Horizontal and vertical coordination of the European Youth Guarantee

The European Union launched the Youth Guarantee (YG) in 2013 to combat the increase in youth unemployment following the financial and economic crisis. The goal of the YG was to ensure that all young people under the age of 25 would receive a high-quality, concrete offer of employment or training within four months of leaving formal education or becoming unemployed. The measures at national level were to combine various activities: early intervention and activation, supportive measures enabling labour market integration, assessment and continuous improvement of the scheme, and its swift implementation. An additional emphasis was to be placed on building up partnership-based approaches and effective coordination across policy fields such as employment, education, youth and social affairs (Council of the European Union, 2013). To advance these goals, the EU for the first time dedicated a specific budget to youth employment policy, creating the Youth Employment Initiative (YEI), which supplements the financial aid provided under the European Social Fund (ESF). The YEI budget was directed primarily at young people not in employment, education or training (NEETs) who were living in regions where youth unemployment rates were higher than 25 per cent in 2012 (European Commission, 2017c). In addition, the incorporation of the YG into the Country Specific Recommendations of the European Semester indicated that the scheme would be monitored more closely compared with the implementation of other EU social policies. Altogether, decision makers combined high expectations with particular support for the YG at EU level. The goals confirm that the YG was rooted in the normative paradigms of an activating labour market policy (Gilbert and Van Voorhis, 2001; OECD, 1989) and a social investment welfare state (Giddens, 1998; Morel


INTRODUCTION
The European Union launched the Youth Guarantee (YG) in 2013 to combat the increase in youth unemployment following the financial and economic crisis. The goal of the YG was to ensure that all young people under the age of 25 would receive a high-quality, concrete offer of employment or training within four months of leaving formal education or becoming unemployed. The measures at national level were to combine various activities: early intervention and activation, supportive measures enabling labour market integration, assessment and continuous improvement of the scheme, and its swift implementation. An additional emphasis was to be placed on building up partnership-based approaches and effective coordination across policy fields such as employment, education, youth and social affairs (Council of the European Union, 2013).
To advance these goals, the EU for the first time dedicated a specific budget to youth employment policy, creating the Youth Employment Initiative (YEI), which supplements the financial aid provided under the European Social Fund (ESF). The YEI budget was directed primarily at young people not in employment, education or training (NEETs) who were living in regions where youth unemployment rates were higher than 25 per cent in 2012 (European Commission, 2017c). In addition, the incorporation of the YG into the Country Specific Recommendations of the European Semester indicated that the scheme would be monitored more closely compared with the implementation of other EU social policies. Altogether, decision makers combined high expectations with particular support for the YG at EU level.
The goals confirm that the YG was rooted in the normative paradigms of an activating labour market policy (Gilbert and Van Voorhis, 2001;OECD, 1989) and a social investment welfare state (Giddens, 1998 , 2012). Comparative research has demonstrated different approaches within these paradigms, namely a pro-market or work-first approach in contrast to a human-capital development/enabling approach (Bonoli, 2010;Dingeldey, 2009Dingeldey, , 2011a. A work-first approach involves the prioritization of immediate labour market integration of young unemployed people, stressing, for example, subventions to employers. By contrast, an 'enabling' approach treats improved qualifications or upskilling as the dominant goal of youth employment policies. The EU goals gave no priority to one particular approach. The YG recommended the reduction of labour costs and subsidies to employers, but also suggested improving the quality of employment services and strengthening education and vocational training (Council of the European Union, 2013). The YG overall acknowledged the diversity of member states regarding youth unemployment and institutional arrangements, financial constraints and the capacity of the various labour market players (Council of the European Union, 2013). In addition, the Commission stressed that in most member states the implementation of the YG would require longterm, in-depth structural reforms of training, job-search and education systems to improve school-to-work transitions (European Commission, 2015b). Commission officials characterized the YG as being a policy approach rather than a programme with fixed money and milestones. Its value was to 'oblige everybody to think globally about youth employment programmes' (Interview EC).
Despite its rather ambitious goals, the YG was subject to the Open Method of Coordination. This method has been in use since 2001 in sensitive areas such as European social and employment policies where member states have not been willing to grant the EU political powers. Hence, the 'Recommendation' is non-binding, rather it encourages overall intergovernmental coordination, benchmarking and best practice without threats of sanctions (Heidenreich, 2009;Heidenreich and Zeitlin, 2009).
To investigate the implementation of these ambitious goals by means of rather limited, albeit strengthened forms of social policy governance, we focus on the following questions: • • How did instruments of vertical and horizontal coordination linked to the YG work in a multi-level governance system?
• • To what extent do the goals and ideas related to the YG translate into changes of national policies and institutions relevant to combatting youth unemployment?
To answer these questions and to be able to mirror the diversity across member states, we chose contrasting cases for an in-depth comparison: Germany as a country with low youth unemployment rates, and two Southern European countries, Greece and Spain, as cases with high unemployment rates and YEI eligibility. These countries also differ as regards the institutional setting for supporting young people in their school-towork transitions via the vocational education and training systems and the public employment service (PES). Our presentation in this chapter proceeds as follows: Drawing on existing research, we outline our theoretical framework and research approach. Next, we describe the institutional preconditions and the national approaches to youth employment policy prior to the YG in the three countries studied. We present findings related to the coordination and implementation of the YG as well as to the procedural change aligned to European policies (Section 4). Finally, we analyse the policy development in search of substantive change in the three member states (Section 5). We end with a brief recommendation for future policies (Section 6).

THEORETICAL FRAMEWORK AND RESEARCH APPROACH
Research on the European YG is still rather poorly developed. Yet, even this limited research involves a controversy regarding the relevance and impact of the YG. Critics have highlighted issues like underfinancing with respect to the number of young unemployed in the different countries (Cabasés Piqué et al., 2015), as well as the limitations of soft modes of governance seeking to influence member states mainly through voluntarism (De la Porte and Heins, 2015). Such criticism has culminated in claims that YG implementation is dominated by a path-dependency logic and does not lead to convergence of the specific elements addressed by the Council Recommendation (Dhéret and Roden, 2016;Madsen et al., 2013). Other scholars have argued that the link between European funding instruments and the YG points to stricter forms of vertical coordination, generally in combination with the European Semester. It follows that the YG is likely to foster a greater degree of Europeanization, at least in countries eligible for YEI funding (Costamagna, 2013;Dhéret and Roden, 2016).
To provide new knowledge about how the YG has worked and possibly changed national policies and institutions, we drew on historical institutionalism theory (Pierson, 2000;Thelen, 1999) in combination with a multi-level governance approach (Marks and Hooghe, 2004) and the Europeanization literature (Börzel and Risse, 2006). Historical institutionalism points to the importance of institutions in shaping policy  (Pierson, 2000;Thelen, 1999). Accordingly, external shocks like crisis, war or critical junctures may lead to pathbreaking changes (Pierson, 2000). Others, however, have pointed out that policy learning may be a driver of change. More recent literature also points to gradual institutional shifts that may lead to path-breaking changes in the long run (Streeck and Thelen, 2005). To some extent, a multi-level governance approach and the Europeanization literature have integrated these explanations. A multi-level governance approach outlines the institutional background of regulations and competences between different levels. Furthermore, the particular mode of governance is relevant. For instance, one would consider the open method of coordination for the YG as a soft form of governance. Zeitlin et al. (2014) have identified five mechanisms for influencing member states' social policies. These mechanisms include: (1) external pressure (to meet commitments); (2) external support (financial or technical); (3) socialization and discursive diffusion (internalization of common cognitive frames via reviewing); (4) mutual learning (awareness of policies, practices and performance in other member states); and (5) creative appropriation (strategic use by different actors).
The Europeanization literature focuses on European integration as a driver for domestic change. In order to explain different reform trajectories, however, scholars point to domestic institutional settings as relevant influence factors. Such settings may include the political system of a country, the particular forms of vertical governance structures within federal states (Pierson, 1995) and the established institutions of the particular welfare regimes (Weishaupt, 2014: 227). Moreover, this literature also sees weak economic and administrative capacities as influential factors (Weishaupt, 2014) and mostly as obstacles to successful and unitary implementation of EU policies. However, another research strand contests this view, arguing that adaption pressure increases with the misfit between European policy goals and national preconditions (Cowles et al., 2001;Falkner et al., 2005). Beyond these different understandings of the misfit between European goals and national conditions, the Europeanization approach tends to neglect the fact that EU policies are not the only influence on national policies. For instance, we have to consider whether European-level policies really have initiated the adoption of the YG in a member state. We need to carry out an empirical investigation to clarify which of these competing assumptions best captures the situation in different member states.
To characterize the kinds of changes in the member states, we need precise concepts of change. Building on the work of Weishaupt (2014), we make a distinction between substantive and procedural changes. Substantive changes are defined as ideational (shifts in positions of actors), agenda-setting (weight that actors place on particular issues) or programmatic (shifts in legislative and administrative rules and practices). Procedural changes are related to changing governance and policymaking arrangements. They include horizontal coordination (between different policy fields/administrations) and involvement of non-state actors, but also enhanced national steering capacity (monitoring evaluation) and improved vertical coordination (i.e., between governance levels). As we focus our analysis on the implementation of the YG, we give less attention to decision-making processes. To explore substantive changes, we limit our analysis of the YG to agenda-setting and programmatic changes. With respect to procedural changes, we focus on two aspects, namely reinforced coordination between different policy fields or administrations and the increased involvement of non-state actors, particularly of the social partners. In contrast to Weishaupt (2014), but in accordance with YG guidelines (European Commission, 2017c), we see these two aspects as being related to horizontal coordination (see Table 9.1). Drawing on the theoretical outline, the YG combines mechanisms of external support with external pressure. We therefore expected a general willingness of member states to comply with the respective YG goals and started the empirical investigation with the following two hypotheses: • • First, in line with the historical institutionalist approach, we expected a path-dependent implementation of the YG according to the previously established youth employment policies of member states. To assess the empirical support for the two hypotheses, we adopted a three-pronged research approach: First, to set a starting point in how to assess 'change', we examined national institutional settings and policies in the field of school-to-work transitions. In line with Bonoli (2010) and Dingeldey's (2011a) typologies of activating employment policies, we distinguished between a work-first approach and an enabling approach. These approaches can be found in combination with the institutionalization of different school-to-work transitions guided by primarily school-based or dual vocational training systems (Eichhorst et al., 2015;Solga et al., 2014). Gangl (2001) and Hora et al. (Chapter 7 this volume) suggest that overall a dominant (dual) apprenticeship system implies rather smooth transitions from school to work, whereas in school-based systems a large proportion of low-skilled labour market entrants and a lack of in-work experience together lead to high youth unemployment. Additionally, the PES is important in providing unemployment benefits and services like counselling, placement in training measures and jobs. Hence, relevant indicators for the efficient implementation of such policies were the administrative capacity of the respective PES (caseload and financing), the incentives provided for young unemployed people to register and the governance structure of the PES (centralized/decentralized).
Second, building on the Europeanization literature, we assessed the particular European instruments of vertical coordination in relation to the YG: the provision of EU funding and the 'YG implementation plans'. We treated EU funding through the YEI as a mechanism of external support. We examined whether the EU achieved the goal of providing particular support to countries with the highest problem pressure or if specific regulations might have caused problems. We considered the YG implementation plans both as mechanisms of external support as well as external pressure. On the one hand, the plans provide a framework for country-specific goal-setting, developing indicators and policy assistance by European actors. On the other hand, these plans also put pressure on member state governments, given that the European Semester monitors their implementation. We understand the monitoring as a sort of external pressure in the form of 'naming, faming and shaming' (Zeitlin et al., 2014). We investigated whether the YG implementation plans were built on a partnership approach, delivered according to the required rules (punctuality) and/or whether a country had to reformulate the plans. The investigation provides insights into how the mechanisms of vertical coordination used influenced procedural processes in the member states.
Third, we analysed whether the YG implementation was in line with the previously established approach for youth policies or whether it followed a different path, implying a need for substantive change. This assessment is based on the analysis of national policy discourses, the newly introduced programmes, and reform initiatives in the field of employment and vocational training.
As sources, we used secondary literature and official documents as well as data from seven interviews with EU officials and stakeholders conducted in spring 2017. In addition, we drew on national reports written by NEGOTIATE project partners based on four to five expert interviews conducted in summer 2016 in each country at national and local level.

CONTRASTING APPROACHES OF YOUTH EMPLOYMENT POLICIES
Member states' active labour market policy varied according to the level of youth unemployment and the institutions of school-to-work transitions in the three countries (see Table 9.2). When youth unemployment was high, member states tended to prioritize quick labour market integration, while member states with lower youth unemployment rates were more likely to provide support for vocational training. Differences in education and vocational training systems reinforced such contrasts. Accordingly, Spain and Greece pursued a work-first approach, where employment policy was aimed at integrating young people quickly into jobs, for instance by providing subsidies for employers (Ayllón and Ferreira-Batista, 2016;Kominou and Parsanoglou, 2016;OAED, 2013). By contrast, Germany had practiced an enabling policy approach, where measures were focused on the attainment of school or vocational training certificates as intermediate steps towards labour market integration (Dingeldey et al., 2017). The historical development of the respective member states' institutions supporting young people in their school-to-work transitions had influenced these contrasts. Germany has a long-established dual vocational training system. Since the initial vocational education and training system at upper-secondary level became popular, more than 50 per cent of all students have enrolled in it. Although the social service professions have relied on school-based vocational education and training systems, still more than 40 per cent of all students have enrolled in the dual-track employment-based systems (Autorengruppe Bildungsberichterstattung, 2014;OECD, 2014).
By contrast, in Greece and Spain, a school-based vocational education and training system has dominated, while work-based training has played a minor to marginal role (OECD, 2014). Dual tracks either did not exist or the governments provided only a few dual-track places (ReferNet Greece, 2009;ReferNet Spain, 2012; also see Dingeldey et al., 2017).
Despite trends towards municipalization in labour market policy and decentralization in the context of New Public Management, the German PES has remained a centralized national agency. Vertical coordination has been strong, meaning the level of flexibility in delivery at regional or municipal level has been low (Dingeldey, 2011b;Mosley, 2008Mosley, , 2011. In 2014 the capacity of the PES in Germany was high, with a comparably low annual average caseload of less than 150 clients (of all clients served by staff in the PES). The expenditure on such services (as a percentage of GDP) was above the EU average of 28. The regulation of access to unemployment benefits gives moderate incentives for young people to register with the PES, resulting in coverage rates of the young unemployed of 50 or more per cent in Germany (Matsaganis et al., 2013).
In Greece the structure of the PES and of labour market policy has also remained centralized notwithstanding decentralization trends in recent decades (Kominou and Parsanoglou, 2016;Kyvelou and Marava, 2017). By contrast, the PES in Spain has been decentralized. The level of flexibility in delivery at the regional or municipal level has been medium and the autonomous communities have even had their own vocational education and training systems (Mosley, 2008(Mosley, , 2011. Both in Greece and Spain there were indications of weak capacity in the PES. The expenditure on PES was under the EU average in both countries, although Spain spent slightly more (0.144 per cent of GDP) on labour market services than Greece (0.012 per cent of GDP; European Commission and ICON Institute, 2016; Eurostat, 2016). Nevertheless, annual average caseload was very high in Spain at 2683 in 2014, while we may regard the annual average caseload of 488 in Greece as 'medium' but still too high to provide effective counselling. With coverage rates of unemployment benefits under 15 per cent in Greece and Spain, the incentives to register were very limited (Matsaganis et al., 2013).
These indicators suggest that even before the YG was launched, Germany not only followed an enabling labour market policy approach, but also combined a rather well-established PES and a comparatively high rate of registration with a vocational education and training system. By contrast, young Spaniards and Greeks were more likely to be unsupported in their transition from school to work. School-based vocational training systems were established in combination with a practiced work-first approach and a rather overloaded PES that did not register all unemployed young people.

YOUTH GUARANTEE COORDINATION AND IMPLEMENTATION -PROCEDURAL AND SUBSTANTIVE CHANGE
Before addressing the procedural and substantive changes, we briefly outline selected instruments of European vertical coordination, namely funding and the preparation of YG implementation plans. In 2013 the external funding supplied by the EU within the framework of the newly created YEI -the 6.4 billion euros provided initially in 2014−15 − were extended by another 2.4 billion euros for the period 2017−20 to support the member states actively in implementing the YG (Council of the European Union and EMCO, 2016). This money is available to regions that had a youth unemployment rate above 25 per cent in 2013. Thus, the amount of funding provided relates to the level of problem pressure in the different member states. Spain therefore received 881.44 million and Greece 160.24 million euros, while Germany did not receive any money from the YEI (European Commission, 2014a). EU financial support makes up a substantial share of total spending on youth employment policies in the Southern European countries. Furthermore, the EU recommended using money from the ESF for the YG implementation (see Bussi et al., Chapter 10 this volume). The YEI has been part of the ESF framework and control structure and thus required co-financing from the member states (Council of the European Union, 2013). Accordingly, they developed 'operational programmes' that had to be approved by the Commission. Later the member states had to submit implementation reports (European Commission, 2014b; Interview EC; Interview UEAPME). Particularly the principle of reimbursement, which has meant that member states had to finance projects in advance, caused difficulties to countries with deficit targets and led to delays (European Commission, 2015a). In particular, Spain had problems with the principle of reimbursement, especially as it was also under EU pressure to cut the public deficit (Ayllón and Ferreira-Batista, 2016). Many countries claimed that they did not have the national budget to release advance funding for YEI measures (European Commission, 2015a). In 2015, in reaction to the delays and problems indicated, the Commission increased the 'pre-financing' from part of the EU to member states by around one billion euros (European Commission, 2016b). Subsequently, the YEI's financial resources allocated to selected projects rose between 2015 and 2017 from 36 per cent to 68 per cent. Nevertheless, about one third of the total budget has not yet been allocated (European Commission, 2017b).
In Greece, delays in the withdrawal of funding were connected to the role of the 'YG National Coordinator' (Ministry of Labour): 'Responsible for the distribution of resources is the "National Coordinator" and we do not know why they did not proceed so that funding could be absorbed. The other stakeholders had few and poor proposals, but the "National Coordinator" should put some pressure on them' (interviewee in the Greek PES, cf. Kominou and Parsanoglou, 2016: 13).
Other reasons for late or no withdrawal might be that the complexity of the application process created uncertainty amongst national decision makers as to whether they would receive reimbursement of the costs of the presented projects. This might have led to so-called 'gold-plating', meaning that member states and public administrations might refuse good projects or initiatives if they were not sure whether the projects would meet the EU criteria (Interview UEAPME).
The YG implementation plans have represented country-specific goalsetting supported at EU level. The plans have described the measures and reforms that the countries intended to implement in order to comply with the YG, including the time frame as well as the foreseen funding and responsibilities. As mentioned before, the European Semester has monitored the implementation of the YG. The YG National Coordinator has been the main point of contact to communicate with the European Commission and has led the establishment and management of the YG (European Commission, 2017a). The preparation of the YG implementation plans has been a crucial element in designing and realizing member states' involvement of non-state actors as well as horizontal coordination across policy fields. Hence, the YG implementation plans might have led to changes with respect to established procedures and systems of coordination.

PROCEDURAL CHANGE IN HORIZONTAL COORDINATION?
In the countries under study, the ministries of labour acted as National Coordinators to manage and coordinate the design and implementation of the YG, while each respective PES has been the central operative institution. The involvement of different state and non-state actors such as relevant ministries, social partners and other stakeholders has been important for implementing the partnership approach and launching structural reforms. Whereas Greece and Spain created new formal or informal institutions, Germany used previously existing bodies. For the YG design and implementation, Germany made use of several already established forms of cooperation between schools and vocational guidance services, PES and industry organizations (YGIP-Germany, 2014: 21-6). When designing the German YG implementation plan, the Ministry of Labour invited several ministries, social partners, welfare associations, PES and representatives of municipalities to discuss and provide written feedback to a draft version. The participation of social partners in single policies in Germany has varied but it has been intensive in vocational training (Assmann et al., 2016;YGIP-Germany, 2014). In line with established procedures, the role of non-state actors in the consultation process to design and implement the YG has been of a quite participative character. Nevertheless, the national trade union confederation criticized the denial of a proposed apprenticeship guarantee and also the timing of the hearing for giving them little opportunity to prepare remarks (Bussi, 2014: 33).
Furthermore, already existing horizontal coordination forms across policy fields have been further strengthened and refined under the YG scheme. An important innovative reform has been the establishment of one-stop youth career agencies to combine PES, educational measures, social youth services and other relevant institutions to support school-towork transitions at local level (Assmann et al., 2016: 6, 24-33 however, has not been the trigger, rather has had an overall supportive impact in these developments as the respective reforms were begun in 2010.
The centralized PES's governance structure in Germany may have been advantageous for implementing the YG. However, some of the actors involved have criticized centralization because sometimes the local employment agencies had to wait for the consent of the national PES, which has hindered the rapid implementation of some measures at local level.
Although in Spain the Labour Ministry has been the formal point of contact for the Commission, it does not have the centralized power to coordinate the YG. Due to decentralized coordination of the YG implementation, we find strong regional differences in the YG design and implementation in combination with a poorly equipped PES. Spain did not set up formal coordination committees for implementing the YG, rather has made use of informal multi-stakeholder bodies, including nonstate and state actors, ministries, the Youth Council, youth organizations, autonomous communities, the Spanish Federation of Municipalities and Provinces, and the PES (European Commission, 2016a: 24; YGIP- Spain, 2013: 14-16). When further sources are taken into account (Ayllón and Ferreira-Batista, 2016;BusinessEurope, 2014BusinessEurope, , 2015BusinessEurope, , 2016Bussi, 2014), there seems to be a gap between what is described in the Spanish official documents and statements from regional actors and trade union and employer representatives: On the one hand, the Spanish 'YG implementation plan' indicates that it has received various contributions from interested parties and that it has passed a prior consultation before approval. The reason could be that in some autonomous communities several YG pilot projects were conducted prior to 2013 where stakeholders had been consulted (Ayllón and Ferreira-Batista, 2016). On the other hand, Spanish trade unions, employers and autonomous public employment services have noted that participation in the YG at national level has been poor, notwithstanding their requests for information and involvement. Consultation meetings gave information about the finalized YG implementation plan but did not allow for feedback (Ayllón and Ferreira-Batista, 2016;Bussi, 2014). Hence, the Spanish social partners expressed very strong dissatisfaction with the social dialogue in the YG process (BusinessEurope, 2014(BusinessEurope, , 2015(BusinessEurope, , 2016. Thus, the informative character seemed to dominate when it came to the involvement of non-state actors. Moreover, it appears that decentralization in Spain has not only led to regional differences but has also counteracted the coordination of different administrations. The competition of power between the autonomous communities and the national government even led to parallel registration systems for the YG implementation, creating complex bureaucratic procedures. Young Spaniards who had already registered as unemployed at the regular PES system additionally had to register in a particular system for the YG. Furthermore, the inscription modalities were criticized as complicated and not target-group oriented since it was the young people who needed to get actively involved. The general weakness of the governance structure of the PES has created overall obstacles to efficient implementation and coordination of youth employment policies as well as to evaluation and monitoring (Ayllón and Ferreira-Batista, 2016: 11).
In Greece informants did not report on major regional differences. However, the EU Employment Committee has addressed the inefficiency of its PES in the implementation of the YG in the Country Specific Recommendations (Council of the European Union, 2013; Council of the European Union and EMCO, 2016). In contrast to the other two countries, the Greek Ministry of Labour as National Coordinator has concentrated on the establishment of a formal institution with a particular focus on the YG. The secretaries of relevant ministries set up a 'Coordination Committee for the implementation of the Youth Employment Initiative, and in particular of the YG programme'. However, deficiencies in the social dialogue, amongst other problems, resulted in a revision of the YG implementation plan (Bussi, 2014: 40;Kominou and Parsanoglou, 2016: 5).
The revised version of the plan demanded non-state actors' involvement and stated that the Committee was to be comprised of social partners, civil society representatives and youth employment experts, ignoring the optional character of their participation. Additionally, the government established a 'Working Group on implementing the Youth Employment Initiative and YG' that also included representatives from relevant ministries as well as the Association of the Regions and the Central Union of Municipalities of Greece. However, the involved stakeholders assessed the participation in these coordination bodies in contrasting ways. First, since the government did not consult certain stakeholders in the process of designing the YG implementation plan, we may see the inclusion of non-state actors as of a mainly informative nature. Nevertheless, for the Greek Ministry of Labour, horizontal coordination was quite a challenge, since it was the first time that they had to work effectively together on a specific basis (Kominou and Parsanoglou, 2016: 9). Thus, the setting up of coordination bodies for the YG encouraged Greece to address youth employment policy from a more holistic perspective by reinforcing horizontal coordination across policy fields.
In summary, we can identify several procedural changes in horizontal coordination in all three member states, although in Germany the YG was not a trigger for these changes. We cannot identify a clear change in the involvement of non-state actors in Germany since this country used preexisting bodies. It seems that external support via the YEI, the preparation of the YG implementation plans and external pressure due to monitoring processes have together triggered procedural changes in Greece and Spain. However, it is not clear whether the procedural changes will have a sustainable character in these countries since the creation of coordination bodies related only to the YG and the YEI.

SUBSTANTIVE CHANGE OF EMPLOYMENT APPROACH AND INSTITUTIONAL SETTINGS?
The YG triggered changes in the discourse and promoted a stronger focus on youth unemployment at national level and, occasionally, in local administrations in Germany, Spain and Greece. Moreover, the term 'young NEETs' has received more attention from policymakers since the implementation of the YG in the three countries (Assmann et al., 2016;Ayllón and Ferreira-Batista, 2016;Kominou and Parsanoglou, 2016).
Overall, this has contributed to putting topics related to youth employment policies at the top of the political agenda and therefore to substantive change in agenda-setting. However, in Germany, we see no substantive change concerning the policy approach to youth employment, rather a path-dependent implementation of the YG according to the already dominating enabling approach for the young. The majority of educational and labour market measures in the YG remained preventive and aimed at pupils, jobseekers or training seekers, or young unemployed with a focus on the attainment of school or certified vocational training qualifications. Similarly, Spain and Greece developed the YG through a path-dependent implementation, albeit by pursuing a work-first approach. For instance, labour market policy measures have given financial incentives to companies to hire adolescents in times of uncertainty. The Greek YG has included several voucher programmes combining short training periods with work experience. Spanish policymakers considered an incentive for hiring often in combination with the provision of atypical contracts (Ayllón and Ferreira-Batista, 2016: 21;Kominou and Parsanoglou, 2016: 9, 18).
Beyond path dependency and according to our hypothesis, we were able to identify steps towards substantive programmatic change concerning school-to-work transition systems in both countries. The Greek government sought to make the vocational education and training system more attractive. A new legal framework for apprenticeships (Law 4186/2013) was created in 2013 (Kominou and Parsanoglou, 2016: 6;YGIP-Greece, 2014: 24-5). This framework seeks to connect vocational education and training more strongly to the economy and the labour market. Another innovation in line with these objectives was to introduce a dual system by establishing the 'apprenticeship class' (YGIP- Greece, 2014: 25). Vocational training schools offered a fourth optional year of an apprenticeship programme to provide workplace experience that led to a higher qualification for upper-secondary vocational graduates. Furthermore, to strengthen the link between labour demand and supply, vocational training schools set up career offices (Kominou and Parsanoglou, 2016: 6-8;YGIP-Greece, 2014: 25). However, according to national reports, the recession was a limiting factor. Many Greek companies lacked the necessary structures and financial resources for apprenticeship training (Kominou and Parsanoglou, 2016: 12, 28).
Furthermore, institutional reforms in Greece restructured the PES (OAED) according to the 'Business Model Reengineering Plan', highlighted as a crucial factor for delivering the YG. The reform aimed at internal changes such as a better alignment between organizational units, but more importantly concerned the way in which the PES approached unemployed people. The objective was to treat them in a more individualized way and to set up Individual Action Plans (IAP; YGIP- Greece, 2014: 17-20;Kominou and Parsanoglou, 2016: 4-8). The reform also included the establishment of an online portal and a call centre for employers and jobseekers. However, these reforms had already started before the implementation of the YG, promoted by international institutions as part of fiscal discipline policies under the Memoranda of Understanding between Greece and its creditors (Kominou and Parsanoglou, 2016: 23).
In Spain a foundation was set up in 2012 to 'establish the basis for the progressive implementation of a dual training system' (Royal Decree 1529/2012). The aim was to facilitate labour market integration for young people by matching vocational skills with labour market needs (Ayllón and Ferreira-Batista, 2016: 6, 22;YGIP-Spain, 2013: 27-8). The reform offered several modalities of vocational education and training. These offers included the option to provide training combined with employment exclusively within an educational institution or within an enterprise. An alternative was to offer young people training by a training centre in combination with work-based training at an accredited company (ICF GHK, 2012: 6). Again, since these developments started before 2013 we cannot see the YG as the trigger. Overall, financial resources within the YG may have supported the increase in the number of participants from 4292 in 2012/13 to 15 304 in 2015/16. Within that time also the numbers of companies offering work-based learning rose from 513 to 5665 (European Commission, 2016a: 59). Nonetheless, at the time of writing it was too early to estimate the outcome of a proper vocational education and training system.
If we consider all the changes observed in Germany, we see that they were primarily related to procedural changes and we can characterize them as 'system refinement'. By contrast, we regard substantive changes in the Southern countries as cases of 'system-building' (see Table 9.3). These may represent initial steps of transition towards a different institutional setting and policy approach.

CONCLUSION AND DISCUSSION
Overall, the findings presented support our expectation that the implementation of the YG would be path dependent, in the sense of largely reproducing and strengthening a pre-existing approach to domestic youth unemployment policies. German YG measures built on a previous enabling orientation, whereas Spain and Greece tended to broaden the scope of an already dominating work-first approach to labour market policy measures for young people. If we adopt a broader perspective including relevant institutions like the PES as well as the system for vocational education and training, we can observe different forms of procedural and substantive changes. Although the YG Recommendation did not necessarily trigger these changes, mechanisms of vertical and horizontal coordination have supported them. Furthermore, there are reasons to assess the changes in the Southern European countries as bigger and more consequential than the changes in Germany. A more noticeable institutional misfit with respect to settings for providing smooth school-to-work transitions in Southern Europe than in Germany may have contributed to this difference. Moreover, both the external pressure and the support from the EU have been stronger in the Southern European countries than in Germany.
However, according to stakeholders in all countries, both bureaucratic rules to claim money and the principles of reimbursement created problems overall in accessing EU funds, especially for member countries with deficit targets. This emerged as an area where there was scope for improving vertical coordination at EU level. Although the EU has already responded to these problems, for example with so-called 'pre-financing', further improvements seem to be necessary to ensure support for the member states and the organizations implementing ESF and YEI programmes (also see Bussi et al.,Chapter 10 this volume).
Within this context one may also ask whether other forms of external pressure not dealt with in the present analysis (such as requirements formulated by the Troika as part of crisis management) might have hampered substantial reforms in the respective countries. For instance, austerity policies may have prevented a necessary increase in the capacity of PES that might have enabled improvement of the staff−client ratio and an increase in the social protection to which young people would have access.
Probably the most positive influence of the YG in all three countries has been a greater awareness of the negative consequences of youth unemployment and job insecurity. At best the YG has encouraged member states to address youth employment policy from a holistic perspective. Improved horizontal coordination between different ministries and administrations, as well as the participative involvement of social partners and other stakeholders, are likely to be crucial for establishing new institutional settings that can provide smooth transitions from school to work. Finally, continuity of the financial commitment to and political interest in the support of youth employment policies through the Commission and other European actors are essential for member states' ability to combat youth unemployment in effective ways.