and one was also efficient (Baumgartner and Caliendo 2008). The challenge with this kind of research is that the specific elements in the two start-up programs are not considered in the analysis and that oversight makes it impossible to assess what it is specifically in the program that works. For instance it is unclear if it is a particular good case worker, a certain element in the programs or the programs as a whole that are effective in terms of moving people into self-employment. And further, as with the argument in the preceding section, the specific target group is not defined, which makes the results even more muddy as you might assume that the population is consistent made up of people who in order to gain self-employment one has to be fairly ‘job-ready’ and without severe problems beside unemployment. This is not made clear in the evaluation, that end up concluding that ALMP focusing on self-employment may be the most effective, both in Germany and elsewhere.

As the above examples illustrate the typical methodology for studies regarding ALMP programs is a quantitative methodology using administrative data or e.g. randomized controlled trials – both with a linear understanding of causality underlying the research. Though, a few exceptions are present within this category of the mapping review. One is an evaluation of three cases of activation projects for young unemployed with problems beside unemployment. Here Bredgaard & Hansen (2012) apply a diversified evaluation method to specify and supplement effect evaluations and the commonly used controlled experiments on the one hand, and process evaluations and case studies on the other hand. They use a combination of focus group interviews, individual interviews and statistical data in order to evaluate the three projects. Also important in this approach is the aim at building contextualized and at the same time generalized knowledge about what works, for whom and under which conditions. Not surprisingly, the results are not as clear-cut as the ones presented in the research above because more elements and contextual terms are taking into account. Thus, the evaluation of the three activation projects show that it is a complex interaction between different mechanisms that contribute in creating results for a very complex group of unemployed. Among other things, dimensions like self-reflection, motivation and making positive relations are highlighted as important in order to improve the employability of young people with problems beside unemployment (Bredgaard and Hansen 2012). Yet, this study is distinctive in its nuanced and diversified approach.

Despite the exceptions regarding methodology in this study, the dominant trend in these studies is quantitative analysis that focuses on highlighting linear causality, where different program features are tested for their effects on the measureable client outcomes regarding employment.

Moving on to the second sub category is the studies with no clear operationalization of the ALMP being investigated. The studies in this category represent a plurality of research regarding the methodology, the understanding of outcomes measures and the research objects. Before reviewing this plurality of research, it is important to state that in this case ‘unemployment programs’ or ‘active labor market programs’ are considered general terms which are not  really adequate to explain the specific content of particular programs. Thus the studies in this category are investigating more than just ALMP programs and also include studies that use one of these generic concepts without providing further details and descriptions of the specific program.

An example of a study applying one of these overall concepts is Graversen & Jensen (2010) who evaluate private sector employment programs for Danish benefit recipients. They use a latent variable model that allows for heterogeneous treatment effects among observationally identical people to estimate commonly defined mean treatment effects and the distribution of treatment effects. Graversen & Jensen find no significant mean treatment effect of the private sector employment programs when comparing to other programs intended for the same group of unemployed. What is unclear is the composition of the private sector employment programs. Both public and private sector programs vary significantly depending on the provider (e.g., the specific private organization or the local municipality responsible for the public employment programs), the different elements in the programs (e.g., focusing on job searching skills, vocational training or personal appearance), the employees working on the programs and their relation with the unemployed, the group of clients participating in the program (with reference to the arguments presented in the above section) etc. Given the limitations of the study and in particular its lack of attention to program diversity, if in fact the study showed that there was a significant effect when using private sector programs, this knowledge would unfortunately be unclear and difficult to apply in practice (Graversen and Jensen 2010).

A similar example is found in Albert & King’s study from 2011, where exits from the Temporary Assistance for Needy Families (TANF) are studied by longitudinal monthly administrative data and a Cox proportional hazard model. The primary focus of TANF is to assist low income families meet basic needs by providing cash assistance, supportive services and training as well as employment opportunities. But how and with what weight on the different elements TANF is being implemented across the country, is unknown. Therefore, findings regarding high exit rates (within the first eight months) or that families headed by noncitizens remain on TANF longest leave unanswered questions limiting the relevance to improving ALMP programs (Albert and King 2011) (See also Arellano 2010; Behncke, Froelich, and Lechner 2008; Bernhard and Kopf 2014).

As stated, this sub-category also includes a variety of research not focusing on specific interventions or programs. This goes for Rivera et al. (2012) studying unemployed people’s perception of their employability competences, London’s research on college attendance effect on employment chances and Cheng (2005), who investigates factors affection the return to TANF programs by former participants who left welfare benefits (Rivera et al. 2012; Cheng 2005; London 2005). Much of the research not aiming directly at specific ALMP focus on more soft measures and consequences of unemployment like subjective well-being and mental health issues. For instance, Ervasti & Venetoklis (2010) analyses the level of subjective well-being and its determinants among unemployed compared to people in paid labor. With a hypothetical deductive starting point, they test whether unemployment is a psychological stressor, whether the level of well-being among unemployed is sufficiently high to discourage them from actively and effectively searching for a new job, and finally whether there is adverse effects of financial stress for subjective well-being during unemployment. Ervasti & Venetoklis use data from the European Social Survey in 21 countries and finds evidence that the financial strain of unemployment is crucial for the well-being of unemployed and that unemployment is related to psychological factors (Ervasti and Venetoklis 2010).

Unlike the sub-category representing the studies with a clear intervention focus, a great number of studies in this sub category are applying a qualitative methodology. And, even more unlike the studies in the other sub-category this includes an approach where the unemployed perspective or point of view is addressed (Leibetseder and Kranewitter 2012; Eskelinen and Olesen 2010). For example Cooney examines the perspective of welfare recipients as they experience the policy enacted at the street level. The purpose of the study is not to determine what works in the ALMP, but to provide information about the gaps between the assumptions embedded in welfare legislation that become apparent when we learn in-depth about the complexity of welfare recipients’ lives. Through focus group interviews the analysis identifies two coping strategies invoked by women in the welfare-to-work programs: respectively affirming motherhood and strategic acceptance of sanctions (Cooney 2006). These kinds of ‘bottom-up’ studies provide a critical missing ingredient not found in the quantitative studies. Another study ‘taking a view from the bottom’ is Breitkreuz & Williamson, who base their examination of social assistance recipients’ attempt to make the transition from social assistance to employment on a longitudinal, institutional-ethnographic study with 17 welfare-to-work participants in Canada. Through in-depth interviews with the participants, they find that there is a considerable gap between what welfare-to-work programs are intended to do and what they in fact do (Breitkreuz and Williamson 2012). Though looking across the body of research, it is often the case managers being interviewed in the studies applying qualitative methods (Coop 2010; Weibel and Madsen 2012).

The findings in the initial analysis of the target group can be mirrored when focusing on the nature of the interventions within the research. Thus the prevailing tendency is studies applying a general concept of activation programs or ALMP when describing their research object. When detailed descriptions of the content of a specific program is lacking, the potentially gained knowledge is embedded in what amounts to an empty shell not adequate when it comes to either assessing the validity of the research or using the knowledge to improve current programs and policies. Once again, the one-size-fits-all understanding of the field predominates.

6 To be or not to be unemployed?

In this third and final section of analyzing the body of research extracted from the mapping review, focus is on the understanding of effects and outcomes of ALMP. Just by glancing through everyday communication in newspapers, listening to political speeches and conversations between scholars and non-scholars it becomes obvious, that these are everyday concepts, often used without any kind of explanation or definition attached to it. Further, in academia the understanding of effects and outcomes depends on the research field one operate within and/or the methodological approach applied in the studies. Narrowing the scope down to only social sciences, this is still the case. Therefore, in order to provide a clear overview and identify the potential knowledge gaps, it is pivotal to dissect the different definitions and outcomes present in the body of research. As mentioned in the descriptive analysis, the categorization of the studies changed during the review process, ending up with two initial sub categories; hard outcomes or soft outcomes.

More than two-thirds of the studies in the outcome-category are coded into the hard-outcome sub-category. The prevailing understanding of outcomes in this sub-category derives from the econometric and statistical research, where effects and outcomes are closely linked to the linear understanding of causality, and the purpose of the research is to explain the relationship between a program and its outcomes using observations on a group of individuals. Different methods are applied – e.g. randomized control trials (RCT), where the outcomes between an experimental group exposed to the program and a control group not participating in the program, are compared. This is the case in Carling & Richardson’s study from 2004, where the relative efficiency of eight Swedish labor market programs are estimated in terms of reducing the unemployment duration for participants. They use a rich administrative data set of all adult unemployed workers and find, that programs in which the participants obtain subsidized work experience and training provided by firms have better outcomes than classroom vocational training. Further, they find that the relative efficiency is similar across demographic and skill groups and independent of the timing of the placement (Carling and Richardson 2004). A similar example of this classic outcome understanding is found in Crepon, Dejemeppe & Gurgand’s (2005) research on the effects of intensive counseling schemes designed to improve the quality of job-referrals. At the time of the study, the schemes were provided to about 20 percent of the unemployed in France. Crepon, Dejemeppe & Gurgand identify heterogeneous and time-dependent causal effects of the schemes, using a rich data set and duration models. They find significant outcomes with positive impact on unemployment recurrence stronger than on unemployment duration (Crepon, Dejemeppe, and Gurgand 2005). As these two examples illustrate, a focus on outcomes with a linear approach to causality requires a research object which can be formulated and operationalized into straight-forward measurable variables (something that is often not immediately available for many studies).

Moving beyond the simple classical understanding of outcomes, many of the studies in this sub- category distinguish between different kinds of hard outcomes. The most frequent ones in this regard are outcomes in terms of exit-rates, income increasing, lock-in effects, spillover and tax effects as well as threat effects. To give a few examples, Arntz (2008) examines the lock-in effects of unemployed people in Germany choosing search strategies that favor migrating out of declining regions or not. Using a competing-risk framework of exiting unemployment to jobs in local or distant labor market areas, Arntz finds that unemployed in West Germany are responsive to local labor market conditions and are more likely to leave regions with unfavorable re-employment opportunities (Arntz 2008). Further, in an experimental setting Graversen & van Ours (2009) study the threat effects of mandatory activation programs for unemployed workers in Denmark. The results imply, if not definitively prove, a significant threat effect since the unemployed in the experimental group found a job more quickly than the unemployed in the control group. In addition, they find that the quality of the post-unemployment jobs was not affected by the activation program, and on this basis concludes, that activation programs mainly work because they are compulsory and unemployed do not like them (Graversen and van Ours 2009).  They do not however provide evidence of what participants actually do and do not like. So, once again, the bottom-up approach is the missing ingredient in providing findings of sufficient specificity for policy change.

When distinguishing between these different hard outcome measures, it is important to be aware of what they do not say in addition to what they do.  For instance, several studies use exit rates and administrative data on unemployed leaving welfare benefits when determining the effects of ALMP. However, leaving welfare is not necessarily equal to finding a job (Berg, Klaauw, and van Ours 2004; Bryner and Martin 2005). On the contrary, unemployed exiting welfare can do so without entering the job market. In some cases, the unemployed refuse to be subject to the public restrictions related to being eligible for benefits and thus decide to get by without the financial support from the state. In other cases, the unemployed might decide to depend on a working spouse, or the elderly unemployed might feel pressured to retire before planned. On this basis, some studies in this category have applied a more refined research design, where more variables, details and sophisticated analyses are taken into use. For example Kemp & Davidson (2010) investigate the implications of a new distinction between unemployed who potentially could do paid work and unemployed for whom that outcome is much unlikely, which was introduced in United Kingdom in 2008. Drawing on baseline and follow-up surveys, they examine the employment status trajectories of a cohort of incapacity benefit claimants by asking the three questions: (1) Who returns to work and what factors are associated with that outcome? (2) How employable are the clients not returning to work and does their employability change? and (3) What distinguishes clients who classify themselves as permanently unable to work from those who do not (Kemp and Davidson 2010)? Bencke et al. (2008) use a detailed linked jobseeker-caseworker dataset to investigate the effects of caseworkers’ cooperativeness on employment probabilities of their clients. By using a modified statistical matching method, they find that caseworkers who place less emphasis on a cooperative and harmonic relationship with their clients increase their employment chances in the short and medium term (Behncke, Froelich, and Lechner 2008). These studies point to the need to consider a wide variety of factors that can affect employment-related outcomes.

Based on the analysis of the studies within the first of the two sub-categories, it is clear that even when focusing only on hard outcomes, there a lot of variation and differences in the understanding of this concept. And, when proceeding to the next sub-category, these understanding only become more differentiated and thus the picture gets even muddier.

Among the studies with a focus on soft outcomes, it is possible to make a distinction between research that quantify traditionally soft variables, research not immediately associated with statistical calculations, and a little off-category of research not applying an outcome focus at all. The vast majority is research related to the first of these three distinctions and is primarily examining the correlation between ALMP and health or subjective well-being. For example, Ayala & Rodriguez (2013) evaluate the effects of welfare-to-work programs on physical and mental health status. They base their research on data from the minimum income program in Madrid and match the administrative data from the program with a survey of former welfare recipients who took part in different work related activities. By calculating propensity scores, they find that both health status and health behaviors were modestly better for people that had taken part in the work related activities. On this basis, Ayala & Rodriguez concludes that welfare-to-work policies may have positive unintended health outcomes (Ayala and Rodriguez 2013). A similar study by Coutts, Stuckler & Cann (2014) also finds that government training programs can have a positive effect on the participant’s well-being. They further find that ALMP can be designed to have a double effect in terms of improving participant’s basic skills and education, and thereby increase the potential for entering the labor market.  This study points out that psychological health is both a necessary part of an individual’s portfolio of employability and a step towards labor market integration—something that many studies fail to consider (Coutts, Stuckler, and Cann 2014).

Not all of the research in this sub-category is quantitative even though there is an outcome focus. This goes especially for studies where the research object is to gain in-depth knowledge about a complex and differentiated target group, and where the context of the research object plays an important role in understanding and explaining the outcomes. Here, an econometric or statistical model of explanation falls short, and a different understanding of outcomes is needed. Very few studies in the mapping review has engaged in research were qualitative methodology is combined with an attempt to investigate outcomes of ALMP. One of the few is Eskelinen & Olesen’s work from 2010, where an empirical study of the content and consequences of ALMP is examined from the perspective of the unemployed (See also Leibetseder and Kranewitter 2012). They use short narratives to trace changes in welfare recipients’ experience of work identity when participating in welfare-to-work programs. Among other things, they find that clients have important knowledge of the ALMP and they as actors have a significant role in designing the ALMP being implemented. On this basis Eskelinen & Olesen (2010) conclude that research on the effects of ALMP is not necessarily limited to studies on employment efforts, but can also be applied in other fields in order to gain knowledge from the point of view of the clients (Eskelinen and Olesen 2010). A similar research focus is found in King’s study of current and former welfare recipients living Ohio who are in transition from welfare to work. By performing in-depth qualitative studies, King (2008) identifies those services, resources, policies and procedures that the participants perceived to be effective in assisting them to make the transition from welfare to work.

Finally, the last group of studies includes those that actually exceed the initially intended criteria for this sub-category, i.e., studies with no outcome focus whatsoever. Since there turned out to be several studies matching this definition it was necessary to make this adjustment. Even though most of this research points in very different directions (though all within the boundaries of ALMP) there is a preponderance of studies concerned with the lived experiences of unemployed. For example Alzate et al. (2009) investigate the realities of poor women living in Georgia face in the pursuit of employment. Through a survey undertaken by the Welfare Reform Project in Georgia 107 women report on their confidence in leaving welfare. Alzate et al. finds that health followed by age are the strongest predictors of confidence to leave welfare and on this basis they conclude that the confidence of female welfare recipients may be greater than can be justified by the realities of their lives and the society they live in (Alzate et al. 2009). Two different studies without any outcome focus relevant to highlight are, respectively, a study of the case managers options when organizing and coordinating activation programs and a study the interactions between clients and case workers in conversations at selected job centers. In the latter Caswell, Eskelinen & Olesen show how clients try to evade the demands of labor market identities in their negotiations with the caseworkers. The results of in-depth analysis of several qualitative observations of meetings between clients and case managers, Caswell, Eskelinen & Olesen (2013) find that clients are in fact active participants who resist, protest against and sometimes even avoid the identities offered by them by the employment system. Regarding the former, Weibel & Madsen (2012) interview more than 50 case managers, managers and employees at activation programs in a qualitative study of the work in Danish job centers. They conclude that the case managers lack knowledge on the effects of the activation programs, that in general there is a lack of knowledge about the content of the available activation programs and finally, that there a great challenges in securing continuity and focus in the programs (Weibel and Madsen 2012). These studies commonly focus on processes and interactions rather than on outcomes. This means that the detailed knowledge about the ALMP revealed is not directly linked to specific interventions and, I would argue, therefore is not likely to be taken into consideration when it comes to improving the ALMP.

The above findings from the mapping review pose significant challenges for reframing how to best connect the research to efforts to improve ALMP policy. The analysis has shown that studies often had: (1) unclear target groups , where no distinctions are made between different social problems and causes of unemployment; (2) a linear understanding of causality that failed to address the dynamic effects of various features of specific programs; (3) results that were primarily focused within a range of different quantitative outcome measures; and (4) a top-down approach that left behind a knowledge gap about how clients experience these programs.  Across the board, there is an urgent need for in-depth knowledge and sensitivity toward the individual client’s role as an active participant in ways that affect outcomes and could be the basis for forming improved policies.

Despite these glaring limitations, there has been an increasing emphasis on researching ALMP as the golden way to get people off benefits. It could be argued that this is actually not a problem since politicians often do not bother to take into account research when making the legislation. But, as the following case will illustrate if/when research is included in the making of policy, it can have serious consequences when the knowledge base is biased. This is most especially a problem when the one-size-fits-all logic is relied on (even if only tacitly). Drawing on a Danish case, it is demonstrated how ALMP in Denmark has expanded the logics from one area of policy to a much broader and less homogeneous area. Using the example of research on sanctioning in Denmark, it is argued that there has been a tendency to take policy shown to have positive outcomes for one target group and expand this policy artlessly to encompass a much larger group of unemployed people.

7 Expanding policy beyond the limits of the knowledgebase

The idea of using economic sanctions as an instrument to discipline clients to focus more intently on getting a job is inherently punitive. It becomes less attractive to receive benefits when benefit receipt becomes precarious and can be curtailed or removed as a consequence for not meeting the duties demanded by the welfare system. When introduced in Denmark in the 1990ts, sanctions were initially reserved for clients with no personal problems besides being unemployed. But following the reforms of the welfare system and the implementation of mandatory activation programs for all unemployed, sanctioning became an inevitable consequence for all clients who were not adhering to the prescribed terms for receipt of assistance, even though evaluations show that the most vulnerable clients rarely were activated (Brogaard & Weise 1997: 11-14). By 1998 all unemployed clients were included in mandatory activation programs and thus they all were at risk of being sanctioned with reduction or termination of benefits if absent or if refusing job openings and/or participation in activation activities. Since then, legislation on sanctions has continued to become more restrictive, including increasing the benefit reductions from a maximum of 20 to 30 percent and by making it an absolute duty for the municipalities to sanction when clients were absent from activation.

Several studies have investigated the effectiveness of sanctions in Denmark as an instrument to move unemployed people away from welfare benefits. The research mirrored findings from other parts of the world (Arni, Lalive, and van Ours 2009; Berg, Klaauw, and van Ours 2004; Bong, Slack, and Lewis 2004).  Sanctions were generally found to be an effective tool of active employment policy. But, the one-size-fits-all mentality undermines these studies, since they analyzed only data on people with no problems beside unemployment. Representative is Svarer (2011) who argues that even moderate sanctions have rather large effects (Svarer 2011). However, Svarer’s study also shows that the effect of sanctions decreases over time and they have very heterogeneous effects varying across even this relatively homogeneous client population.  Further, when we look at these studies overall, these effects are often presented as effects on the transition from welfare to work, but what is generally measured in this as well as in international studies is not whether a client moves into paid employment, but only if the client no longer receives welfare benefits. As described in the analysis of the understanding of ALMP outcomes, this can also mean that a client decides to retire before planned, tries to manage on a spouse’s income or even seeks to go without any income at all.

The results from these studies reflect how the Danish authorities have prematurely come to view the use of economic sanctions as an effective tool for motivating unemployed people in general to get back into employment, but without little or limited differentiation between the various groups of unemployed. Following the implementation of the new and harsher ALMP regime in Danish municipalities, there has been a substantial resistance from social workers and other frontline workers (Caswell and Høybye-mortensen 2015). Among other things, caseworkers now must of necessity use a lot of precious time trying to understand the legislation and then assessing whether a sanction is to be given and if so, what kind of sanction. Less needed attention can be given to clients. At the same time, the unemployed themselves often are left with confusion, finding it difficult to understand the reasoning behind a sanction, why they are sanctioned, for how long and with what economical consequences.

As mentioned above, in the period from the early 1990s up until today, we have seen a gradual but substantial widening of the target group for economic sanctions as well as a development of operational reforms ensuring wide and consistent implementation of the sanctioning policies in Denmark (Caswell and Larsen 2015). The beginning of 2013 saw yet another reform of the social benefit system (Ministry of Employment, 2013) which emphasizes the political confidence in the positive effects of economic sanctions. It introduces harsher punitive sanctions and focuses on enhancing clients’ availability for work. Sanctions are identified as the solution for a certain hypothetical group of clients, that is, those who “systematically avoid the demands placed upon them” (Ministry of Employment, 2013).  Specification of who these unemployed actually are however is not provided. This mystery can cast a long shadow over the diversity of clients participating in labor activation. While the research points in the direction that sanctioning can create positive results clients with no problems besides unemployment, there has been no evidence as to whether sanctions are also effective for more complex client groups with problems besides unemployment. Instead, these more complex clients are either placed in the same category of ‘unemployed workers’ or they are recognized to the extent that the results for this group are too muddy to provide a clear picture of the outcomes. Further, the most recent research has shown that sanctioning can have perverse effects undermining program goals related to activation (Fording, Schram, and Soss 2013; Caswell, Andersen, and Sieling 2011). Ultimately, this means that sanctioning is used as a tool to get clients off benefits without knowing if in fact this is the best tool for all unemployed. Failure to seriously consider this more recent research allows punitive economic sanctions to be expanded beyond the limits of the knowledge base to include clients for whom this tool could be directly harmful. Thus, it is possible to argue that sanctioning the unemployed with, for instance, mental problems would contribute to pushing them even further off the edge of the job market if the consequence is that they cannot afford to by their medicine or continue with their physiological treatment or just that they become more emotionally distressed (Fording, Schram, and Soss 2013; Caswell, Andersen, and Sieling 2011).

What this example from the Danish unemployment system shows us is the consequences that can follow when knowledge about implementing new policies is expanded beyond its limits. There is a need for research that addresses the complexity of the unemployed as a group, the nuances of tailoring services for them, and the non-linear dynamics operating in activation programs. Further, research should focus on the variety of positive outcomes beyond employment. We need to emphasize both hard and soft outcomes, for the short as well as the long term, and the outcomes not immediately apparent in administrative registers but nonetheless important, which can become apparent when we take a bottom-up perspective that includes consideration of how clients experience programs.

8 When one size doesn’t fit all

Mapping research on ALMP leaves behind a clear picture that the research is extensive in terms of the number of studies available, but it is also lacking nuanced and does not reflect the diversity of the field, both in terms of programs and clients. These lapses in research can have serious consequences for the clients exposed to policies that are based on this faulty knowledge base.  In other words, it is usually the case as the cliché suggests that one size doesn’t fit all, and that is hopefully now apparent when it comes to using ALMP as a tool to get unemployed off welfare benefits. Yet from the research, we find the predominant picture is that there is still a one-size client (the unemployed workers), a one-size activation scheme reflecting a linear causality without consideration of program dynamics, and a very superficial focus on a one-size outcome targeting the quantifiable measures of effects, largely related to employment as an end in itself (even when it is not necessarily right for particular clients).

The foregoing analysis reveals these gaps in ways that hopefully will lead to better research in the future. This new research should reframe how we look at activation, starting with the realization that one size in fact doesn’t fit all when it comes to ALMP. Policy evolution suggests that ALMP is in fact still treated as the ‘one size fit all’ concept. On the basis of the analytical findings in the review, it can be argued that too many people are being expected to participate in activation policies when it is in fact doubtful and very unclear if this is the best or even a helpful response for addressing the challenges related to their unemployment. The trend is in the negative direction and hopefully new and better research will begin the process of reversing it.

Appreciating the need for nuanced approaches to serve diverse populations can have substantial consequences for policy development. This is the challenge for ALMP research going forward. Although difficult, it should not be neglected by clinging to the one-size-fits-all mentality. Instead, we must embrace the challenge in the name of servicing clients better.

References:

Aguilar Rivera, M. del C., Villardon Gallego, L., Alvarez Alvarez, M., Moro Inchaurtieta, A., Elexpuru Albizuri, I., & Alvarez de Eulate, C. Y. (2012). Perceived Employability and Competence Development. Procedia Social and Behavioral Sciences, 69, 1191–1197.

Albert, V., & King, W. (2011). Citizenship Status and TANF Exits: A Proportional Hazard Model. Journal of Social Service Research, 37(3), 294–308.

Alzate, M. M., Moxley, D. P., Bohon, S. A., & Nackerud, L. (2009). The Influence of Perceived Health Status on Poor Women’s Confidence in Leaving Welfare: Implications for Social Work. Social Work in Health Care, 48(1), 57–75.

Arellano, F. A. (2010). Do training programmes get the unemployed back to work? a look at the spanish experience. Revista de Economia Aplicada, 18(53), 39–65.

Arni, P. P., Lalive, R., & van Ours, J. C. (2009). How Effective are Unemployment Benefit Sanctions? Looking Beyond Unemployment Exit. SSRN Working Paper Series. University of Lausanne - Department of Economics (DEEP) ; Tilburg University - Department of Economics: Social Science Research Network.

Arntz, M. (2008). The Geographical Mobility of Unemployed Workers. SSRN Working Paper Series. Rochester, Rochester: Social Science Research Network.

Ayala, L., & Rodriguez, M. (2013). Health-related effects of welfare-to-work policies. Social Science & Medicine, 93, 103–112.

Bauld, L., Templeton, L., Silver, K., McKell, J., Novak, C., & Hay, G. (2013). Pathways back to work for problem alcohol users. Policy Studies, 34(3), 360–376.

Baumgartner, H. J., & Caliendo, M. (2008). Turning unemployment into self-employment: Effectiveness of two start-up programmes. Oxford Bulletin of Economics and Statistics, 70(3), 347–373.

Behncke, S., Froelich, M., & Lechner, M. (2008). Unemployed and Their Caseworkers: Should They Be Friends or Foes. SSRN Working Paper Series. Universität Mannheim, Chair of Econometrics ; University of St. Gallen - Swiss Institute for Empirical Economic Research: Social Science Research Network.

Berg, G. J. van den, Klaauw, B. van der, & van Ours, J. C. (2004). Punitive Sanctions and the Transition Rate from Welfare to Work. Journal of Labor Economics, 22(1), 211–241.

Bernhard, S., & Kopf, E. (2014). Courses or individual counselling: does job search assistance work? Applied Economics, 46(27), 3261–3273.

Bong, J. L., Slack, K. S., & Lewis, D. A. (2004). Are Welfare Sanctions Working as Intended? Welfare Receipt, Work Activity, and Material Hardship among TANF-Recipient Families. Social Service Review, 78(3), 370–403.

Bonoli, G. (2010). The Political Economy of Active Labor-Market Policy. Politics & Society, 38(4), 435-457.

Boone, J., & van Ours, J. C. (2006). Modeling financial incentives to get the unemployed back to work. Journal of Institutional and Theoretical Economics-Zeitschrift Fur Die Gesamte Staatswissenschaft, 162(2), 227–252.

Booth, A., Papaioannou, D., & Sutton, A. (2012). Systematic Approaches to a Succesful Literature Review. London: Sage Publications Ltd.

Bredgaard, T., & Hansen, C. (2012). What works in activation policies for young unemployed with problems besides unemployment? Tidsskrift for Arbejdsliv, 14(1), 73–91.

Breitkreuz, R. S., & Williamson, D. L. (2012). The self-sufficiency trap: A critical examination of welfare-to-work. Social Service Review, 86(4), 660–689.

Brodkin, E. Z., & Larsen, F. (2013). The Policies of workfare: At the boundaries between work and the welfare state. In E. Z. Brodkin & G. Marston (Eds.), Work and the Welfare State: Street-level Organizations and Workfare Politics (pp. 57-67). Washington, DC: Georgetown University Press.

Brogaard, S., & Weise, H. (1997). Evaluering af Lov om kommunal aktivering. Kommuneundersøgelsen. København: Socialforskningsinstituttet.

Bryner, G., & Martin, R. (2005). Innovation in Welfare Policy: Evaluating State Efforts to Encourage Work among Low-Income Families. Review of Policy Research, 22(3), 325–343.

Butler, A. D., Alson, J., Bloom, D., Deitch, V., Hill, A., Hsueh, J., Jacobs, E., Kim, S., McRoberts, R., & Redcross, C. (2012). What Strategies Work for the Hard-to-Employ? Retrieved May 2015, from OPRE: http://www.acf.hhs.gov/programs/opre/resource/what-strategies-work-for-the-hard-to-employ-final-results-of-the-hard-to

Carling, K., & Richardson, K. (2004). The relative efficiency of labor market programs: Swedish experience from the 1990s. Labour Economics, 11(3), 335–354.

Caswell, D., Andersen, H. L., & Sieling, S. M. (2011). Økonomisk straf bringer ikke kontanthjælpsmodtagere i job. København: AKF, Anvendt Kommunal Forskning

Caswell, D., & Larsen, F. (2015). Frontlinjearbejdet i leveringen af aktiv besjkæftigelsespolitik: hvordan faglige, organisatoriske og styringsmæssige kontekster påvirker dette. Tidsskrift for Arbejdsliv, 17(1), 9-28.

Caswell, D., & Høybye-Mortensen, M. (2015). Responses from the Frontline: How Organisations and Street-level Bureaucrats Deal with Economic Sanctions. European Journal of Social Security (EJSS), 17(1), 31–52.

Cheng, T. (2005). Recidivism Among Former Recipients Who Remain Unemployed: The Vicious Circle Created by Welfare Reform. Families in Society, 86(1), 134–142.

Cooney, K. (2006). Mothers First, Not Work First: Listening to Welfare Clients in Job Training. Qualitative Social Work, 5(2), 217–235.

Coop, A. (2010). Veje til beskæftigelse - En kvalitativ undersøgelse af indsatser over for ikke-arbejdsmarkedsparate kontanthjælpsmodtagere. Retrieved May 2015, from SFI: http://www.sfi.dk/resultater-4726.aspx?Action=1&NewsId=2565&PID=9422

Coutts, A. P., Stuckler, D., & Cann, D. J. (2014). The health and wellbeing effects of active labor market programs. In F. A. H. C. L. Cooper (Ed.), Interventions and policies to enhance wellbeing (Vol.VI) (pp. 465–482). Wiley-Blackwell: Wiley-Blackwell.

Crepon, B., Dejemeppe, M., & Gurgand, M. (2005). Counseling the Unemployed: Does it Lower Unemployment Duration and Recurrence? SSRN Working Paper Series. National Institute of Statistics and Economic Studies (INSEE) - National School for Statistical and Economic Administration (ENSAE) ; Catholic University of Louvain - School of Economic and Social Research ; National Center for Scientific Research (CNRS): Social Science Research Network.

Crépon, B., Ferracci, M., Jolivet, G., & van den Berg, G. J. (2009). Active labor market policy effects in a dynamic setting. Journal of the European Economic Association, 7(2-3), 595–605.

De Bruijn, H. (2007). Managing Performance in the Public Sector.London: Routledge.

Dingeldey, I. (2007). Between workfare and enablement - The different paths to transformation of the welfare state: A comparative analysis of activating labour market policies. European Journal of Political Research, 46(6), 823–851.

Ervasti, H., & Venetoklis, T. (2010). Unemployment and Subjective Well-being. An Empirical Test of Deprivation Theory, Incentive Paradigm and Financial Strain Approach. Acta Sociologica, 53(2), 119–139.

Eskelinen, L., & Olesen, S. P. (2010). Beskæftigelsesindsatsen og dens virkninger set fra kontanthjælpsmodtagernes perspektiv. København: AKF, Anvendt KommunalForskning.

Fording, R. C., Schram, S. F., & Soss, J. (2013). Do Welfare Sanctions Help or Hurt the Poor? Estimating the Causal Effect of Sanctioning on Client Earnings. Social Service Review, 87(4), 641-76.

Fuertes, V. & McQuad, R. (2014). Personalisation of labour market policies in the UK: the professional, the citizen, and the policy. Paper presented at the SPA Annual Conference, 2014 July Sheffield.

Gaure, S., Roed, K., & Westlie, L. (2012). Job search incentives and job match quality. Labour Economics, 19(3), 438–450.

Graversen, B. K., & Jensen, P. (2010). A Reappraisal of the Virtues of Private Sector Employment Programmes*. Scandinavian Journal of Economics, 112(3), 546–569.

Graversen, B. K., & van Ours, J. C. (2009). How a Mandatory Activation Program Reduces Unemployment Durations: The Effects of Distance. SSRN Working Paper Series. Danish National Centre for Social Research (SFI) ; Tilburg University - Department of Economics: Social Science Research Network.

Heidenreich, M., & Aurich-Beerheide, P. (2014). European worlds of inclusive activation: The challenges of coordinated service provision. International Journal of Social Welfare, 23(1), 6-22.

Hetrick, S. E., Parker, A. G., Callahan, P., & Purcell, R. (2010). Evidence mapping: illustrating an emerging methodology to improve evidence-based practice in youth mental health. Journal of Evaluation in Clinical Practice, 16, 1025–1030.

Herbig, B., Dragano, N., & Angerer, P. (2013). Health in the Long-Term Unemployed. Deutsches Arzteblatt International, 110(23-24), 413–U17.

Kemp, P. A., & Davidson, J. (2010). Employability trajectories among new claimants of Incapacity Benefit. Policy Studies, 31(2), 203–221.

Lechner, M., & Gerfin, M. (2004). Microeconometric Evaluation of the Active Labour Market Policy in Switzerland. SSRN Working Paper Series. University of St. Gallen - Swiss Institute for Empirical Economic Research ; University of Bern: Social Science Research Network.

Leibetseder, B., & Kranewitter, H. (2012). Activation and the Austrian social assistance: permanent entry. International Journal of Sociology and Social Policy, 32(7/8), 448 - 460.

London, R. A. (2005). Welfare recipients’ college attendance and consequences for time-limited aid. Social Science Quarterly, 86(5), 1104–1122.

Ministry of Employment. (2013). Alle kan gøre nytte - udspil til en kontanthjælpsreform. Retrieved May 2015, from Copenhagen: Ministry of Employment:   http://bm.dk/da/Aktuelt/Pressemeddelelser/Arkiv/2013/02/Kontanthjaelpsreform%20Alle%20kan%20goere%20nytte.aspx

Schram, S. F. (2006). Welfare Discipline. Discourse, Governance and Globalization. Pennsylvania: Temple University Press.

Soss, J., Fording, R. F., & Schram, S. F. (2011). Disciplining the Poor. Neoliberal paternalism and the persistent power of race. Chicago: The University of Chicago Press

Svarer, M. (2011). The Effect of Sanctions on Exit from Unemployment: Evidence from Denmark. Economica, 78(312), 751–778.

Torfing, J. (2004): Det stille sporskifte i velfærdsstaten. En diskursteoretisk beslutningsprocesanalyse. Århus N: Aarhus Universitetsforlag. 

Van Berkel, R., De Graaf, W. and Sirovátka, T. (Eds.). (2011). The governance of active welfare states in Europe. Palgrave Macmillan.Wildavsky & Pressman

Weibel, K., & Madsen, M. B. (2012). Delt viden. Retrieved May 2015, from København: SFI: http://www.sfi.dk/rapportoplysninger-4681.aspx?Action=1&NewsId=3658&PID=9267

Author´s Address:

Sophie Danneris
Aalborg University, Department of Sociology and Social Work

Kroghstraede 7, 9220 Aalborg Oest, Denmark

+45 4043 2667

sophiedj@socsci.aau.dk