Refugee integration policy the Norwegian way – why good ideas fail and bad ideas prevail

Abstract

German

French

Integrating non-Western refugees into the highly specialised Scandinavian labour markets has proven difficult. This highly ideological policy field is an interesting case for the study of policy learning versus ideas as drivers for institutional change or continuity. Using the Norwegian Introductory Programme as a case study, we show that the application of core programme measures remains largely unaffected by evaluations that show that such measures tend to have very modest effects on the labour market integration of refugees. Concurrently, incremental changes in the disciplining elements of the programme have resulted in an increasingly controlling activation regime. Our interpretation is that a major driver behind the intensification of disciplinary elements has been the assumption that participants lack the motivation to integrate into the labour market. Moreover, we find that this assumption presents an obstacle to policy learning with regard to programme quality. Within activation, policy ideas seem to function simultaneously as path-reinforcing cognitive locks and as drivers for political change.

Keywords

Integration policy welfare state change policy learning refugees Norway

Introduction

In the Scandinavian countries – Denmark, Norway and Sweden – in recent decades, the number of migrants and the percentage of the population born outside the European Union (EU) and the European Economic Area (EEA)¹ have increased substantially (Djuve and Grødem, 2014). Refugees have struggled to find a foothold in local labour markets despite full legal access to the labour market and significant integration efforts. Participation rates increase with duration of residence but in most refugee groups they remain significantly lower than among natives (Djuve and Grødem, 2014). Traditionally, the Nordic welfare model² has been characterised by universal welfare services, but the three Scandinavian countries have responded to the integration challenge by setting up extensive programmes to facilitate refugees’ labour market integration.³ Launched in 2003, the Norwegian Introduction Programme (NIP) lasts for two (sometimes three) years and consists mainly of language and on-the-job training similar to that provided through traditional active labour market programmes (ALMPs).⁴ Djuve (2011) argues that this fairly sharp deviation from previous integration policy can be explained by a combination of critical-juncture and path-dependency theory (Pierson, 2000), alongside theories of policy learning and the importance of ideas. In this article, we contend that, particularly since 2013, ideas about the need to control the behaviour of new migrants have overshadowed other drivers for change, such as policy learning, and explain a tightening of behavioural requirements, as well as the limited consideration paid to the need to develop training measures.

The NIP is a classic example of activation policy. Activation policy has a dual nature. It combines measures aimed at increasing participants’ labour market opportunities with sanctions or incentives intended to regulate their behaviour. This duality is at the core of a number of discussions of the ‘true face of activation’. Are activation measures intended mainly to curb rights or to increase employability through social investment (Larsen, 2005)? This tension makes it particularly interesting to investigate whether political change is focused on one rather than two pillars, and, if so, whether the ‘social investment’ pillar or the ‘behavioural control’ pillar is emphasised.

In this article, we outline how the NIP has evolved since 2003 and demonstrate that in this period, and particularly since 2013, the main policy changes have concerned the disciplining elements of the programme. We also document that research on the effectiveness of activation measures does not seem to support this development, but rather emphasises shortcomings in the social investment measures.

In the discussion, we consider possible explanations of why core measures are maintained despite disappointing results and new measures with undocumented effects are introduced. We draw on path-dependency theory (Pierson, 2000) and on theories of the importance of ideas and their construction (Blyth, 2001; Hemerijck, 2013; Levitas, 2005; Schmidt, 2008; Taylor-Gooby, 2007). Theories of policy-making predict deviations from a straightforward policy learning process; politicians will not necessarily suggest additional measures that work, nor the removal of measures that do not. Path dependency might influence policy-makers to suggest ‘more of the same’ even if there is no evidence of success and, when action is taken, ideological framing of what problems need solving may affect which policy instruments are chosen. Moreover, refugee integration is very high on the political agenda, which may lead to policy initiatives that are particularly influenced by ideology, as politicians may be eager to put forward initiatives that resonate with the perceived ideology of their electorate.

Theoretical framing: problems and their perceived solutions

Within the ‘new politics’ tradition that became dominant in the 1990s, political change was generally understood to be slow and path dependant (Pierson, 2000). In the 2000s, a new interest in the theory of political change emerged, bringing forward new perspectives on policy learning and ideational change (see, for example, Blyth, 2001; Schmidt, 2008). In his comprehensive analysis of welfare state change Hemerijck (2013) acknowledges the important contributions of the ‘new politics’ tradition, but identifies a number of shortcomings. For example, it fails to identify social change, partially because of its bias towards large-scale social insurance programmes, meaning that other critical social institutions, such as labour market policy, education and vocational training, are overlooked. While cost containment, path dependency, veto points and adaptations to external drivers such as globalisation and demographic change (Pierson, 1994) may explain institutional inertia, these factors are less able to explain real change. Hemerijck (2013: 15) presents a more optimistic view of the adaptive and policy learning capacities of the welfare state:

It is my contention that dimensions of policy learning, the readiness to use information feedback from past performance, the input of new ideas and expertise, and the proliferation of inspiring reform successes across countries should count as important conduits of the welfare state.

This does not mean that Hemerijck’s understanding of the transition from policy evaluations to policy learning is simplistic. He underscores the importance of policy-makers’ cognitive and normative orientations, as well as positional interests. They may make mistakes, but such mistakes are not fatal if information feedback remains possible, allowing mistakes to be corrected. Their interest in learning may, however, depend on electoral competition: ‘without the chance of losing power and authority, there is no existential political incentive for policy-makers to learn’ (Hemerijck, 2013: 100).

The scope for policy learning is also addressed by Schmidt (2008), who contends that studies of policy ideas and discourse tend to have a bias towards assuming that good ideas that are supposedly more adequate to the task are likely to be chosen before bad ideas. As Schmidt states, sometimes good ideas fail and bad ideas prevail. One reason for this may be that in politics, programmatic success is not judged by social scientists! ‘Good’ ideas must be perceived as such first and foremost by policy-makers and citizens. Receptiveness to a new idea depends not only on its inherent persuasiveness, but also on how well the new idea resonates with individuals’ wider understandings. Taking Hemerijck’s point into account, this means that if the electorate is not persuaded by policy evaluations, politicians have no incentive to learn from such evaluations.

Other scholars have underlined that evaluations of policy measures in general, and of labour market activation measures in particular, can be both used and misused. They are methodologically challenging, often inconclusive; results are frequently adapted from one context to another; and research programmes and their interpretations are influenced by (politicians’) ideological frameworks and assumptions (Deeming, 2017; Dahler-Larsen, 2017). This clearly undermines the role of evaluations both as sources of ‘good’ ideas and in reducing the uncertainty that policy-makers and their electorate operate under. According to Blyth (2001), experiences with former policy measures can be decisive for whether ideas function as path-reinforcing ‘cognitive locks’, or as blueprints that provide political actors with a model for reform. The identification of failure or success is rarely straightforward, leaving a critical role for the ‘construction’ of success or failure in political discourse. Policy failure is also the main driver in Hall’s (Hall, 1993) version of paradigm shifts. According to Hall, a paradigm involves:

the framework of ideas and standards that specify not only the goals of policy and the kind of instruments that can be used to attain them, but also the very nature of the problems they are meant to be addressing. (Hall, 1993: 279)

Ideological standpoints shape actors’ understandings of problems and the solutions they arrive at. If policy measures are developed according to prevailing ideas about causal mechanisms, but still fail to produce the anticipated outcomes, politicians may be more prone to assuming that the treatment is correct, but insufficient, and opt for ‘more of the same’.

Discourse is the process of conveying ideas (Schmidt, 2008). Agenda setting and framing the problems that need solving is a source of political power in itself, as expressed by Bacchi (1999); the construction of what the problem ‘is’ is crucial for policy development. In this article, we do not conduct a discourse analysis of policy changes, but ideas (in discourses) will be discussed as one of several possible drivers for integration policy change. In line with Blyth (2001), we also consider the possibility that ideas might serve as cognitive locks that hinder policy change. In the United Kingdom, Levitas (2005) has identified three political discourses aimed at explaining social exclusion, each suggesting different solutions to the problem. In the ‘Redistribution Discourse’ (RED), social exclusion is understood to be a result of poverty, discrimination and demand-side problems in the labour market, resulting from imperfections in the capitalist economic system. In the ‘Moral Underclass Discourse’ (MUD), social exclusion is understood to be a result of lack of a work-ethic culture. Finally, in the ‘Social Integrationist Discourse’ (SID), the focus is on labour market participation as the main tool for social inclusion. In SID, the emphasis is typically on (short-term) activation measures, while structural reasons for social exclusion are given less attention. The relevance of these discourses for activation policy is evident: proponents of MUD would clearly support measures pertaining to the behavioural control pillar of activation policy, while proponents of SID would support social investment measures. Proponents of RED would probably be inclined to dismiss activation policy altogether and concentrate on demand-side measures and redistributive welfare benefits.

Labour market participation among refugees

As a group, refugees have lower employment rates than migrants who come to Norway for work or education. Employment differences between refugees and natives are even more significant. The labour market challenges faced by refugees are often attributed to trauma related to their motive for migration (flight). The context of reception can also delay their economic integration, as lengthy asylum and settlement procedures hinder early efforts to find employment. Although refugees make a slower start than other migrants, several studies indicate that this ‘refugee gap’ is much smaller for refugees who have been resident for many years (Bakker et al., 2017; Bevelander and Pendakur, 2014). Nevertheless, the gap does not disappear completely and refugees as a group do not achieve the same level of labour market integration as labour migrants or natives (Bevelander, 2016). There are also substantial differences in labour market integration between refugees from different countries and between male and female refugees, even when years of residence are accounted for (Bevelander and Pendakur, 2014; Bakker et al., 2017; Djuve et al., 2017). In addition, once labour market integration is achieved, the quality of the employment secured may vary between different refugee groups and between refugees and non-refugees.

The overall employment rate among refugees varies between 40 per cent in Spain and approximately 60 per cent in Italy and Sweden, with 55 per cent as the EU average (see Figure 1). In Sweden, Denmark and Norway, refugees’ employment levels are all equal to or above the EU average, while the rate in Finland is far lower. That said, in the Scandinavian countries, given high employment levels generally, the gap between refugees and the overall population is actually higher than in most other countries. As in the population in general, the employment level among refugees varies according to their level of formal education. Refugees with formal education enter the labour market to a greater degree than refugees with limited or no education (Connor, 2010). Individual human capital tends to fall when people cross national borders, however (Chiswick et al., 2005). This may be due to limited knowledge of the language and labour markets of the destination country, to different formal demands regarding occupational competence, to information problems or to discrimination (Borjas, 1995; Chiswick et al., 2005; Kanas and van Tubergen, 2009). Even though refugees with higher education seem to obtain lower returns on their education compared with natives, the positive link between formal education and paid employment is clear. According to Statistics Norway, in 2015 primary school (or lower) was the highest completed education of 53 per cent of adult refugees in Norway (aged 15–74). The same is true for just 26 per cent of natives (Olsen, 2017). The significantly lower level of formal education among refugees compared with natives has serious implications not only for their position in the labour market, but also for the effectiveness of existing introductory programmes.

Figure 1.

Employment rate by reason for migration (16–64 years), by country of residence, 2014.

What works for whom? A review of the literature

Effect evaluation of labour market programmes is methodologically challenging and, depending on the position of the reader, the results can be interpreted in different ways. The resulting uncertainty allows room for ideas and discourse to play a role. For our analysis, it is therefore important to survey the existing literature in order to assess the degree of certainty and uncertainty surrounding the effectiveness of different measures. We focus mainly on studies from the Nordic⁵ countries, as the distinctiveness of integration programmes and policies makes results difficult to compare across (very different) countries (González Garibay and De Cuyper, 2013). Also, the majority of effect evaluations are from the Nordic countries (Bilgili, 2015).

Language training

A central component of all introduction programmes is language training. This investment might not increase employment in the short term but can have positive long-term effects. Chiswick and Miller (2014: 87) argue that investment in destination-language proficiency is beneficial for immigrants and for society because immigrants who are proficient in the language of the destination country will be more successful in adjusting to its labour market. The scope, organisation and content of language training vary between countries. How many hours are offered and to whom? Is participation voluntary or compulsory? Does it take place in classrooms and/or at workplaces? Is language training free or do participants have to pay? All these variables are relevant in evaluating the effects of language training. The Nordic countries provide free and fairly comprehensive language training to refugees who participate in introductory programmes. Studies in Finland (Sarvimäki and Hämäläinen, 2016), Sweden (Delander et al., 2005), Denmark (Clausen et al., 2009) and Norway (Djuve et al., 2017) indicate that language training is an important component of introductory programmes leading to employment, but also that the quality of language training is subject to variations. Furthermore, results differ significantly according to participants’ education level. Djuve et al. (2017) highlight that only a very small proportion of refugees and immigrants with limited or no formal education pass even low-level language tests upon arrival in Norway.

Formal education

Human capital is not always easily transferrable across borders (Borjas, 1995). Formal education acquired in the host country, in particular vocational education, seems to be a more important predictor of employment than education acquired abroad (De Vroome and Van Tubergen, 2010). Studies that have investigated the employment effects of formal education report encouraging findings. For example, means-tested and subsidised vocational training in Germany significantly increased transfers to employment both among natives and immigrants (Huber et al., 2009). This finding is corroborated by Thomsen et al. (2013), who find that while training to improve job search skills is largely ineffective for immigrants, practical skills training at an early stage of the integration process has positive effects. The efficacy of vocational training is further supported by Dahlstedt and Bevelander (2010), who find that immigrants with a vocational education have a higher probability of employment than those with a general education. They also find that immigrants with destination country education are more successful in the labour market than those with an education from their country of origin. In Norway, vocational education for refugees and immigrants has so far only been a very minor part of the introductory programme (Djuve et al., 2017). Moreover, immigrants are generally underrepresented in these types of training programmes (Bernhard and Kruppe, 2012).

Active labour market programmes

Studies of active labour market programmes generally show that the most effective measures are ones that are closely linked to the labour market (see for example Butschek and Walter, 2014; Rinne, 2012). Because it is typically the most employable unemployed who are selected for such programmes, the findings may not reflect the efficacy of the measures generally. The same finding has however been produced by studies that control for this kind of selection. In Denmark, Clausen et al. (2009) analysed six active labour market programmes offered to newly arrived refugees and family immigrants, but found positive employment effects only in the context of private sector subsidised employment. Among non-Western immigrants with a longer residency in Denmark, Heinesen et al. (2013) found positive and fairly large employment effects from a wider range of ALMPs. In addition to wage subsidies, positive effects were related to direct employment programmes, as well as other programmes, including education, training and counselling. In Norway, Hardoy and Zhang (2010) found positive employment effects for immigrants both in the contexts of subsidised employment and shorter training activities, while on-the-job work training – the most common measure – had no effect. This finding has been replicated in a more recent study of the NIP, in which Djuve et al. (2017) found no effect of on-the-job work training.

As already mentioned, evaluation of individual programme elements is complicated by a strongly biased selection of the different measures; contributions that do not take selection bias into consideration report a strong correlation between certain programme elements and transition to employment (Blom and Enes, 2015; Fasting and Riekeles, 2016). After controlling for selection, Djuve et al. (2017) found no statistically significant effects of the same measures.

Figure 2 demonstrates the weak correlation between the most commonly applied programme measures and transition to the labour market, using a municipal-level benchmark analysis. Based on the characteristics of the participants in each municipality, we have estimated an expected municipal rate of labour market participation four years after participants started the NIP. This estimated labour market participation rate is then deducted from the actual municipal labour market participation rate four years after starting the programme. We then add the average labour market participation rate for all participants in order to avoid the curve crossing the X axis. The resulting curve can be termed the ‘benchmark-adjusted municipal labour market participation rate’. When we also add the curves that illustrate the municipal rates of refugees who have participated in on-the-job work training or on-the-job language training, two findings emerge: (i) there is a large variation in the municipal use of these (core) measures, and (ii) there seems to be no consistent effect of the use of either measure on labour market outcomes.

Figure 2.

Percentage in paid employment four years after starting the NIP (benchmark adjusted) and percentage of participants who have participated in on-the-job work training or on-the-job language training, by municipality (x-axis).

In a meta-analysis of the employment effects of active labour market programmes for immigrants, Butschek and Walter (2014) condense findings from 33 empirical studies on the effects of training, job-search assistance, wage subsidies and subsidised public sector employment. They concluded that ‘subsidized employment in the private sector is significantly more likely to have a positive effect on immigrants’ labour market outcomes than training’ (2014: 2). For the most commonly used ALMPs, there are mostly insignificant results. Subsidised employment in the private sector is offered mainly to unemployed immigrants with a high degree of employability and seldom to newly arrived refugees (Clausen et al., 2009; Djuve et al., 2017). Outcomes of training and education differ according to background characteristics. This means that effects identified for one group may not be transferable to other groups.

Counselling

An important aspect of introduction programmes in the Nordic countries is the ambition to offer participants training tailored to their individual needs. Studies in Sweden, Finland and Norway indicate that a stronger focus on individually adapted training is effective. In Sweden, a six-month workplace introduction programme (SIN), targeting immigrants as well as refugees, showed increased transitions from unemployment to work experience schemes, as well as improved future employment probabilities (Åslund and Johansson, 2011; Joona and Nekby, 2012). A central component in the programme set-up was to reduce the caseload of caseworkers so that they could provide better counselling to each participant, combined with a careful matching between job-seekers and employers. The effects of aptitude tests assessing the suitability of participants in terms of skills, capability and labour market opportunities for specific occupations are also found to be positive (Thomsen et al., 2013). Sarvimäki and Hämäläinen (2016) argue that the introduction of ‘integration plans’ for unemployed immigrants in Finland had substantial employment effects because immigrants could now attend programmes that were better adapted to their training needs. Immigrants spent more time in language courses and training specifically designed for immigrants, and less time in traditional ALMP schemes, such as job-seeking courses. In Norway, close follow-up of NIP participants (Kavli et al., 2007) and limitations on the workload of caseworkers (Djuve et al., 2017) are positively related to the transfer of participants to employment.

Conditionality and sanctions

In activation programmes, behavioural control typically comes in the shape of economic sanctions. Research generally indicates that a reduction in benefit payments, or the expectation of such reductions, will push benefit recipients towards desired behaviours (McVicar et al., 2016). Sanctions affect welfare recipients in different ways, however. In 2002, the Danish government lowered benefit payments to refugees by approximately 35 per cent. While this had a (slight) effect on employment rates, it was those with the poorest labour market prospects (women and low educated) who were least responsive to reductions in benefit levels (Rosholm and Vejlin, 2010). In the overall population, Qureshi (2013: 244) shows that welfare recipients from less developed countries respond less to punitive sanctions than Danes. He suggests that welfare recipients may not have the qualifications necessary to secure employment, meaning that sanctioning them will have limited effect. In Sweden, adult immigrants who finished their language training on time were given an economic reward. But while there was a clear impact of the bonus among the most skilled students on the most advanced courses, no impact was found among weaker students with poor prospects of completing in time (Åslund and Engdahl, 2017). Larsen (2005) states that the motivational effect of activation is considerable, but that it has been shown to work only among unemployed people with insurance, not for social assistance claimants. In other words, both positive and negative economic sanctions seem to push primarily those participants who are closest to the finish line in the desired direction. Also, for individuals who do respond to the sanctions, Berg and Vikström (2014: 330) find that on average they ‘accept jobs with a lower hourly wage and fewer working hours per week’.

When the target group consists of newly arrived immigrants, behavioural control increasingly comes in the form of civic requirements, such as participation in introductory programmes or passing a language or citizenship test (Goodman and Wright, 2015). There seems to be limited evidence of the effects of these measures, however. Based on data from the European Social Survey, Goodman and Wright found limited links between civic requirements and long-term integration and concluded that these policy initiatives serve a symbolic and strategic narrative rather than a functional one.

In summary, language training is an important component of effective introduction programmes, but the quality of training varies and there is limited knowledge of the effects of various methods. Formal education in general, and vocational education in particular, shows promising results, but immigrants and refugees are seldom offered these measures. Subsidised employment shows promise for some groups, but little is known about the potential effects for participants with low employability. On-the-job training, a more common measure, yields poor results. Moreover, while a stronger emphasis on individually adapted training increases transfers to employment, conditionality and sanctions show varied effects, with higher transitions out of unemployment for the most resourceful and limited or no effects for others.

Changes in refugee integration policy since the introduction of the NIP

The 2003 Introduction Act introduced more rights, but also more obligations for refugees who receive residence in Norway. Newly arrived adult (18–55 years) refugees and their families were given the right to participate in a two- to three-year, full-time (30–37.5 hours per week) training programme. They receive a modest fixed income during the programme period (NOK193,766 / €20,257 per year in 2018). This income is contingent on full participation in set programme activities and undocumented absences result in an hour-by-hour deduction from benefits. In this way, the government effectively made the programme obligatory, as refugees and their families have limited alternative income opportunities upon arrival in Norway (Hagelund and Kavli, 2009). The NIP represents a far more extensive human-capital investment than prior policies, but also a more explicit system of economic sanctions. In this way, the dual nature of activation has become more evident with regard to a target group previously less exposed both to intensive training and to economic sanctioning.

Between its introduction in 2003 and June 2018, several amendments have been made to the NIP, and other policy changes have affected the rights and obligations of NIP participants (see Table 1). We categorise the most important of these changes as belonging to either the social investment pillar of activation or to the behavioural control pillar. We then discuss whether the policy initiatives represent change or ‘more of the same’.

Table 1.

Policy changes and initiatives in the introductory programme, 2003–2018.

	Social investment	Behavioural control
1. Language training	–	2005: 300 hours obligatory 2011: 600 hours obligatory 2013: obligatory to take test 2017: obligatory to pass test
2. Formal education	2016: formal right for participants to attend full-time secondary education	–
3. Counselling	2018: underscoring of the need for coherent qualification mapping	2018: central government may instruct local municipalities on how to sanction participants
4. Wage subsidies	–	–
5. Other ALMPs	2017: ‘fast track’ for refugees with relevant labour market qualifications 2018: underscoring of work- or education-oriented measures in addition to language training	–
6. Income support/welfare benefits for programme participants	2018: underscoring of possibility of ‘third year’	2016 and 2017: curbing of rights to other welfare benefits

Language training

At the administrative level, a number of initiatives have been launched to facilitate local efforts to improve language training for immigrants. The efforts of administrative units are obviously to some extent politically initiated, but at our level of analysis – changes in laws and regulations – no new initiatives on the quality or organisation of language training were launched in the period of analysis. Obligatory participation in and completion of language training have, however, been targeted (see the section on sanctions below).

Formal education

Since the implementation of the NIP, participants without completed primary education have had formal access to primary education for adults while receiving the introductory benefit. It has also been possible to include individual subjects from secondary education in the programme. In practice, very few participants have benefited from either of these opportunities. Participation in secondary education is often halted because the participants have not completed primary education and because the caseworkers are reluctant to enrol NIP participants in educational tracks that they will not be able to finish before the programme period and the income support run out. Also, a lack of adaptation of the curriculum for participants with poor Norwegian language skills is a challenge at both primary and secondary levels (Djuve et al., 2014). In 2016, a policy change was implemented that made it formally possible to participate in secondary education full time, while still being enrolled in the NIP. The other barriers to participation have not been targeted. We consider this change to belong within social investment; it provides a formal opportunity.

Active labour market programmes

At the administrative level, efforts have been made to increase the use of on-the-job work training and on-the-job language training; meanwhile policy initiatives have recently been introduced to support these efforts. From September 2018, the introductory law will be amended to make it clearer that measures to prepare participants for employment or education are part of the minimum requirements for the NIP. This option is already specified in the act, but the ministry responsible for the NIP interprets the municipal variations in the use of these measures as a sign of ambiguities in the present law text. A new wording is therefore used to underscore that language training alone will not satisfy the requirements specified in the legislation.⁶ Furthermore, the ministry specifies that local integration authorities must explain and justify the choice of work- or education-oriented measures for each participant. In 2017, an initiative was launched to increase the use of employment subsidies for highly skilled refugees, with the so-called ‘fast track’. This was one of the initiatives from labour market organisations and labour authorities that came in response to the 2015 refugee situation. With the exception of the ‘fast track’ (which, according to NAV, and as referred to in the Norwegian media, included only 120 refugees in 2017⁷), these proposals hardly reflect actual policy changes, but are rather path-dependent ‘more of the same’ initiatives. They clearly pertain to the social investment pillar and are ‘soft’ regulations.

Counselling

The law contains no regulation of the maximum number of clients per counsellor. In the latest policy initiative from the Ministry of Education, however, it is underscored that individual training programmes should be based on a mapping of the qualifications acquired before resettlement in a municipality. This is a ‘soft’ policy initiative that can be interpreted as an attempt to improve the efficiency of counselling.

Conditionality and sanctions

Between 2003 and 2018, the behavioural control elements in the programme have increased in both number and consequence. In 2005, the ministry responsible for the NIP for the first time specified a minimum number of hours the participants had to complete in order to obtain a permanent residence permit. It was set at 300 hours, and in 2011 increased to 600 hours. The changes did not affect the right to participate; NIP participants already had the right to participate in a lot more than 600 hours of free language training (up to 3000 hours for participants with little or no former education). We therefore categorise the introduction of a 300-hour and then a 600-hour minimum as a change to a sanctioning measure rather than a change to language training. The tightening of the sanctioning regime continued with the 2013 obligation not only to participate in training, but also to take a language test. In 2017, permanent residence permits were made contingent on passing the test. In 2018, another change specified that the ministry can instruct local municipalities on how to sanction participants who do not comply with the rules on the completion of tests in Norwegian language skills and citizenship studies.

A few changes have also been made to the NIP participants’ rights to other welfare benefits. In 2016, the introductory benefit was coordinated with the single parent benefit (overgangsstønad) so that single parents in the NIP in effect suffered a reduction in total benefit levels and in 2017 eligibility for the cash support (kontantstøtte) benefit for parents whose children do not go to nursery was limited to people with more than five years of residence.

No changes have been made to the size of or entitlement to the introductory benefit, but in the amendment initiated by the Ministry of Education (adopted June 2018) it was emphasised that the programme (and benefit) can be extended to three years. This possibility already existed and the amendment does not suggest any measures to encourage municipalities to make use of it. It is a ‘soft’ regulation and belongs to the social investment pillar.

Discussion: good ideas or bad?

In this section, we discuss our findings in light of the theories of institutional change presented in the theoretical framing. Table 1 offers a starting point. Changes have been made in both the social investment pillar of activation and the behavioural control pillar. It is, however, striking that all the changes made to the social investment pillar are soft regulations, and, in reality, will affect a small number of participants. All the changes in the behavioural control pillar are hard regulations that will affect all participants, with the possible exception of the 2018 amendment concerning the ministry’s authority to instruct local municipalities. It seems fair to conclude that the behavioural control pillar has been strengthened, while not much has happened with the measures in the social investment pillar. Djuve (2011) pointed to a mix of SID, MUD and RED discourses in the original articulation of the NIP as one of the reasons why this fairly radical change to integration policy gained sufficient political support. Since the NIP was introduced in 2003, however, we find that the amendments to the NIP seem to reflect MUD more than SID; in other words, the policy initiatives seem to focus more on control and sanctions than on programme content and programme quality.

In the national context, several evaluations have pointed to large variations in local programme quality, limited use of primary and secondary education as part of the programme, and less focus generally on programme content and quality than on participation in and completion of the programme (see, for example, Bredal and Orupabo, 2014; Djuve et al., 2017; Kavli et al., 2007; Tronstad, 2015). Administrative integration authorities (Directorate of Integration and Diversity, Skills Norway) are actively using research results in the shaping of activation measures, but the Introductory Act and political signals limit the level and scope of these changes. At the political level, regulations directed at the quality or composition of the programme have been scarce and soft since the more substantial changes made when the law was adopted in 2003.

Within the social investment pillar, policy initiatives are firmly positioned within a path-dependent trajectory; they prescribe ‘more of the same’. Accordingly, types of ALMP are still being applied in the NIP that national and international evaluations more or less unanimously consider to have little or no effect on the transition to employment for newly arrived refugees, as are methods of language training that have not been systematically evaluated. The universal finding that formal education is of crucial importance has not led to political initiatives directed towards increasing educational activities, other than to formally introduce secondary education as part of the programme. Given the significant obstacles to newly arrived refugees enrolling in ordinary secondary schools (Djuve et al., 2014), however, this initiative might not have much impact on actual educational activities. Also, no policy initiatives have been made in order to test systematically alternative methods of language training for participants with little or no former education, even though it is well known that mastering the host-country language represents a long-term and substantial challenge for this group.

In other words, the policy initiatives poorly reflect mainstream understandings among social scientists. It seems that policy learning is not reflected in either the social investment pillar or the behavioural control pillar. Policy changes in the behavioural control pillar can, however, be seen to reflect mainstream understandings in a social policy discourse that tends to explain social exclusion as resulting from a lack of motivation at the individual level. If the lack of motivation is conceived as the reason for social exclusion, sanctions seem to be an appropriate measure. Furthermore, if poor programme results are interpreted within this paradigm, there is no enticement to search for alternative explanations. Here, ideas seem to function both as a cognitive lock resulting in path-dependent inertia, and as a blueprint for changes that are little supported by social science.

Following Hemerijck (2013), the absence of policy learning indicates that policy-makers in office are not particularly worried that their combination of ‘more of the same’ and more sanctioning may cause a decline in support among their electorate. As long as ideas belonging to ‘MUD’ resonate well with their electorate, politicians have little incentive for policy learning. It can also be argued that the introduction of more conditionality and sanctions in the NIP is motivated by a desire to communicate a stricter approach in immigration- and refugee-integration policy rather than an ambition to increase the employment of refugees. The large and – for the general public – surprising influx of refugees in 2015 moved immigration and integration policy even higher up the political agenda. Changes may also reflect an ambition to deter further migration by making Norway a less attractive destination. The potential for uncertainty must also be considered when evaluating effects. There will always be some uncertainty concerning the effects of programme elements, particularly within ALMPs. The great variety of measures and participants, coupled with methodological difficulties and sometimes mixed research findings, are reasons for caution when it comes to dismissing ALMPs altogether. With regard to the effects of formal education, however, there does not seem to be much uncertainty. Here, it seems that policy-makers have chosen not to learn, or at least not to act. They cannot claim not to have been informed; the 2018 proposition quotes several studies that underscore the importance of formal education (Djuve et al., 2017; Norges offentlige utredninger (NOU), 2017).

There are a number of other path-reinforcing mechanisms in play that complicate the launching of policy initiatives that would make formal education a national and large-scale ambition of the NIP. Cost containment is one of them. Municipal autonomy with regard to integration is another. In addition, the division of responsibility between levels of governance and policy sectors has so far been a severe obstacle to local initiatives aimed at providing tailored secondary education to newly arrived refugees.

Conclusion

Across Europe, labour market programmes and initiatives to integrate refugees and other immigrants into the labour market have grown into a substantial industry. Evaluators struggle to identify effective methods and programmes, however, and their evaluations do not necessarily lead to policy learning. Policy learning is not absent, but seems to have been hindered by cognitive locks. One ‘cognitive lock’ in the case of integration policy appears to be the lack of a discussion on the measures’ suitability, given the characteristics of their target group. It seems that policy-makers have forgotten to pose one very basic question: what kind of problem is on-the-job training intended to solve? In the Norwegian labour market, in which wages are relatively high even at the lower end of the wage distribution, very few jobs are available for job-seekers without a formal education. As already indicated, however, it may be the case that policy-makers’ primary concerns are electoral support and/or the influx of migrants, and not necessarily unemployment among refugees.

Either way, developments in Norwegian integration policy since 2003 do not seem to be characterised chiefly by processes of policy learning, at least not in the rationalist understanding of learning through processes of ‘feedback’ and ‘steering’ (Deutsch, 1963). Neither do they seem to be the result of ‘puzzling out’ in situations of uncertainty (Heclo, 1974). Although competing ‘regimes of truth’ are persistently present in the public debate (Clarke, 2004), uncertainty about what works does not seem to be a critical factor. The political changes we have identified seem to be compatible with understandings of unemployment as the ‘problem’ and a lack of motivation as its cause, indicating that ideas and discourses on the reasons for social exclusion matter in policy-making. This is happening within a fairly extensive and costly integration regime, with few signs of a general retrenchment of training activities. Policy-makers do not disregard the need for training, but when activation programmes fail, their emphasis on individual behaviour as the main reason for social exclusion leads them to look for solutions associated with stronger incentives rather than to review critically the quality of programme contents, even when policy evaluations suggest that the latter would be a good idea. Also, it is likely that they believe that the electorate has a similar understanding.

Policy learning is often initiated by poor programme outcomes. It may be, however, that the ‘MUD’ (‘Moral Underclass Discourse’) is inherently resistant to results-initiated policy learning. In ‘MUD’, poor programme results may always be interpreted as the result of poor work incentives and hence as affirming the initial views of its proponents. This way, the paradigm becomes self-reinforcing and resistant to policy learning.

As a final point, it should be mentioned that recent political signs indicate that formal education might be given a more prominent role in future Norwegian integration policy. In 2018, political responsibility for the NIP was transferred from the Ministry of Justice and Public Security to the Ministry of Education and Research. This might indicate a new approach to the role of formal education in integration policy. Also, both the Minister of Labour and Social Affairs and the Minister of Education recently announced their ambition to increase the share of newly arrived refugees who participate in formal education. Political initiatives designed to fulfil these ambitions have however yet to be launched.

Footnotes

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Notes

Appendix

Benchmark model

	B	Sig.
Women	−.658	.000
Age 20–30 (ref. 50+)	1.418	.000
Age 30–40	1.493	.000
Age 40–50	.978	.000
Norwegian language track 1 (ref. 2 & not registered)	−.402	.000
Norwegian language track 3 (ref. 2 & not registered)	.203	.079
Primary school	.089	.309
Upper secondary school	.369	.001
University/University college	.217	.049
Eritrea	.614	.000
Somalia	−.698	.000
Afghanistan	.184	.061
Iraq	−.381	.000
Palestine	−.585	.000
Married (2014; ref. not married)	.293	.000
Female × number of children after 2011	−.631	.000
Number of children under 18	−.393	.000
Number of children after 2011	.517	.000
Child abroad	−.211	.022
Year of arrival in Norway	−.074	.000
Rural/urban municipality	−.197	.005
Number of relocations between municipalities	.074	.000
Municipal unemployment level (2014)	−.180	.000
Constant	147.514	.000

References

Åslund

Engdahl

(2017) The value of earning for learning: Performance bonuses in immigrant language training. Economics of Education Review 62(issue C): 192–204.

Åslund

Johansson

(2011) Virtues of SIN: Can intensified public efforts help disadvantaged immigrants? Evaluation Review 35(4): 399–427.

Bacchi

(1999) Women, Policy and Politics: The Construction of Policy Problems. London: Sage.

Bakker

Dagevos

Engbersen

(2017) Explaining the refugee gap: a longitudinal study on labour market participation of refugees in the Netherlands. Journal of Ethnic and Migration Studies 43(11): 1775–1791.

Berg

Vikström

(2014) Monitoring job offer decisions, punishments, exit to work, and job quality. The Scandinavian Journal of Economics 116(2): 284–334.

Bernhard

Kruppe

(2012) Effectiveness of further vocational training in Germany: empirical findings for persons receiving means-tested unemployment benefit. IAB Discussion Paper 2012 10. Nuremberg: IAB.

Bevelander

(2016) Integrating refugees into labor markets. IZA World of Labor 269. DOI: doi: 10.15185/izawol.269

Bevelander

Pendakur

(2014) The labour market integration of refugee and family reunion immigrants: A comparison of outcomes in Canada and Sweden. Journal of Ethnic and Migration Studies 40(5): 689–709.

Bilgili

(2015) Evaluating Impact: Lessons Learned from Robust Evaluations of Labour Market Integration Policies. Barcelona: The Barcelona Centre for International Affairs and the Migration Policy Group.

10.

Blom

Enes

(2015) Introduksjonsordningen – En Resultatstudie. Rapporter 2015/36. Oslo/Kongsvinger: Statistisk sentralbyrå.

11.

Blyth

(2001) The transformation of the Swedish model: Economic ideas, distributional conflict, and institutional change. World Politics 54(1): 1–26.

12.

Borjas

(1995) Assimilation and changes in cohort quality revisited: What happened to immigrants’ earnings during the 1980s? Journal of Labour Economics 13(2): 201–245.

13.

Bredal

Orupabo

(2014) Drammen Som Introduksjonsarena: En Gjennomgang av Kommunens Introduksjons- Og Kvalifiseringsarbeid for Nyankomne Innvandrere. Institutt for samfunnsforskning Rapport 2014: 4. Oslo: Institutt for samfunnsforskning.

14.

Butschek

Walter

(2014) What active labour market programmes work for immigrants in Europe? A meta-analysis of the evaluation literature. IZA Journal of Migration 3: 48.

15.

Chiswick

Lee

Miller

(2005) Immigrant earnings: A longitudinal analysis. Review of Income and Wealth 51(4): 485–503.

16.

Chiswick

Miller

(2014) International migration and the economics of language. IZA Discussion Paper No. 7880.

17.

Clarke

(2004) Dissolving the public realm? The logics and limits of neo-liberalism. Journal of Social Policy 33(1): 34.

18.

Clausen

Heinesen

Hummelgaard

. (2009) The effect of integration policies on the time until regular employment of newly arrived immigrants: Evidence from Denmark. Labour Economics 16(4): 409–417.

19.

Connor

(2010) Explaining the refugee gap: Economic outcomes of refugees versus other immigrants. Journal of Refugee Studies 23(3): 377–397.

20.

Dahler-Larsen

(2017) Critcal Perspectives on Using Evidence in Social Policy. In: Greve

(ed.) Handbook of Social Policy Evaluation. Cheltenham: Edward Elgar, pp. 242–262.

21.

Dahlstedt

Bevelander

(2010) General versus vocational education and employment integration of immigrants in Sweden. Journal of Immigrant & Refugee Studies 8(2): 158–192.

22.

De Vroome

Van Tubergen

(2010) The employment experience of refugees in the Netherlands. International Migration Review 44(2): 376–403.

23.

Deeming

(2017) Uses and misuses of evaluation in social policy. In: Greve

(ed.) Handbook of Social Policy Evaluation. Cheltenham: Edward Elgar, pp. 161–182.

24.

Delander

Hammarstedt

Månsson

. (2005) Integration of immigrants: The role of language proficiency and experience. Evaluation Review 29(1): 24–41.

25.

Deutsch

(1963) The Nerves of Government: Models of Political Communication and Control. New York: The Free Press.

26.

Djuve

(2011) Introduksjonsordningen for nyankomne innvandrere: Et integreringspolitisk paradigmeskifte? [The introductory programme for newly arrived refugees: A paradigm shift in integration policy?] Phd Thesis, University of Oslo.

27.

Djuve

Grødem

(eds) (2014) Innvandring og arbeidsmarkedsintegrering [Immigration and Labour market integration]. Fafo-rapport 2014:27. Oslo: Fafo.

28.

Djuve

Haakestad

Sterri

(2014) Rett til utdanning? Grunnskoleopplæring og videregående opplæring som tiltak i introduksjonsordningen for nyankomne innvandrere. Fafo rapport 2014:34. Oslo: Fafo.

29.

Djuve

Kavli

Sterri

. (2017) Introduksjonsprogram og Norskopplæring. Hva virker for hvem? [Introductory programs and language training. What works - for whom?]. Fafo rapport 2017:31. Oslo: Fafo.

30.

Dølvik

Fløtten

Hippe

. (2015) The Nordic model towards 2030. A new chapter? NordMod2030. Final report. Fafo-report 2015:07. Oslo: Fafo.

31.

European Commission (2016) Employment and Social Development in Europe 2016. Brussels: European Commission.

32.

Fasting

Riekeles

(2016) Flyktninger og arbeid [Refugees and employment]. Oslo: Civita.

33.

González Garibay

De Cuyper

(2013) The Evaluation of Integration Policies Across the OECD: A Review. Antwerp: Steunpunt Inburgering en Integratie.

34.

Goodman

Wright

(2015) Does mandatory integration matter? Effects of civic requirements on immigrant socio-economic and political outcomes. Journal of Ethnic and Migration Studies 14(12): 1885–1908.

35.

Hagelund

Kavli

(2009) If work is out of sight. Activation and citizenship for new refugees. Journal of European Social Policy 19(3): 259–270.

36.

Hall

(1993) Policy paradigms, social learning, and the state: the case of economic policymaking in Britain. Comparative Politics 25(3): 275–296.

37.

Hardoy

Zhang

(2010) Innvandrere i arbeid: Hjelper arbeidsmarkedstiltak? Søkelys på arbeidslivet 27(4): 343–364.

38.

Heclo

(1974) Modern Social Politics in Britain and Sweden: From Relief to Income Maintenance. New Haven and London: Yale University Press.

39.

Heinesen

Husted

Rosholm

(2013) The effects of active labour market policies for immigrants receiving social assistance in Denmark. IZA Journal of Migration 2: 15.

40.

Hemerijck

(2013) Changing Welfare States. Oxford: Oxford University Press.

41.

Huber

Lechner

Wunsch

. (2009) Do German welfare-to-work programmes reduce welfare and increase work? IZA Discussion Paper No. 4090.

42.

Joona

Nekby

(2012) Intensive coaching of new immigrants: An evaluation based on random program assignment. The Scandinavian Journal of Economics 114(2): 575–600.

43.

Kanas

van Tubergen

(2009) The impact of origin and host country schooling on the economic performance of immigrants. Social Forces 88(2): 893–915.

44.

Kavli

Hagelund

Bråthen

(2007) Med rett til å lære og plikt til å delta. En evaluering av introduksjonsordningen for nyankomne flyktninger og innvandrere. Fafo rapport 2007:34. Oslo: Fafo.

45.

Larsen

(2005) The Active Society and Activation Policy: Ideologies, Contexts and Effects. In: Goul Andersen

Guillemard

Jensen

Pfau-Effinger

(eds) The Changing Face of Welfare: Consequences and Outcomes from a Citizenship Perspective. Bristol: Policy Press, pp. 135–150.

46.

Levitas

(2005) The Inclusive Society? Social Exclusion and New Labour. Basingstoke: Palgrave.

47.

McVicar

Wooden

Leung

. (2016) Work-related training and the probability of transitioning from non-permanent to permanent employment. British Journal of Industrial Relations 54(3): 623–646.

48.

Norges offentlige utredninger (NOU) (2017) NOU 2017:2. Integrasjon og tillit. Langsiktige Konsekvenser av høy Innvandring. Oslo: Justis- og beredskapsdepartementet.

49.

Olsen

(2017) Innvandrere i og utenfor arbeidsmarkedet [Immigrants in Norway]. Oslo: Statistics Norway.

50.

Pierson

(1994) Dismantling the Welfare State? Reagan, Thatcher and the Politics of Retrenchment. Cambridge: Cambridge University Press.

51.

Pierson

(2000) Increasing returns, path dependence, and the study of politics. The American Political Science Review 94(2): 251–267.

52.

Qureshi

(2013) The effect of punitive sanctions on the transition rate from welfare to work: Evidence from Denmark. Nationaløkonomisk Tidsskrift 151: 225–246.

53.

Rinne

(2012) The evaluation of immigration policies. Discussion Paper series. Forschungsinstitut zur Zukunft der Arbeit, No. 6369. Available at: http://nbn-resolving.de/urn:nbn:de:101:1-201206146426 (accessed 7 May 2018).

54.

Rosholm

Vejlin

(2010) Reducing income transfers to refugee immigrants: Does start-help help you start? Labour Economics 17(1): 258–275.

55.

Sarvimäki

Hämäläinen

(2016) Integrating immigrants: The impact of restructuring active labor market programs. Journal of Labor Economics 34(2): 479–508.

56.

Schmidt

(2008) Discursive Institutionalism: The Explanatory Power of Ideas and Discourse. Annual Review of Political Science 11(1): 303–326.

57.

Taylor-Gooby

(2007) Ideas and Welfare State Reform in Western Europe. Basingstoke: Palgrave.

58.

Thomsen

Walter

Aldashev

(2013) Short-term training programs for immigrants in the German welfare system: Do effects differ from natives and why? IZA Journal of Migration 2: 24.

59.

Tronstad

(2015) Introduksjonsprogram for flyktninger i norske kommuner. Hva betyr organiseringen for overgangen til arbeid og utdanning? NIBR-rapport 2015:2. Oslo: NIBR.