A social investment perspective on labour market policies for vulnerable groups

. This article conceptualises the notion of “social investment’ within the context of social protection and active labour market policies. This paradigm is confronted to the neoliberal ‘making work pay’ paradigm and the conservative workfare doctrine built on the idea that a stronger work ethos needs to be inculcated through reduced rights and more duties. The social investment paradigm advocates generous social protection combined with enabling investments and is thus perfectly congruent with social inclusion objectives. The empirical analysis shows remarkably small effects of both types of policies in the past 15 years: ALMPs have had small positive effects on employment, but also on exits into inactivity. The most positive employment effect of ALMPs is a redistribution of employment opportunities towards older, female and less educated groups. The Re-inVEST research analyses in-depth participatory case studies of policies targeting disadvantaged groups in different European countries, using an ‘enriched’ social investment model, building on a human rights and capabilities approach. The findings indeed show very wide differences in quality – and a lot of room for improvement. In order to improve their effectiveness as well as inclusiveness, the right to decent ALMPs should be put on the policy agenda.


Introduction
The idea of the 'active welfare state' dates back to the post-war period in the Nordic countries, but conquered the rest of Europe during the 1980s, as the traditional welfare state got off-track. Massive unemployment, concentrated among vulnerable groups, put a high burden on the social protection budgets and necessitated a shift from redistribution of income to work as the main leverage for social inclusion. In the next decades, the discourse on the active welfare state diversified under the influence of various ideologies. Whereas all Western governments agreed on the need for more active labour market policies, neoliberal governments put more emphasis on 'incentives'; conservatives on duties and individual responsibility; while social-democrats tend to use a rights discourse and justify their approach in terms of social investment. This paper aims to develop the concept of social investment in the context of labour market integration, building on a human rights and capabilities approach. It draws lessons with regard to quality criteria for labour market integration services.

The neoliberal view: making work pay
The economics of social security (Atkinson & Micklewright 1991;Barr 2001) are closely connected with the neoclassical theory of labour supply, which considers individual labour supply decisions as the result of a trade-off between income and leisure. In this context, generous social benefits are seen as potentially harmful because such regulations may reduce the incentives to re-enter the labour market (moral hazard effect). Hence, the theory prescribes that a sufficient gap must be maintained between earnings from work and social benefits for non-working people in order to 'make work pay'. Note that this argument applies to any kind of replacement income (unemployment benefit, sickness benefit, minimum income, early retirement etc.).
An additional, typical problem of guaranteed minimum income (GMI) schemes highlighted in this approach is the poverty trap effect. Given that benefits are means-tested and just meant to top up primary income, up to a fixed income threshold, each euro earned below that threshold is deducted from the benefit. This means that beneficiaries do not gain (and indeed, actually lose leisure time) when they take up small jobs. In economic jargon, the 'effective marginal tax rate on earnings' below the GMI threshold equals 100%, which supposedly completely discourages work.
These assumptions concerning human behaviour have informed the 'making work pay' agenda, which made up the core of the OECD's Jobs Study (OECD, 1994). The message was copied by the European Commission as one of the four pillars of its agenda for the modernisation of social protection (EC, 1997). The EU Employment Guidelines, developed around the turn of the millennium, urged Member States to "review and, where appropriate, reform their benefit and tax systems to reduce inactivity traps, and provide incentives for unemployed or inactive people to seek and take up work".
Neoliberal 'making work pay' policies are largely based on the assumption that individual behaviour can be 'steered' by financial incentives. Examples of such incentives include: benefit cuts, in-work benefits, negative income tax schemes, limited duration of benefits etc. The most notorious examples of such reforms in the EU are the UK's Universal Credit and the French Revenu de Solidarité Active.
The effectiveness of 'making work pay' policies is a hot topic in the literature. From a social inclusion perspective, the notion of 'effectiveness' refers to two key criteria: the employment effect and the poverty reduction effect. To some extent both effects reinforce each other, but in many countries rising employment went along with rising poverty in the past decades. In its reassessment of the Jobs Strategy after a decade, the OECD openly admitted that cutbacks in social benefits may be held responsible for increased poverty in some countries: "Cutting benefit levels and their duration have succeeded in raising work incentives but, beyond certain thresholds, this may compromise social objectives" (OECD 2006: 10). As a consequence, the revised OECD Jobs Strategy has shifted away its emphasis from restricting unemployment benefits to other activation measures.
Whereas by and large empirical research appears to confirm that there is some effect of benefit design on labour supply, the effect is often small to insignificant and indeed less positive on the macro level than on the individual level. (EC 2012: 91-95;EC 2013b: 161-167;Lehweß-Litzmann & Nicaise, 2020). For example, at country level, a slight positive correlation was found between the coverage rate of job seekers by unemployment benefits and the rate of re-entry into work -even after controlling for other determinants. Likewise, a positive correlation was found between the replacement rate (in unemployment benefits) and the probability of re-entry into work. More importantly, generous benefits systems (with decent replacement rates and a longer duration of entitlements) appear to protect job seekers from poverty and/or to foster exits from poverty.
The Commission's Employment and Social Developments in Europe 2013 report also focussed on the 'poverty trap effect' in social assistance and found the opposite of what economic textbooks predict, namely, that more generous social assistance (GMI) systems act as springboards out of poverty: more generous GMI benefit systems tend to result in quicker exits from poverty, rather than more persistent dependency.
At least five explanations can be given for these -at first glance, paradoxicalfindings: -Firstly, as pointed out by Immervoll et al. (2009), work incentives targeting exclusively the supply side of the labour market (job seekers) have little (if any) effect as long as demand constraints dominate. When jobs are not available, incentives for job seekers to step up search efforts may only lead to greater frustration rather than higher employment. Bargain et al. (2005) come to the same conclusion in their evaluation of the German Mini-job reform (Hartz IV).
-Secondly, the inverse correlations in figure 2 may be spurious, as unobserved third factors may affect both covariates in each of the graphs. For example, countries with more developed social protection and GMI schemes may also spend more on active labour market policies. After all, this is not really a spurious correlation, because more generous social protection generates a stronger sense of belonging and citizenship on the part of recipients as well as a stronger pressure on governments to avoid the cost of unemployment through active labour market policies. Hence, efforts on both sides are encouraged for re-integration into the labour market. -Thirdly, a distinction must be made between incentive effects at micro-level and the macro-economic impact of social protection, and between short-and longer-term effects. In less developed social protection / assistance systems, the financial pressure from low benefits and short duration of entitlement on individual job seekers may indeed accelerate their exit from unemployment, but this may be offset by substitution and revolving door effects at macro-level. Substitution effects mean that the waiting queues for jobs are simply re-shuffled as desperate job seekers just 'jump over the heads of others', with no net job creation effect. Revolving door (or 'churning') effects mean that job seekers take up the first job within their reach but fall back into unemployment soon, as their skills and aspirations do not match with the job requirements and content. -Fourthly, extreme competition for jobs between under-protected job seekers exerts a downward pressure on wages and working conditions at the 'bottom' of the labour market, resulting in a further erosion of jobs and increased (inwork) poverty risks. -Finally, maybe some key assumptions of the 'making work pay' paradigm do not hold in reality. Maybe -for the majority of the job seekers, and the poor in particular -job search behaviour is driven more by other motives (the quest for autonomy, participation and social inclusion, the sense of citizenship) than by financial incentives. Moreover, although inactivity and poverty traps may be 'biting' for some categories (early retired, large families, single parents, marginal workers) they do not affect the vast majority of welfare recipients. And most importantly, the majority of all unemployed people in Europe simply do not draw any unemployment benefit. In Spain in 2016, for example, barely one-third of all job seekers were covered by unemployment (insurance or assistance) benefits (OECD, 2018: 194). Across the OECD as a whole, the picture is roughly the same, though with a lot of variation between countries.
Overall, we must conclude from the empirical literature that the role of financial incentives has been grossly over-estimated. The 'making work pay' paradigm appears to rest on over-simplified assumptions, while other important determinants of employment are being overlooked.

The conservative view: workfare, rights and duties
The rights and duties approach is based on the perception that financial safety nets do not just discourage job search, but even contribute to a 'dependency culture', i.e. a value system that is supposedly inherited from generation to generation in poor families. The fact that some benefits are paid selectively to groups with specific charac teristics allegedly incites people to adapt their behaviour in order to obtain (or retain) those benefits. In the American literature, for example, we encounter extensive debates on the former AFDC 2 welfare scheme, centred around the claim that AFDC encouraged young women to remain unemployed, to separate from their partners and even to bear children with a view to being able to claim the AFDC benefit for as long as possible (Murray, 1984(Murray, , 1990. The resulting dependency culture was said to breed in ghettos, where it is allegedly passed on from generation to generation. This claim is then linked to the observation that many daughters of AFDC-clients in turn began claiming benefit as pregnant teenagers. A culture such as this, it is argued, should be combated by changing the behaviour of the poor: for example, by linking income support to mandatory unpaid community service ('workfare') or enforcing alimony payment by fathers (Mead, 1986). This view informed the 1996 Welfare Reform in the US, which abolished entitlements and limited the duration of GMI-benefit recipiency (TANF) to a maximum 5 years in a life-time, under the telling title 'Personal Responsibility and Work Orientation Reconciliation Act'.
The debate about the dependency culture has been less intense in Europe. It was relatively lively in the UK a few decades ago (Runciman, 1990) and was re-fuelled in particular by the writings of Dalrymple (2001), who saw a strong link between poverty and moral decay (dependency and laziness, teenage pregnancy, out-of-wedlock childbearing, family disruption, child abuse, addiction and criminal behaviour) and blamed social services for perpetuating the underclass culture instead of combating it. It is probably no coincidence that the Cameron Government was the first to introduce large-scale workfare programmes in the UK during the 2008-2013 crisis.
In continental Europe, the shift was less abrupt. Nevertheless, the main trend in the unemployment insurance has been towards tightening of the access to -and conditionality of benefits, in four ways: -restrictions in eligibility conditions (age, family income, prior work intensity in the period preceding application, assets…), -obligation to accept less attractive job offers (longer commuting distances, lower-paid jobs, offers below one's level of qualification, more flexible working hours), -tighter follow-up of job search activity (obligation to report weekly or monthly on one's job applications, training efforts etc.) and -'contractualisation' of social protection (e.g. through compulsory participation in activation programmes, including community work, in exchange for benefits).
The number of sanctions increased across the board in Europe too, with 'first sanctions' generally being moderate and repeated sanctions more harsh (Knotz, 2014). As European economies got stuck in a prolonged economic and employment crisis, the tendency for governments to introduce workfare schemes gained ground. The examples of the UK and Germany informed tight activation programmes, particularly 2 AFDC = Aid to Families with Dependent Children. In 1996 this scheme was replaced with TANF = Temporary Aid to Needy Families in Norway, Bulgaria, Latvia, and The Netherlands. In a number of EU countries, the Youth Guarantee programmes have turned into workfare programmes due to poor funding and substandard remuneration of participants. 'Pure' workfare programmes have also been implemented in Central-Eastern European countries with large Roma minorities (Slovakia, Hungary, Bulgaria) as a way of combating their perceived dependency on unemployment benefits or minimum income schemes (Lakner & Tausz, 2016;Vidra, 2018;Sirovatka, 2016;Csomor, 2017).
Two noteworthy examples of workfare-like programmes are Community Work Placement in the UK and the Hartz reforms in Germany.
In April 2014, the British Government introduced strict eligibility conditions for benefits, coupled with sanctions, on the long-term unemployed under the 'Help to Work' programme. In concrete terms, applicants are subject to an assessment and subsequently allocated to one of three pathways: mandatory training, community work placement, or daily monitoring. Community Work Placement involves 30 hours of unpaid community work, as well as 4 hours of job counselling, during a period of six months. Non-compliance results in a 4-weeks suspension at the first infringement and 13 weeks the second time. The community work must provide useful work experience and include individualised support. The pilot preceding the law ('Community Action Programme' -CAP) involved 15,000 participants and was rigorously evaluated, with the third ('monitoring') subgroup serving as control group. 3 In the 91 months following its kick-off, the (un)employment trajectories of the three subgroups barely deviated from each other, finishing with a 2% 'advantage' of the CAP group. Yet, this difference in benefit retention was not even attributable to differences in employment rate. Moreover, given that the jobs carried out by the CAP participants competed directly with regular jobs, we can assume high displacement effects, which means that the net employment effect of the CAP was very negative.
In Germany, the Hartz reforms that were introduced between 2003 and 2005, combine a drastic cut in benefits for the long-term unemployed with an integration contract and workfare-like (low-wage) employment. The unemployed henceforth fall back on very low and means-tested 'unemployment assistance' benefits after 12 months, substantially below the poverty line. On the other hand, they are 'allowed' (but also obliged) to take up part-time and low-paid 'Mini-jobs' yielding up to 450€/month and get a tapered reduction of their benefit. The Mini-jobs are not subject to social security contributions but do not build up rights either. The scheme was later extended with Midi (up to 1300€) jobs which are also partly tapered and exempted from social security contributions. Approx. 7.5 million Germans worked in Mini-and 1.2 million in Midi-jobs in 2019. In addition to the low benefits and low quality of the jobs offered, the 'workfare' nature of the reform is also characterised by its emphasis on 'duties' ('Fördern und Fordern'): any (legal) job must be accepted, and non-compliance can be sanctioned with benefit cuts.
Altogether, the reforms appear to have generated a 'social dumping' effect on the German labour market, resulting in increased employment at the expense of more poverty. The price for the employment creation was also borne to a large extent by the remaining long-term unemployed, whose poverty risk has risen to 67%, i.e. the highest level in the EU (Gutierrez, 2016), and partly also by trade partners whose terms of trade relative to Germany deteriorated (Akyol et al., 2013). 3 The first subgroup received a treatment called 'ongoing (intensive) case management' (OCM); the third subgroup (control group: Job Centre Plus -JCP) just had to report regularly to the Job Centre.

The social investment approach
The European Commission's Social Investment Package, launched in 2013, can be seen as more or less representative of a third welfare state paradigm. It includes active inclusion as one of the pillars of social investment along with early childhood services, education, health care, the Youth Guarantee plan, the EU Skills Agenda, social housing etc. Social investment is defined as 'investing in people' (as opposed to financial or material investments): the common feature to all forms of investment is that costs are incurred in the short and medium term, with a view to yielding returns in the longer term. For example, investing in skills training may shorten the expected duration of unemployment and raise workers' productivity, and may therefore be profitable for the individual as well as for the government, tax-payers and society at large. Apart from skills training, investing in active labour market policies may also include health care, psychological assistance, child care, mobility services etc.
The EC's Communication on the Social Investment Package (EC, 2013a) refers in the first place to social investment as an economic strategy to make people more employable and to 'ensure the best use of economic resources', and many passages continue to transpire a neoliberal ideology (focusing on the individual worker, calling for private investment etc.). Yet, the introduction also frames the initiative as a means to foster social cohesion and implement the Charter of Fundamental Rights of its citizens. Hence, it seems that different political schools of thought may develop their interpretations of social investment.
In the Re-InVEST research, 4 which aimed to strengthen the philosophical and scientific underpinnings of the social investment paradigm, employability was defined as one dimension of human capabilities; and social investment as a strategy to extend human capabilities with the aim of implementing fundamental rights (Sen, 2004;2005). This view of active labour market policies sharply contrasts with the 'making work pay' and workfare approaches in several ways: -A humanistic vision of vulnerable people as citizens with equal rights and aspirations, as opposed to a vision based on distrust or underclass culture. -A multidimensional framework for government action, as opposed to a narrow view based on financial sticks and carrots. -A focus on the long-term (including intergenerational) sustainability of work, as opposed to a short-term 'work first' perspective: the investment in human resources may take time but should result in the capability to carry on a job, and for working parents to invest in their children. -Finally, an emphasis on choice as a key element of welfare. One of the novelties of Sen's theory of capabilities is the explicit link between freedom and welfare. Persons with the same level of functioning(s), but with unequal freedom, have a different degree of welfare, because the 'capability' of the less free persons (i.e. the set of functionings that are feasible for them) is more restricted than the capability of the more free persons. This means that the room for negotiation in activation pathways becomes a key criterion in evaluating the welfare-enhancing 4 The acronym Re-InVEST stands for 'Rebuilding an inclusive, value-based Europe of solidarity and trust through social investment'. See www.re-invest.eu for a full description of the project and its outputs. effect of measures. From this perspective, workfare can be seen as reducing the participants' well-being, despite a positive net employment effect.
Note that with this lens on labour market policies, the notion of social investment coincides only partly with active labour market policies (ALMPs). On the one hand, social protection itself (which is often erroneously labelled as 'passive labour market policy') partly serves a social investment purpose: decent social benefits should allow job seekers to invest in their own human and social capital so as to maximise their chances of finding a decent job. On the other hand, cheap and coercive ALMPs may not entail any form of investment in people -or indeed disinvest by destroying the job seekers' human and social capital.

Measuring effectiveness: econometric analysis
One of the work packages of the Re-InVEST research aimed to re-assess European labour market policies from a social investment perspective (Lehweß-Litzmann & Nicaise, 2018a-b; Lehweß-Litzmann & Nicaise, 2020). The first step consisted of a dynamic econometric model to estimate the impact of the generosity of social benefits as well as the intensity of ALMPs on sustainable labour market integration. We used the EU-SILC (rotating) panel data from 2006 to 2014 for the EU28, combined with aggregate data on ALMP expenditure and other macro-economic control variables by country and year, in a multilevel analysis. The model was applied to the active population as a whole as well as -separately-on households living in poverty. Given our focus on 'social investment', we ignored immediate anti-poverty effects and concentrated on the medium-term impact of social protection and ALMPs on the degree of economic self-reliance and employment of households. Although it was not possible to capture effects on a full set of capabilities with the available data, we managed to measure more than just short-term employment effects. More precisely, the following criteria were used to assess the effectiveness of policies: -For households experiencing poverty: -The effect of social transfers in year t on the reduction of the pre-transfer household poverty gap 5 (among households at risk of poverty in year t) between year t and t+1; -The effect of social security transfers in year t on the improvement in work intensity 6 between year t and t+1, among households with very low work intensity in year t. -For the population in active age as a whole: 5 The poverty gap, measured at household level, is the percentage difference between a household's actual level of income and the poverty threshold for that household type. Note that this analysis examines the 'pre-transfer' poverty gap, i.e. abstracting from social security transfers, as a measure of 'lack of financial self-reliance' of the household. 6 Work intensity is the share of the overall amount of time, available to all adults in the household, spent in paid employment. Very low work intensity is defined as work intensity below 20% -The reduction in financial poverty risks 7 between two / three consecutive years through unemployment or early retirement benefits or ALMP; -The shift in work intensity among households with very low work intensity, through unemployment or early retirement benefits or ALMP.
Our empirical findings show remarkably small effects of both types of policies. Drastic increases (decreases) in the level of social benefits generate extremely small decreases (increases) in employment and 'self-reliance' of poor households. The good news is that the fear for adverse effects of generous social protection appears ungrounded: more decent social benefits will not trap people into long-term unemployment. They generate immediate poverty-alleviating effects without increasing, nor perpetuating dependency. This also disqualifies the 'making work pay' assumption that more generous benefits prevent labour market reintegration. The bad news is that our findings do not support the social investment view of social transfers either, as higher benefit levels do not automatically produce sustainable exits from poverty. Other tools may be needed in addition for that purpose, such as education, health care, etc.
As regards active labour market policies, the lessons are also ambiguous: at country level, a 50% increase in ALMP spending appears to result in less than 0.3 percentage points increase in the aggregate employment rate in the following year, albeit with a positive redistribution effect towards vulnerable groups. At the same time, however, the increased ALMP spending also has a (small) negative effect on the economic activity rate. This suggests that ALMPs also drive job seekers into inactivity (e.g. through sanctions or suspensions, or because too demanding activation offers take a toll on people's health). In other words, in the period 2006-2014, ALMPs simultaneously seemed to invest and disinvest: some groups moved into work more swiftly, while other groups were excluded from the labour market. The most positive employment effect of ALMPs is a redistribution of employment opportunities towards older, female and less educated groups.
All in all, the findings from our econometric analysis raise doubts about the effectiveness of ALMPs in the observed period: the fact that their effect is so ambiguous may be due to poor targeting and/or poor quality of the services offered. Therefore, the Re-InVEST team also carried out a qualitative analysis, based on seven case studies and a more comprehensive concept of quality. More concretely, the case studies related to disabled persons in Latvia, users of mental healthcare in Liverpool (England), older job seekers in Salzburg (Austria), immigrants in Antwerp (Flanders) and Seine-Saint-Denis (France), and NEETs in Espinho (Portugal) and Geneva. Like often in qualitative research, the cases examined here do not pretend to be representative for the average type of provision across Europe. The aim was rather to reflect the diversity of systems and target groups in the best possible way, and to understand the interactions and processes that produced the outcomes. Therefore, it would be unwise to try and generalise the research conclusions. Yet, we found some commonalities that allow us to draw relevant lessons for labour market policies in Europe, particularly from the perspective of vulnerable groups.
Building on the definition of social investment elaborated in section 4, three key questions were examined: (a) to what extent do labour market policies respond to the 7 A household is at risk of poverty (AROP) when its equivalised disposable household income is below 60% of the national median human rights of citizens (in particular, the right to a decent living standard and the right to work) ? (b) To what extent do they enhance the beneficiaries' capabilities ? (c) To what extent are freedom and voice (two essential dimensions of capabilities) realised in these policies ?
In all case studies, a co-construction method was used, with mixed groups of academic researchers, 'peer researchers' from the target group, local NGO workers and sometimes also professionals from public services 'merging their knowledge' and evaluating the case together, through a series of focus group meetings spread over approx. six months.

Labour market policies and human rights
In the realm of social protection and ALMP, human rights refer mainly to the right to social security, work, free choice of work, decent working conditions and protection against unemployment (art. 22-23 of the Universal Declaration of Human Rights). The European Social Charter elaborates further on several aspects such as the right to vocational guidance and vocational training; and the European Pillar of Social Rights (EPSR) spends two full chapters to a detailed list of rights related to work. However, human rights are unseparable, and very soon it appears that other rights such as health, protection of family life, and even the right to good governance are at stake.
Our co-researchers experienced a big gap between principles and reality in this regard -beginning with information and awareness about rights. The most striking example was the Portuguese Youth Guarantee Plan, which seeks to guarantee an individualised pathway to work within 4 months of unemployment. A senior officer from the public employment service (PES) declared in an interview that he did not see the interest of informing young people about this 'guarantee' given the lack of resources to implement it (Costa & Araújo, 2017: 20). Young people wanting to register for unemployment assistance did not even get correct information about the eligibility rules and registration procedure. Similar complaints are found in the Latvian report about employment subsidies for disabled persons: neither the target group, nor employers knew about the existence of the scheme (Lace & Rungule, 2017: 24). In Austria, job referrals under a temporary wage subsidy scheme were initially not disclosed to the job seekers themselves: several co-researchers discovered being allocated to such jobs only when they were fired due to expiration of the subsidy. They felt that they had been used as 'cheap disposable workers' to enhance their bosses' profits (Buchner & Lessmann, 2017: 16). In the case of the Employment Support Allowance (ESA) for people with long-term sickness in England, withholding information even appeared to be part of the strategy of social services to reduce their caseload, resulting in unequal treatment of clients: Of course, information about rights is no more than a starting point; accessibility is another stumbling block, particularly for the most disadvantaged. The lack of personnel and resources in public services was exacerbated during the 2008-2013 crisis by the fact that many countries even economised on expenses for unemployment while the caseload was booming. Whereas digitalisation would supposedly provide an efficient response to staff reduction, it also meant more impersonal contact, and even unbridgeable distance for low-literate clients and those who cannot afford access to the internet. In the French case study, Abder (24, released from prison and living in a rehabilitation centre) felt that officers from the PES and the 'Mission Locale' distrusted him and did not believe he can be integrated. He sketches his relationship with his 'personal advisor' at the PES rather cynically: "I cannot reach my councillor at the Employment Agency. I asked her for appointments, but she didn't schedule any (...) I have never seen her. I was told 'Eh you have Mrs so and so as a councillor'. I said 'ok! It is cool, give me an appointment' but she never scheduled any with me." (Abder, in: Droy, Dubois & Ricardou, 2017: 30).
For the most vulnerable, free choice is also less of an option. Case managers decide for them what training they can get, and what they can not. An immigrant in Flanders testified that his social worker denied him the opportunity to take a Dutch language course because he thought it was too difficult (Debruyne & Put, 2017: 23).
A more subtle way of conditioning support is contractualisation. Although theoretically, integration contracts linked with social benefits provide some room for negotiation and individualised, tailored services, in practice the asymmetry in information, status and power is so large that contracts turn into prescriptions that are imposed at the discretion of case workers. The immigrants participating in the research in Flanders complained that they did not understand the integration contracts they had to sign, due to language barriers; however social workers declined questions and insisted on signing the contracts, arguing that they had no time "and that questions could be asked later by email" (Debruyne & Put, 2017:20-21).
Although discrimination often occurs in subtle ways and is therefore hard to prove, immigrants believe that language barriers are used unnecessarily to deny them access to particular job offers or training schemes. Language barriers are also used by co-workers on the work floor to discriminate against immigrants.
Sanctions are never far away under such a regime. Among people with mental health problems in the UK, the fear for withdrawal of benefits, combined with workshops organised in Job Centres to 're-motivate' welfare recipients in the context of the ESA are even experienced as a form of 'psycho-compulsion' rather than a recognition of rights (Moth & Lavalette, 2017: 23).
Summing up, restoring decent rights in social protection and labour market services requires investment in awareness raising, accessibility for all citizens, free choice and respectful treatment of claimants.

Investing in capabilities
According to our theoretical framework, the capabilities of the 'beneficiaries' of national policies are affected in three ways: via their resources (supplemented through social benefits), the conversion factors (nature and conditions of the services offered) and the room for negotiation and autonomy.
Resources are determined by the generosity of social protection schemes and the earnings linked to job opportunities. In Portugal, not only social benefits, but also public employment services themselves were affected by austerity measures during the 2008-2013 crisis. On both sides, the resources are too low to enable investment: "When I left school, I was unemployed, I applied for a benefit and they offered me €24 and I rejected it. What is €24 going to do for me? (...) I also asked her what she would do with the €24 if she were in my shoes. She said nothing, just that those are the rules. I told her I did not want it. I couldn't do anything with €24 and so I didn't accept it." (school leaver, Espinho -in: Costa & Araújo, 2017: 24).
At the same time, the PES had to cut back on staff while the caseload continued to increase: "(...) I would say, with more than 30 clients in a day, the quality is getting worse, which is normal, it is only human as there are limited staff available to meet demand (...) Personal assistance is only available in crisis scenarios as we have experienced, and given the human resources available, it is no mean feat." (PES officer, in: Costa & Araújo, 2017: 22).
As mentioned earlier, the ESA and PIP 8 reforms in the UK were based on the premise that the previous schemes were too generous, particularly towards people with mental health problems: hence, regular re-assessments carried out by private subcontractors aimed to exclude a number of beneficiaries. The co-researchers emphasised that social protection is often indispensable for sheer survival: they reported about the difference benefits make on people's material living conditions, as well as the impact of suspensions on food poverty, destitution and even suicide as a consequence of reduced or suspended benefits:  (Matt, claimant -ibid.) Looking at the services offered to clients, the key question is to what extent they respond to the complex needs of vulnerable groups. On the negative side, 8 PIP = Personal Independence Payment, a living allowance covering the extra costs of people with health problems.
inadequacy of offers was a recurrent theme in several case studies. The Austrian co-researchers complained about the obligation to repeat the same job application workshops at regular intervals. In Portugal, young people saw registration at the PES and the job training offered there as useless: course contents are limited in scope, the courses are poorly organised, and the training does not improve their chances in the labour market. This view seems to be shared by employers. Even vacancies continue to be advertised after they have been filled (Costa & Araújo, 2017: 20-21). The immigrants in Flanders accepted the obligatory introduction and language courses, but criticised the ineffectiveness of these schoolish entry-level courses. Several participants would have preferred to take more intensive courses or to learn the language more informally on the work floor, but their requests were turned down by social services due to the costs or risks involved. The Flemish integration policy has a reputation of rigidity, moulded too narrowly to the profile of low-educated immigrants, with a sequential approach based on 'language first', followed by housing, activation into (entry-level) work, etc. Moreover many case workers in social services appear to start from the premise that immigrants can only do seasonal work in the agricultural sector, low-skilled work in the construction industry or cleaning -to the point that one might suspect them of ethnic prejudices. Blended pathways tailored to more diverse capabilities and aspirations would probably be much more effective. The Flemish PES and municipal social services are currently experimenting with such alternative pathways for high-educated immigrants (Debruyne & Put, 2017: 22).
The mental health patients in Liverpool were extremely critical about the detrimental effects of the 'services' they have to undergo on their psychological wellbeing. They described 'motivation courses' (where welfare claimants are taught through positive psychology techniques how to adopt more positive behaviours) as 'blame workshops' because the sessions explain unemployment exclusively in terms of personal deficits and thus inculcate a sense of guilt to participants. Combined with intrusive, painful interviews about their mental health problems, the workshops re-traumatise participants and further undermine their self-image. Moth & Lavalette (2017: 23) label this approach 'psychocompulsion'.
On the other hand, there were positive stories of measures that changed people's lives. The biggest success story from our case studies was Scène Active, a project of 'pre-vocational training' for socially excluded youngsters that was completely designed from a capabilities perspective. The project has two 'parallel and non-competing goals': the collective project of realising a theatre performance, and 40 individual projects of inclusion. The experiment (repeated yearly) lasts nine months and focuses on the design of the theatre performance, from A to Z, in a real setting. Participants write the theatre plot themselves and develop their own characters; they also have to design the scenography, prepare the costumes, the music and the videos that appear during the play. Within the project, the participants can also choose between six workshops with professional facilitators: theatre, music, video, scenography, costumes and cooking. Despite the apparent 'distance' between the content of the project and its objective of labour market integration, Scène Active appears to produce quite positive results. Through the project, marginalised young people find suitable jobs, decent dwellings, and begin to rebuild 'normal' lives. Most importantly, their self-esteem and social skills increase, their relations with their family and friends improve, and they develop personal projects for the future.
"Before Scène Active I was closed in myself and I spoke to nobody; I stayed alone hidden in a corner. Thanks to Scène Active I opened myself towards other people, I've fought my shyness. Because in all jobs you cannot be shy: you have to discuss with the people, exchange with them. For me it was this that changed the most before and after Scène Active: the shyness and the motivation." "I even dared going to the 'direct recruitment', where you go to present yourself directly in a company with your application. I've never done anything like that before, I was too shy. But now I've done it already two times." "My mother is so happy to see me waking up each morning." and a participant of the previous year testifies: "You cannot see the results immediately. But then you start noticing through some experiences, such as an internship, that you have developed a lot and that you are much more self-confident about your competences. I succeeded in the recruitment interviews and I even received congratulations because -they told me -it is rare to see such a young person who knows herself so well and who puts so much passion in the things she does." (youngsters -in: Bonvin & Laruffa, 2017: 19) One striking commonality across our case study is the participatory and holistic approach used by the NGOs involved in labour market integration services. In comparison with the regulated public services, NGOs are often specialised in services for particular disadvantaged groups, more embedded in local communities, and able to offer a wider set of tailored services. This is the case, for example, with Apeirons, an association specialised in integration through work for persons with disabilities in Latvia. They reach out to their clients through home visits, cooperate very closely with a multidisciplinary team of psychologists, social workers and career advisors, and have accumulated experience in tailored employment (telework, part-time employment, niches in the labour market for deaf and blind persons, etc.). In Seine-Saint-Denis (France), the NGO GRDR has a strong presence in immigrant communities in the suburbs of Paris. GRDR's approach consists in linking pathways with the clients' personal projects for the future: this involves intensive dialogue with the young people, coaching in drafting cv's, training, mediation, financial support 9 , administrative and psychological support, voluntary work offers in the local community or even abroad, theatre workshops, networking with other organisations, community action, sometimes indeed a return to school.
The link between labour market related services and social work is emphasised in several case studies: sometimes as a gap (where collaboration between employment and social services is non-existent, like in Espinho -Portugal), and sometimes as a strength (like in GRDR or Apeirons). In the Swiss case (Scène Active, Geneva), three (part-time) social workers were engaged for a group of 40 disadvantaged young people. Their work was highly appreciated by the participants.

Choice and voice
Section 5.2 on rights already dealt with the issues of contractualisation and compulsion in ALMP, which constrain the right to free choice of work. In this section, we 9 E.g. support to pay the fee for a driving licence course elaborate on the degree of choice and the processes of negotiation between services and users in our local case studies.
The Austrian co-researchers in particular challenged the use of suspensions as a sanction when a client refuses an offer that s/he deems inadequate: e.g. repeating identical training courses, or cases where people are referred to social enterprises not because of their personal problems but as a 'test of their willingness to work'. Lack of information about the available instruments for re-integration also limits the negotiation power of job seekers. The team therefore recommends to systematically inform all new applicants about the full set of available measures they are entitled to. They also claim that a wider choice of educational offers should be made available, while on the other hand employment in temporary work agencies should never be compulsory. Last but not least, the group suggests to foster participation of job seekers in the design and evaluation of policies (Buchner & Lessmann, 2017: 20-29).
Similar aspirations are voiced by the group of immigrants in Flanders. In broad terms, they are grateful for the support they receive, but remain critical about the limited scope of (entry-level) training and jobs offered to them, resulting in systematic under-employment and mismatch. They would also prefer to get better access to further education, for example (Debruyne & Put, 2017:24).
In this regard, much can be learned again from the Swiss case study of 'Scène Active'. The degree of autonomy, choice and participation is maximised in all dimensions of the project: participation is completely voluntary, and the youngsters can in principle opt out at any moment if they want. Participants negotiate the rules of the project with the director. Collective decisions are taken by voting. Sanctions are never imposed (e.g. when participants arrived late for a rehearsal). This far-reaching freedom and autonomy obviously entails a high degree of risk: for example, absenteeism is a serious problem during the project. However, this risk is counterbalanced by a sense of collective ownership and responsibility for the success of the performance, which ultimately results in stronger personality development and better work attitudes (Bonvin & Laruffa, 2017: 16-17).

Conclusion
This article aimed to conceptualise the notion of 'social investment' within the context of social protection and active labour market policies. The social investment paradigm, advocated by the European Commission, is far from dominant in these policy areas in Europe. In the past two decades, it was the neoliberal 'making work pay' paradigm that dominated the public discourse. This framing entails a fundamental dilemma between both objectives: it tends to include those who work at the expense of greater poverty of those who are unable to work. Another popular paradigm -the conservative workfare doctrine -builds on the idea that a stronger work ethos needs to be inculcated through reduced rights and more duties. This approach is not conducive to generous anti-poverty policies either.
The social investment paradigm thus 'rows upstream'. It advocates generous social protection combined with enabling investments and is thus perfectly congruent with social inclusion objectives. But it meets resistance from the other two paradigms, which means that empirical research needs to demonstrate what effects can be expected from shifts in the level of social protection and from ALMP.
Our own empirical analysis -which is in line with a series of other studies, despite differences in methodology -shows remarkably small effects of both types of policies in the past 15 years. On the one hand, it (further) disqualifies the making work pay paradigm: more decent social benefits will not trap people into long-term unemployment. They could generate immediate poverty-alleviating effects without increasing, nor perpetuating dependency. As regards active labour market policies on the other hand, the findings are not convincing either: so far, ALMPs have had small positive effects on employment, but also on exits into inactivity. The latter effect may be the consequence of sanctions, or harmful effects of activation programmes on job seekers with a fragile health. The most positive employment effect of ALMPs is a redistribution of employment opportunities towards older, female and less educated groups.
The findings of our quantitative research raise doubts about the quality of ALMPs in the recent past. We therefore carried out in-depth participatory case studies of policies targeting disadvantaged groups in different European countries, using an 'enriched' social investment model, building on a human rights and capabilities approach. The findings indeed show very wide differences in quality -and a lot of room for improvement. Some ALMPs constrain individual freedom and impose measures that appear to be inadequate, if not counter-productive. In order to improve their effectiveness as well as inclusiveness, the right to decent ALMPs should be put on the policy agenda.
The better examples from our qualitative research teach us some important criteria for such decent-quality policies in the future: (a) more diversified offers (including more education and training, voluntary work, and social support) in order to provide tailored responses to individual needs; (b) more holistic approaches, building on intersectoral partnerships between labour market, education and training, health care and social services; (c) more flexible, open-ended pathways that can be adjusted to changing circumstances; (d) involvement of civil society organisations that are embedded in local communities and have a capacity to innovate; (e) more voice for clients in the design and evaluation of services; (f) more respect for the clients' rights, including room for negotiation and freedom of choice in pathways into work; and last but not least (g) no ALMPs without decent social protection.