Socializing the European Semester: EU social and economic policy co-ordination in crisis and beyond

ABSTRACT This contribution analyses how EU social objectives and policy co-ordination have been integrated into the Union’s post-crisis governance architecture. It argues that between 2011 and 2016, there was a partial but progressive ‘socialization’ of the ‘European Semester’ of policy co-ordination, in terms of increasing emphasis on social objectives in its priorities and key messages, including the Country-Specific Recommendations; intensified social monitoring and review of national reforms; and an enhanced decision-making role for EU social and employment actors. In explaining these developments, the contribution highlights the contribution of strategic agency, reflexive learning and creative adaptation by social and employment actors to the new institutional conditions of the Semester, building on recent theoretical work on ‘actor-centred constructivism’ and the ‘usages of Europe’.


Introduction
Since the onset of the euro crisis, the European Union (EU) has introduced a series of far-reaching changes in its socio-economic governance architecture (see the introduction to this collection). At its core is the 'European Semester' of policy co-ordination, through which the Commission, the Council and the European Council set priorities for the Union; review national performance and reform programs; and issue Country-Specific Recommendations (CSRs) to member states, backed up in some cases by possible financial sanctions. The European Semester (henceforth 'Semester') brings together within a single annual policy co-ordination cycle a wide range of EU governance instruments with different legal bases and sanctioning authority. This process has given the EU institutions a more prominent role than ever before in scrutinizing and guiding national economic, fiscal and social policies, especially within the euro area.
The advent of the Semester has raised a set of contested questions about the relationship between social and economic policy co-ordination within the EU's new post-crisis governance architecture. Has the integration of EU social policy co-ordination, as developed through the Open Method of Coordination on Social Protection and Social Inclusion ('Social OMC'), into the Europe 2020 Strategy and the Semester resulted in its subordination to fiscal discipline and budgetary austerity objectives imposed by the main economic policy actors? 1 Or has such integration instead offered new opportunities for social and employment policy actors 2 to defend and mainstream EU social objectives, such as the adequacy and accessibility of pensions and health care or the fight against poverty and social exclusion, throughout the Semester process?
Critiques of the Semester's social dimension have revolved around three interrelated empirical claims. The first concerns the subordination of social objectives to higher-order economic goals within the EU's post-crisis governance architecture. Thus Crespy and Menz (2015a: 762) contend that 'the slippage of Europe 2020 into the European Semester … has meant the further absorption of social policy into macroeconomic policy'. Hence, in their view, EU 'social policy is becoming increasingly … subsumed to economic objectives focused on competitiveness, narrowly defined as low labour costs … and stringent fiscal discipline' (Crespy and Menz 2015b: 199-200). For Degryse et al. (2014: 70), similarly, 'the CSRs reflect a particular concept of the European economic model … focused on growth and competitiveness, while neglecting … the principal role of social policies … '. More strongly still, Copeland and Daly (2015: 150) assert that 'the European Semester has served as a focal point to organize a form of economic governance in which national budgetary discipline and the correction of macroeconomic imbalances … have been made the principles on which all other policy objectives are dependent'.
The second claim advanced by the Semester's critics concerns the dominance of EU economic policy actors over their social and employment policy counterparts within its decision-making procedures. Thus as de la Porte and Heins (2014: 169-70) contend, 'Europe 2020 is dominated by DG ECFIN and the ECOFIN Council with a very marginal role for the European social policy actors' (cf. also Degryse et al. 2014). Such power imbalances within the Semester, they argue, reflect the fact that the economic actors 'operate in areas where the EU has strong jurisdiction', unlike their social counterparts, who are responsible for fields 'where the EU has only weak legislative competence' (de la Porte and Heins 2015: 12). For Copeland and Daly (2015: 150, 153), likewise, the structure of the Semester privileges economic policy actors and 'formalizes the historically disadvantaged position within the EU's political hierarchy of social actors'.
The third claim underpinning the social critique of the Semester concerns its coercive and prescriptive character. Many commentators have seen the process as a centralized mechanism for imposing structural reforms on member states, including in areas of primary national competence such as social and employment policy, through CSRs underpinned by the threat of sanctions under the Stability and Growth Pact (SGP) and Macroeconomic Imbalance Procedure (MIP). Dawson (2015: 984), for example, asserts that the new 'post-crisis economic governance in many ways entails a return to "command-and-control" regulation', whereby 'national diversity … is often placed within strict limits with high levels of supra-national policy prescription'. While Dawson (2015: 984-6) and Degryse et al. (2014: 72, 76) highlight the uniform 'one-size-fits-all' message embodied in the CSRs, other critics such as Scharpf (2013: 137-8) decry instead their discretionary application, involving 'a radical extension of hierarchical European controls over national policy choices'. Either way, however, de la Porte and Heins (2014: 170) contend that 'the European Semester and the instruments designed to re-enforce it are highly intrusive along the dimensions of policy interference, surveillance, as well as coercion'. Within the Semester, they conclude, fiscal and budgetary governance instruments 'have become more precise in terms of objectives, and stricter in terms of surveillance and enforcement', while social and labour market policy instruments remain weaker on these dimensions and thus in their 'potential impact' (de la Porte and Heins 2015: 10).
By contrast, a second body of recent research has drawn attention to the gradual evolution of the Semester in a more socially balanced direction in relation to each of the preceding empirical claims. Thus Bekker has analysed the changing content of the CSRs, finding not only an increasing emphasis on social and employment issues, but also that even recommendations issued under the SGP and the MIP sometimes promote EU social objectives such as accessibility of health care and the fight against poverty and social exclusion (Bekker 2015;Bekker and Klosse 2013). Coding the policy measures contained in the CSRs between those promoting 'social investment' as opposed to austerity or 'social retrenchment', Crespy and Vanheuverzwijn (2016: 77-8, 85-6) estimate that the proportion of the former has increased from 50 per cent to 64 per cent between 2011 and 2016. 3 Focusing on poverty and social inclusion, Jessoula (2015: 496) observes that the growing number of CSRs on this issue was 'the result of the process whereby DG EMPL learnt how to defend and support the Europe 2020 social dimension in inter-DGs bargaining at the various stages of the Semester'. Finally, Bekker has examined the CSRs and National Reform Programs (NRPs) of France, Germany, Poland and Spain between 2009 and 2014, concluding that while the degree of flexibility varies across countries and over time, 'the European Semester, including the SGP, allows for developing alternative socio-economic policies' (Bekker 2016: 62).
Building on this latter body of work, along with two earlier research reports of our own (Zeitlin and Vanhercke 2014;Vanhercke et al., 2015), this contribution analyses the evolution and dynamics of the Semester from its inception in 2010 through the 2016 cycle, bringing new evidence to bear on the main disputed questions about the relationship between social and economic policy co-ordination in the EU's post-crisis governance architecture. The contribution makes two major contributions to this debate. The first is empirical: drawing on 76 interviews with key policy players as well as a systematic analysis of EU documents, we demonstrate that over this period there was a partial but progressive 'socialization' of the Semester, both in terms of its substantive content and its governance procedures. This socialization, as we define it, comprises: 1) a growing emphasis on social objectives in the Semester's policy orientations and messages, embodied in the AGS and especially the CSRs; (2) intensified monitoring, surveillance, and review of national reforms by EU social and employment policy actors and (3) an enhanced role for these actors relative to their economic policy counterparts in drafting, reviewing and amending the CSRs. Finally, regarding the Semester's prescriptive and coercive character, the contribution shows that a concerted pushback by the Employment, Social Policy, Health and Consumer Affairs (EPSCO) Council and its advisory committees against the Commission's perceived efforts to impose uniform, over-detailed recommendations on social and employment issues resulted in significant revisions to the Semester's decision-making arrangements, including the CSRs themselves, which have made them less hierarchical and more interactive. The second major contribution is explanatory and theoretical. In accounting for the empirical developments traced in the contribution, we highlight the contribution of strategic agency, reflexive learning and creative adaptation to the new institutional conditions of the European Semester by the key actors, especially on the social and employment side.
The remainder of the contribution is organized as follows. Section 2 presents the theoretical approach, building on 'actor-centred constructivism' and the 'usages of Europe', along with the data and methods supporting the empirical analysis. Section 3 examines the relationship between social and economic policy co-ordination from the Lisbon Strategy to Europe 2020 and the inception of the Semester. Section 4 analyses the socialization of the Semester under the Barroso Commission (2012)(2013)(2014), in terms both of policy orientations and governance procedures. Section 5 shows how this socialization was further institutionalized under the Juncker Commission. The final section summarizes the empirical evidence and revisits our contribution to the ongoing academic debate.

Theoretical approach, data and methods
In explaining the partial but progressive socialization of the European Semester between 2010 and 2016, this contribution focuses on the strategic agency and creative adaptation to changing institutional conditions of the key actors concerned. Here we draw on and extend recent theoretical work on 'sociological approaches' to European integration and policy-making, notably 'actorcentred constructivism' (Saurugger 2009(Saurugger , 2016 and the 'usages of Europe' (Jacquot and Woll 2003;Woll and Jacquot 2010). These approaches provide valuable conceptual resources for understanding how apparently weaker players, such as the EU social and employment policy actors, can advance their substantive goals and decision-making influence in complex multilevel governance processes like the European Semester, despite the institutional asymmetries and structural power imbalances emphasized by the critical literature discussed in the previous section.
As presented by Saurugger, 'actor-centred constructivist' approaches to the EU diverge from standard rationalist and historical institutionalist analyses that overstate the weight of structural constraints and path dependence on actors' choices, while underestimating the 'possibilities for actors to modify the rules and institutions in which they act'. Such actor-centred approaches, in her formulation, likewise differ from classic constructivist and sociological institutionalist accounts that emphasize how dominant cognitive frames and taken-for-granted norms shape actors' behaviour, thereby downplaying their 'room for manoeuvre in power struggles and social conflicts'. In this view, too, actors' capacities for action are not 'entirely predefined' either by their resource endowments or by the structure of the field in which they operate; instead actors 'are able to adapt and to change their environment' (Saurugger 2016: 73-5).
Work on the 'usages of Europe' gives concrete form to actor-centred constructivism as defined above by analysing how different types of actors identify and strategically exploit opportunities within EU multi-level governance processes to advance their own preferences and objectives. The concept of usages 'highlights how actors engage with, interpret, appropriate or ignore the dynamics of European integration'. This means that actors can fail to seize opportunities for action identified by external observers, either because they do not perceive them as such or because they are more concerned with the possible risks than the potential benefits involved. Moreover, while actors intentionally make use of the opportunities and resources they find in EU institutions and policies to pursue their own goals, the outcome may not be identical to the original objective, since 'the effects of an action are often not entirely predictable or controllable'. Finally, 'as strategic as usages may be initially', in the longer term, as Woll and Jacquot (2010: 115-6) observe, they 'entail cognitive and/or normative adaptation by actors and their political environment, which in turn affects their subsequent behaviour and positioning'. Just as there is 'no impact of Europe without usage' by the actors (Jacquot and Woll 2003: 5-6), so too as another scholar working within this approach remarks, there are 'no usages of Europe without impact' (Sanchez Salgado 2014: 203).
Although the usages of Europe, like actor-centred constructivism, has been conceived as a general conceptual approach to European integration and policy-making, it has mainly been applied to the strategic identification and 'creative appropriation' of opportunities arising from EU governance processes (including both the OMC and the Lisbon Strategy) by national-level actors to advance their own domestic objectives and agendas (Barcevičius et al. 2014;Graziano et. 2011;Jacquot and Woll 2003;Woll and Jacquot 2010). But as their proponents explicitly envisage, these approaches can also be used to illuminate the evolution and dynamics of EU governance processes themselves. Thus, for example, the 'usages of Europe' approach draws attention to how actors situated at the intersection between national and supranational policy-making (such as EU committees of member-state officials) may strategically exploit the ambiguities of European concepts, rules and procedures in order to reframe policy issues, build political coalitions, enhance their institutional influence, and justify decisions taken at the EU as well as the domestic level (Woll and Jacquot 2010: 116-7).
In this contribution, we extend the insights of these agency-focused and possibility-orientated approaches by tracing how reflexive learning from past experience by key actors, especially on the social and employment side, together with creative adaptation of their own organization and practices to the new institutional conditions of the EU's post-crisis governance architecture, has contributed to the partial but progressive socialization of the Semester. Here we focus on how EU social and employment policy actors have deliberately revised their working methods and styles of argument to gain traction in 'evidence-based' deliberation with their economic policy counterparts. At the same time, we analyse how the arguments and demands advanced by the social players through the Semester process have helped to change the broader 'rules of the game' and influence the positions of their economic policy interlocutors.
In terms of data sources, this contribution draws on six rounds of elite interviews, conducted between 2010 and 2016, with current and former members of the European Commission (multiple DGs); Council advisory committees (EMCO, SPC, EPC, EFC); EU social partners and NGO networks; the European Council and Council Secretariat; and the European Parliament. A total of 76 interviews were conducted with 61 separate individuals. Some interviews involved multiple participants, while 17 people were interviewed two to four times in light of their key role in the development of the Semester. Nearly all interviewees agreed to be quoted, provided that the source could not be identified, e.g., by naming only their institutional affiliation rather than their specific position. Interviews cited in the text are referred to by a unique code, the key to which is presented in Online Appendix Section 1, along with a complete list of interviews by date.
This contribution is also based on near-complete access to the papers of the Employment Committee (EMCO) and the Social Protection Committee (SPC) during this period. These papers, which include documents from the EU economic policy actors, allow a fine-grained process tracing of decisionmaking within each Semester cycle, and were used to triangulate information from the interviews. For reasons of space, we have supported empirical claims in the main text with more extensive quotations from interviews and EU documents in Online Appendix Section 4. 4 Given the nature of our sources, the analysis in this contribution focuses on the evolution and dynamics of the European Semester at EU level, rather than its causal influence in the member states, which requires a different research approach, based on comparative process tracing of national policy-making.
The European Semester and the Europe 2020 strategy: advancing or subverting Social Europe?
The relationship between social and economic policy co-ordination has been a hot-button issue at EU level since the late 1990s. A key impetus for the creation of the OMC processes on pensions and health/long-term care in the early 2000s came from EU social policy actors' efforts to ensure that the social objectives of these fields were not eclipsed by the fiscal and budgetary focus of EU economic policy co-ordination. Hence the objectives of the pensions and healthcare OMCs included social adequacy, quality and accessibility as well as financial sustainability (Barcevičius et al. 2014: ch. 2).
The Europe 2020 Strategy, adopted in 2010, was designed to have a stronger social dimension, compared to the preceding Lisbon Strategy, which had focused primarily on growth and jobs after its 2005 relaunch. Europe 2020 featured 'inclusive growth', by 'fostering a high employment economy delivering economic, social and territorial cohesion', as one of three overarching priorities, along with an EU-wide headline target, aimed at lifting 'at least 20 million people out of the risk of poverty and exclusion'. It also incorporated for the first time a separate guideline on 'Promoting social inclusion and combating poverty', which highlighted the contribution to social cohesion of pensions, healthcare and public services. At the same time, however, there was a founding ambiguity within the Strategy about the relationship between fiscal and macroeconomic surveillance by the ECOFIN Council and 'thematic co-ordination' by the other Council formations of progress towards the Europe 2020 goals, targets and guidelines (Barcevičius et al. 2014: 31-4).
Initial experiences under the Semester seemed to confirm critics' worst fears that the new integrated EU policy co-ordination framework would result in the subordination of social objectives to fiscal consolidation, budgetary austerity and welfare retrenchment imposed by economic policy actors (Pochet 2010). The first Annual Growth Survey (AGS) and CSRs 5 focused primarily on fiscal consolidation, while pushing for financial reform of pension and healthcare systems, tax and benefit shifts to 'make work pay', revision of employment protection legislation and measures to contain wage growth (Bieling 2012: 264). Only three countries received CSRs explicitly addressing poverty and social inclusion, despite weak national targets, which the Commission acknowledged would not together meet the EUwide poverty target by 2020 (European Commission 2011: 5). But many member states also received recommendations urging them to enhance the effectiveness of active labour market policies for vulnerable groups (10); expand childcare provision (5); reduce school drop-out rates, improve access to education and training and encourage lifelong learning (9). Ten of the fifteen countries receiving CSRs on reducing early labour-market exit and linking the statutory retirement age to life expectancy were likewise urged to ensure the adequacy of pension benefits, especially through active ageing measures aimed at maintaining the employability of older workers (own analysis based on European Parliament 2013; 6 cf. Clauwaert 2014: 11; Gómez Urquijo 2017: 7.) Overall, Crespy and Vanheuverzwijn (2016: 77-8) estimate that the social and employment policy measures recommended in the 2011 CSRs were evenly divided between those promoting social retrenchment and social investment.
Procedurally, the first Semester cycle followed a prescriptive 'one-size-fitsall' approach, with limited adaptation of the CSRs to the specific situation of individual Member States (Bekker 2013: 15-16). The EU institutions explicitly sought to use multilateral surveillance by Member States of each other's policies and performance to exert peer pressure for implementation of structural reforms (EMCO 2012; President of the European Council 2011). Social actors at both EU and national level conversely found themselves largely excluded from preparation and review of the National Reform Programmes (NRPs) and CSRs (SPC 2011b).
At the same time, the Commission withdrew its support for the Social OMC as an autonomous process. Member states were no longer requested to produce National Social Reports (NSRs), while the Commission also withdrew from the production of the annual Joint Report on Social Protection and Social Inclusion. A key motivation in both cases was the insistence by the Commission, particularly its Secretariat-General (SECGEN), that there should be no parallel policy co-ordination processes outside Europe 2020, which one highranking Commission official called 'the only game in town' for Social Europe, and that social reporting should be channelled exclusively through the Semester (Barcevičius et al. 2014: 34-5; SPC 2011a; COM-COORD6; COM-EMPL7).

Policy orientations
As the sovereign debt crisis within the eurozone morphed into a broader economic downturn, leading to a rapid erosion of public support for the EU, a significant rebalancing between social, economic and employment objectives became visible in the policy orientation of successive European Semesters.
The 2011 AGS had set three overarching priorities for the EU: 'rigorous fiscal consolidation for enhancing macroeconomic stability', 'labour market reforms for higher employment' and 'growth enhancing measures'. The 2012 AGS replaced these with a more socially balanced set of priorities, including 'tackling unemployment and the social consequences of the crisis', which were reaffirmed in 2013 and 2014.
As EU institutions came under increasing pressure to tackle the deepening economic and employment crisis, the Commission and particularly DG EMPL responded by launching a succession of socially orientated 'pacts' and 'packages'. Foremost among these were the Employment Package (April 2012), the 'Compact for Growth and Jobs' (June 2012), the Youth Employment Package (December 2012) and the Social Investment Package (February 2013). Alongside specific policy measures (notably the Youth Guarantee) and EU recommendations (on child poverty and active inclusion), a key feature of these pacts and packages was that implementation of the proposed measures should be reported on, monitored and reviewed within the Semester.
As the Semester's policy priorities, expressed through the AGS and other Commission initiatives, shifted towards a more socially balanced stance, so too did the CSRs, whose social orientation expanded steadily between 2012 and 2014. For reasons of space, we do not analyse the 2012 and 2013 CSRs here, but jump directly to those for 2014, the final year of the Barroso Commission. 7 In 2014, twelve member states received recommendations on poverty and social inclusion, of whom nine were urged to ensure the adequacy and coverage of their social assistance and unemployment benefits. Nineteen received recommendations on reform of their healthcare and/or pension systems, whose purpose according to the Commission was to ensure that they 'continue to be socially adequate' and 'to provide universal access to high-quality care', as well as to be cost-effective and financially sustainable (European Commission 2014a: 8-10). Five countries were specifically urged to safeguard the accessibility and quality of health and long-term care, while eleven were exhorted to promote the employability of older workers, e.g., by expanding lifelong learning. As in previous years, however, not all CSRs dealing with social and employment issues could be considered 'socially orientated', and a number of countries continued to receive recommendations to reform their wage determination systems to improve competitiveness (6), reduce employment protection for insiders (4) or strengthen job search requirements for unemployment benefits (1).
Such recommendations were largely outnumbered by CSRs urging member states to improve their education, training, and activation systems (all except the Netherlands); reduce early school leaving (8); improve skills and training (21); implement the youth guarantee (8); enhance educational outcomes and access to the labour market for disadvantaged groups (12); increase the availability, accessibility and quality of childcare (9) and public social services (6)  Other systematic analyses of the Semester's evolution largely confirm this trend towards a progressive expansion in the number and scope of socially orientated CSRs from 2011 to 2014. Thus, a report by the European Trade Union Institute found that the number of specific recommendations within the CSRs addressing social and employment policy issues rose steadily from 95 in 2011 to 140 in 2014, accounting in each year for 40-47 per cent of the total (Clauwaert 2015: 11). Bekker's (2015) more comprehensive content analysis of the 2013 CSRs, which includes education and health care, concluded that 67 of the 141 recommendations contained at least one item addressing employment or social policies, while 11 more requested recipients to adopt a 'growth friendly approach' to fiscal consolidation. Using an expansive definition of anti-poverty policies (including 'improving social transfers', 'accessibility to social services', 'improving the employability of at-risk-of-poverty' groups' and 'specific measures to alleviate poverty of vulnerable groups'), Gómez Urquijo (

Governance procedures
The substantive reorientation of the European Semester towards a more socially balanced policy stance from 2011 to 2014 was accompanied by a set of procedural developments, which have reinforced the role of social and employment policy actors in its governance.
The progressive opening up of the CSRs to social issues reflects the fact that the process of drafting them became increasingly collaborative within the Commission itself. The groundwork for the CSRs is prepared by Country Teams led by SECGEN, with bottom-up input from desk officers and support from horizontal policy units across a wide range of DGs, including those responsible for health, education, enterprise and justice, as well as the other 'core' DGs: ECFIN, EMPL and (until 2015) Taxation and Customs Union (TAXUD). The CSRs themselves are drafted by the Country Team leaders, drawing on 'intelligence, ideas, and some wording' supplied by the various DGs, which are then discussed by the Directors of the core DGs, before being approved by the College of Commissioners, which often makes final changes. Our interviewees argued that this process has become increasingly deliberative and 'evidence-based'. DG EMPL officials involved in the process underlined how this emphasis on evidence and argument could overturn any implicit hierarchy among the participating units. A clear case in point was the 2014 CSRs on implementation of the youth guarantee scheme, which were initially opposed by both DGs ECFIN and SECGEN, but where DG EMPL eventually succeeded in getting them through based on evidence from the Country Teams and Staff Working Documents (COM-COORD4; COM-EMPL2, 6, 8, 10; EPC4) [1].
The other major developments that reinforced the role of social and employment policy actors during the Barroso era were driven by the EPSCO Council and especially its advisory committees. The first of these changes was the 2011 SPC initiative, subsequently endorsed by EPSCO, to 'reinvigorate' the Social OMC (SPC 2011a). Member states were invited to prepare regular National Social Reports, to be submitted alongside the NRPs, while the SPC also took over responsibility for the production of an annual Social Europe Report, replacing the previous Joint Report with the Commission.
In addition to reinvigorating the Social OMC, the SPC established itself as a significant player in monitoring, reviewing and assessing national reforms within the European Semester, alongside EMCO, EPC and EFC. At the EPSCO Council's request, EMCO and the SPC developed a Europe 2020 Joint Assessment Framework (JAF) for monitoring the Employment Guidelines (European Commission-EMCO-SPC 2010), which since 2013 has been extended to healthcare (SPC 2013; Baeten and Vanhercke 2016: 14). The results of the JAF feed into the Employment and Social Protection Performance Monitors (EPM, SPPM) developed by the two committees. These Monitors include visual representations of member states' comparative performance against a battery of indicators, along with detailed country profiles summarizing key challenges or negative 'trends to watch', as well as good outcomes (EMCO 2016a; SPC 2012b). Both EMCO and the SPC are committed to using their Monitors as analytical tools to underpin multilateral surveillance and support member states in setting reform priorities, identifying good practices and stimulating mutual learning, while also feeding into the broader EU policy debate. They likewise serve as an important point of reference for discussions within the committees about amendments to the Commission's proposed CSRs (COM-EMPL3, 6; EMCO1, 3; SPC2, 3) [2].
This extended social and employment policy monitoring was accompanied by intensified multilateral surveillance and peer review within both committees. By all accounts, multilateral surveillance in EMCO has become progressively more critical, more focused, and more evidence-based (EMCO Chair 2012; EMCO1-5, 7; COM-COORD4) [3]. Whereas until 2012 SPC peer reviews occurred once a year, focused rather superficially on the NSRs, now multilateral surveillance activities are conducted throughout the year, with in-depth thematic and country reviews focused on member states' responses to past CSRs and negative 'trends to watch' in their social performance. SPC members were asked in 2012 if they were ready to make country surveillance 'less descriptive, more critical' and more analytical (SPC 2012a), which subsequent interviewees confirmed has in fact occurred (COM-EMPL6; SPC3, 4) [3]. On overlapping issues like employment and healthcare, SPC reviews of CSR implementation are conducted jointly with EMCO and (since 2013) with the Working Party on Public Health at Senior Level (WPPHSL) (Baeten and Vanhercke 2016: 13). As in EMCO, the written conclusions of these multilateral reviews are submitted to the Council and contribute to discussions on proposed CSRs.
Both EMCO and the SPC have thus been seeking to combine 'tougher' multilateral surveillance in employment and social policies with increased opportunities for deliberation and mutual learning. Commission officials as well as committee members see this mutual surveillance process as a 'game changer', making exchanges within the committees less 'cozy' and more incisive than in the past, while transforming 'the bilateral discussion on the CSRs between Member States and the Commission into a multilateral decision making process' (COM-COORD4; EMCO3, 5; SPC3) [4].
Such intensified monitoring, multilateral surveillance and peer review in turn has formed the basis for enhanced input by EMCO and the SPC into the adoption of the CSRs. In the 2011 Semester, the SPC and the EPSCO Council were largely excluded from the review and adoption of the CSRs. From 2012, however, the SPC began to acquire a more influential place in the process, drawing on the expertise gained through its monitoring, surveillance and review activities, as well as through explicit political challenges by the EPSCO Council to the jurisdiction of economic policy actors over social and employment issues. The 2012 Semester catalysed a vigorous pushback by member states against the Commission's increasingly prescriptive approach to the CSRs, along with its reluctance to deliberate over proposed amendments with national representatives (SPC 2012a; Cypriot Presidency 2012; EMCO Chair 2012; EMCO4, 5, 7; SPC5; Zeitlin and Vanhercke 2014: 43-4) [5].
The mounting pressure for reorientation of the Semester and revision of its governance procedures gained momentum and political salience through the EU-wide debate on the social dimension of Economic and Monetary Union initiated by the President of the European Council in 2013. EPSCO Ministers responded by calling for a strengthening of the governance role of the Social OMC, building on the employment guidelines, together with the monitoring instruments, peer reviews and multilateral surveillance procedures developed within EMCO and the SPC. They also demanded the development of mechanisms to facilitate 'more effective interaction' between the EPSCO and ECOFIN Councils, 'especially as regards the adoption and review of the implementation of country specific recommendations' (EPSCO 2013).
These conflicts resulted in a substantially revised procedural framework for the Semester, based on a clearer allocation of responsibilities and co-ordination in reviewing and amending the CSRs in areas of overlapping competences between the committees and Council formations involved (Irish Presidency 2012; Zeitlin and Vanhercke 2014: 46-9). Underlying this framework were new decision-making arrangements within as well as between the participating committees. In each case, amendments to the CSRs were supported by 'reinforced' qualified majority voting (rQMV) to test the support for changes to the Commission's proposals. 8 As explained above, multilateral surveillance within the committees has likewise become crucial to the review and amendment of the CSRs. Since under the 'comply or explain' rules of the European Semester, the Council is expected to provide a written explanation of its reasons for modifying the Commission's recommendations, both committees also refer explicitly to these multilateral reviews in the formal reports justifying their amendments (EMCO2-6; COM-EMPL7; COM-COORD4) [6].
The revised procedural arrangements for reviewing and adopting the CSRs had a significant impact on the success rate of amendments. In 2012, an EMCO source estimated that proposed social and employment policy amendments had no more than a ten per cent chance of success (EMCO7). In 2013, interviewees estimated that one-third of all such proposed amendments to the CSRs were adopted without objection by the Commission, while the success rate for contested amendments had reached 50 per cent (EMCO4-5; SPC5). In 2013, there were fourteen contested amendments to the CSRs, of which ten concerned social and employment issues, compared to eleven in 2012 of which eight concerned social and employment issues. In 2014, contested revisions to the CSRs on social and employment issues continued at a similar rate, accounting for ten of eleven amendments successfully adopted over the Commission's opposition (own calculations from Council of the EU 2012. Most of these amendments concerned better contextualization of individual CSRs in relation to national challenges and reform measures. But even when amendments concerned only a single country, they could nonetheless involve policy issues of EU-wide significance, as in the case of a 2012 recommendation to Sweden on improving the employment participation of youth and vulnerable groups, which was revised to focus on enhancing the effectiveness of labour market policies rather than wage flexibility, while underlining the responsibility of the social partners for wage determination (Council of the EU 2012). It was also sometimes possible to obtain 'horizontal' amendments to multiple CSRs addressing broader issues, as in the case of pension reform, where the comply-or-explain text emphasized the need to allow Member States to choose among alternative paths to reaching the common objective of raising the effective retirement age (Council of the EU 2013). This message was endorsed by successive Council Presidencies, which concluded that 'Member States need some leeway to choose the implementation path that best suits their national conditions, in particular in areas which remain in [their] competence', urging the Commission accordingly to 'ensure that its CSR proposals are sufficiently precise as regards policy outcomes but not overly prescriptive as regards policy measures so as to leave sufficient space for … national ownership … .' (Cypriot Presidency 2012; Lithuanian Presidency 2013).

The Juncker Commission: streamlining and further socializing the Semester
In a speech to the European Parliament in October 2014, where Commission President-Elect Juncker enunciated his ambition for the EU to achieve a 'Social Triple A' rating, he stated that the Semester is not just an economic and financial process, but should necessarily take into account the social dimension, including the CSRs (Juncker 2015). The new Commission promptly introduced a series of significant innovations to the organization of the Semester. Building on plans initiated by its predecessor and responding to member state demands, the Juncker Commission sought to 'streamline' the Semester process by integrating the MIP In-Depth Reviews with the Staff Working Documents supporting the CSRs into a single Country Report, setting out its analysis of the main national reform challenges and measures taken to address them. These Country Reports were the subject of intensive bilateral discussions with member states both before and after publication. The Commission also released both the Country Reports and CSRs earlier in the Semester cycle, thereby leaving more time for review and debate within the Council advisory committees. But the most fundamental change was the Commission's decision to reduce drastically the number and scope of the CSRs, focusing on 'key priority issues of macro-economic and social relevance' identified as actionable and monitorable within a twelve--eighteen month timescale. In most cases, these streamlined CSRs focused more on challenges and outcomes than on specific policy measures, 'putt [ing] the focus on what to achieve instead of prescribing how to achieve it, in order to guide Member States while leaving the necessary room for manoeuvre'. Countries further advanced in the Excessive Deficit and Imbalance Procedures tended, however, to receive more numerous and sharply targeted recommendations (European Commission 2014b;2015b: 3;2016b: 4-5 The 'streamlining' of the Semester by the Juncker Commission was accompanied by a further socialization of both its substantive policy content and its governance procedures, which became increasingly visible in the 2016 cycle.

Policy orientations
While 'tackling unemployment and the social consequences of the crisis' no longer featured explicitly among the overarching priorities proposed by the Commission, the 2015 AGS emphasized the need for measures aimed at 'improving the dynamics in labour markets and tackling the high level of unemployment'; 'pension reforms … to ensure both sustainability and adequacy'; and 'modernising social protection systems' to 'be efficient and adequate at all stages of a person's life', through better targeting, 'affordable quality childcare and education, prevention of early school leaving, training and job assistance, housing support and accessible health care' (European Commission 2014b: 10-13).
Inclusive social and employment policy objectives figured even more prominently in the 2016 AGS. Central to its policy orientations was a renewed focus on social investment, including not only education and training systems, but also healthcare, childcare and housing; and flexicurity, including measures to foster transitions to permanent contracts, as well as 'well-designed income support systems' and 'policies to … equip job seekers with the right skills and better match them with vacancies' (European Commission 2015c: 4-5, 7-8, 11-12). Compared to the EU's earlier common principles on flexicurity, which were widely criticized for neglecting the security dimension, Bekker (2017) argues that the revised version of this concept has become more socially orientated, acknowledging the need to reduce flexibility in some areas. Another significant development concerned the future of the Europe 2020 Strategy, where the Commission affirmed its commitment 'to make the best use of the existing strategy and its tools'including the headline targets -'by improving its implementation and monitoring in the context of the European Semester'. Hence member states were requested in preparing their NRPs 'to make sure that the Europe 2020 Strategy continues to play a prominent role' (European Commission 2015b: 5-6).
In 2015, the number of CSRs on social and employment issues fell substantially compared to 2014. Using the Commission's classification scheme, the drop was particularly evident in the fields of poverty and social exclusion (from 12 to 6 member states), education and skills (from 25 to 13) and health and long-term care (from 16 to 11 and 7 to 2 respectively) (European Commission 2014a: Annex 1; European Commission 2015a: Annex 1). This development was a direct consequence of the Juncker Commission's decision to reduce the total number of recommendations and focus their content more sharply, while cutting down their length and scope. Considered in relative rather than absolute terms, however, the proportion of social CSRs did not decline (Clauwaert 2015: 10-14). Looking beneath the surface, a striking feature of the 2015 CSRs, as Commission interviewees emphasized, was their more integrated character, whereby social objectives were 'mainstreamed' into recommendations formally focused on other issues, such as labour market policy, education and taxation (COM-EMPL2-3; COM-COORD1-2). Thus, while only six CSRs were coded by the Commission as addressing poverty and social inclusion, a further six member states received recommendations to improve the educational participation and achievement of disadvantaged groups, with particular emphasis on Roma inclusion. Four countries also received recommendations to improve the adequacy, quality and accessibility of their pension or healthcare systems (European Commission 2015a, Annex 1; cf. Gómez Urquijo 2017: 7). 9 In 2016, despite a further reduction in the number of CSRs (to 89 from 102 in 2015), the number of discrete employment and social recommendations contained within them remained nearly stable (114 vs. 118), resulting in an increased relative 'social' prominence. On the employment side, the most frequent recommendations received by member states concerned skills, education and training (16) and activation and employment services (14). In other policy areas, like taxes and undeclared work, youth employment and wage determination, the number of CSRs fell sharply, reflecting perceived progress in these domains. On the social side, eleven countries received recommendations on poverty and social inclusion according to the Commission's classification. Only two CSRs explicitly mentioned poverty reduction, despite limited progress towards the Europe 2020 target and increases in the risk and depth of poverty in more than one-third of member states. But as the SPC observed, 'poverty reduction can also be achieved through policy recommendations on improving the effectiveness of social protection systems', on which fourteen member states received CSRs, including the coverage of social assistance schemes and the adequacy of benefits, as well as access to and quality of social services. While the number of countries receiving CSRs on health and long-term care remained stable, six received recommendations explicitly addressing quality and accessibility challenges, more than in previous years. The number of member states receiving recommendations on pensions, extending working lives and employability of older workers fell from fifteen in 2015 to ten in 2016, reflecting the reforms carried out in this domain, though few CSRs explicitly addressed adequacy and sustainability issues. As in previous years, the vast majority of CSRs specifically focused on poverty and social exclusion were directed to Eastern and Southern countries where the absolute problems were perceived to be most urgent, while social inclusion concerns in Northern and Western countries were more likely to be expressed through recommendations to enhance employment and educational participation of disadvantaged groups, facilitate transitions from precarious to standard forms of employment and improve the tax and benefit position of low-wage earners (European Commission 2016b; EMCO 2016a; SPC 2016). Crespy and Vanheuverzwijn (2016: 77-8) conclude that the proportion of 2016 CSRs promoting social investment rather than retrenchment policy measures rose to 64 per cent, the highest rate since the Semester's inception.

Governance procedures
Already under the Barroso Commission, the production of the CSRs had become an increasingly collaborative process, in which DG EMPL played a key role in both the teams preparing the country analyses and the Core Group responsible for drafting the recommendations themselves. Beginning in 2015, this process was carried a step further, as DG EMPL was charged with drafting the chapter on social and employment policy in the new Country Reports, which incorporated the MIP In-Depth Reviews previously written by DG ECFIN alone. DG EMPL also played an even more prominent part within the Core Group itself, as one of three line DGs working under SECGEN's lead. In so doing, DG EMPL could draw on substantial in-house policy and country expertise built up over the preceding years, as well as on key units dealing with Labour Market Policy and Training and Skills transferred by the Juncker Commission from DGs ECFIN and Education and Culture (EAC) respectively. In healthcare, DG Health and Food Safety (SANTE) contributed both to the Country Reports and CSRs, drawing on detailed country knowledge developed in recent years along with an indicator-based analytical tool for assessing the performance of national systems (COM COORD1; COM-EMPL2-4; COM-ECFIN1-2; COM-OTHER1-2).
A key objective behind the streamlining of the Semester, including both its revised timetable and the refocusing of the CSRs, was to increase national 'ownership' of recommended reforms and thereby improve their implementation (European Commission 2015a: 3, 10). Our interviewees and EU committee documents together confirm that these changes have been broadly welcomed by the member states, especially the opportunity to discuss and sometimes correct the Commission's analysis of national reform challenges and implementation progress in both bilateral and multilateral settings. There was broad agreement that as a result, the quality and accuracy of the analysis in the Commission's Country Reports was much higher than in previous years. National representatives also largely approved of the sharper focus of the CSRs on a reduced number of key priorities, together with the Commission's greater emphasis on outcomes rather than specific policy measures in responding to national challenges, though some recommendations (e.g., on pensions) were still considered overly prescriptive (SPC2; EMCO1; EPC1; COM-EMPL3; COM-ECFIN1; COM-COORD3; EMCO 2016b; SPC 2016) [8].
Streamlining appears to have improved the relationship between the EPSCO and ECOFIN Councils, as well as between the Commission and the member states. The revised Semester timetable gave EMCO and the SPC more time to discuss and review both the Commission's Country Reports and the CSRs, enriching the process of multilateral surveillance and contributing to the formation of internal consensus on proposed reforms. It also allowed for enhanced co-ordination of views across national delegations, which had previously been a major source of tension between social and economic policy actors. Compared to earlier years, there was better co-ordination between the EPSCO and ECOFIN advisory committees, ensuring 'a coordinated and balanced preparation of the CSRs' adoption by the Council', according  But even where EU economic policy actors formally retain a leading role, they appear to have internalized some of the objectives and concerns championed by their social interlocutors. Thus DG ECFIN's 'Horizontal Assessment Framework for Age-Related Fiscal Sustainability Challenges' included indicators for quality and accessibility of healthcare in 2016 for the first time, leading directly to the multiplication of CSRs on these issues (European Commission 2016a). And the Eurogroup itself has begun to discuss how pension reforms can be made 'more acceptable, socially viable and lasting' through flanking policies aimed at extending working lives such as lifelong learning, workplace adaptations and flexible working arrangements for older employees (European Commission 2015d).
Given the reduced number of CSRs, member states managed to obtain in 2015 a similar proportion as in previous years of contested amendments on social and employment issues, which accounted for six of nine successful amendments opposed by the Commission (own calculations from Council of the EU 2015a). In two of these cases (Finland and Italy), the CSRs were amended to recognize explicitly the autonomous role of the social partners in wage bargaining. In 2016, the improved accuracy of the CSRs resulting from intensified dialogue with member states around the Country Reports appear to have reduced demand for amendments, while the Commission also seems to have been more willing than in previous years to accept revisions proposed by EMCO and the SPC based on evidence from their multilateral surveillance reviews. The only contested social amendment adopted over the Commission's objections was to the euro area CSRs, where the EPSCO Council successfully insisted that national reforms should promote 'adequate and sustainable social protection systems that contribute effectively throughout the lifecycle both to social inclusion and labor market integration', rather than focusing more narrowly on work incentives and targeted support for those in need (Council of the EU 2015b; European Parliament 2016; Bekker 2017).

Conclusions
Between 2011 and 2016, a partial but progressive socialization of the Semester occurred. This shift was visible at the level of substantive policy orientations, in terms of a growing emphasis on social objectives in the AGS and especially the CSRs. The latter expanded considerably from year to year in social scope and ambition, placing increasing stress on the need for member states to ensure the adequacy, accessibility, and effectiveness of their social security, pension and healthcare systems; to combat poverty and social exclusion on a variety of dimensions; and to improve their education, training, childcare, employment and social services, especially for vulnerable groupseven if the growing volume and coverage of these social CSRs was still counterbalanced by recommendations to many countries on fiscal consolidation and competitiveness.
The socialization of the Semester was equally visible at the level of governance procedures, in terms of an enhanced role for EU social and employment policy actors in monitoring, reviewing and amending the CSRs. Jurisdictional struggles continued with EU economic policy actors about overlapping issues, especially those linked to the SGP and MIP. But revisions to the Semester's procedural framework, coupled with the detailed knowledge of the social and employment situation in the member states built up through intensified monitoring, mutual surveillance, and peer review, made it possible for EMCO and the SPC, as well as for DG EMPL, to feed their views into the process at key moments. These developments also enabled the EPSCO committees to push through amendments of the Commission's draft CSRs on both larger and smaller issues, despite the high rQMV threshold.
Member state representatives were able to push back against what they perceived as 'over-prescriptive', 'one-size-fits-all' recommendations from the Commission, which sought to lay down not only reform objectives, but also the specific way of reaching them, without taking sufficient account of national contexts and competences. By demonstrating their ability to amend the draft CSRs, these committees compelled the Commission to engage more deliberatively with member states in both multilateral and bilateral forums. Responding to persistent criticisms of its limited social dimension and weak national ownership, the incoming Juncker Commission strategically revamped the Semester process, 'by giving more time for dialogue with the Member States and mainstreaming social considerations alongside economic goals', as the 2017 AGS (European Commission 2016c: 4) proudly proclaims, leading to a further socialization of both its substantive policy content and governance procedures. The governance of the Semester has likewise become less hierarchical and more interactive, while the CSRs, especially in the social and employment field, have become less uniform, less prescriptive, and better adapted to national circumstances. Thus, since its inception in 2011, the Semester has slowly but steadily evolved in a more socially balanced direction in relation to each of the three critical empirical claims discussed in the introduction.
This progressive socialization of the Semester can be interpreted as a response by the Commission and other EU institutions to external developments, notably rising social and political discontent among European citizens with austerity policies, demonstrated by falling levels of public support for the EU and electoral defeats for incumbent governments in many countries. But it should also be understood as the product of strategic agency, reflexive learning and creative adaptation by EU social and employment actors of their own organization and practices in response to the new institutional conditions of the Semester. 10 This process began with the employment policy actors within EMCO and DG EMPL, who were already familiar with the challenges of debating their positions with the economic policy actors under the Integrated Guidelines of the Lisbon Strategy and took the first steps to strengthen their intelligence-gathering and analytical capacities through the development of new monitoring instruments, the intensification of multilateral surveillance, and the introduction of rQMV on amendments to the Commission's draft CSRs. Their example was quickly emulated by the social policy actors within the SPC and DG EMPL, who not only followed suit in developing new instruments to monitor and analyse the Semester's social dimension, deepening multilateral surveillance of CSR implementation and adopting rQMV for amending the CSRs, but also proactively used the institutional acquis and flexibility of the Social OMC to introduce new social reporting initiatives at both national and EU level and to foster mutual learning on promising approaches to tackling common policy challenges beyond those flagged in the CSRs. These innovations in turn helped EU social and employment policy actors not only to gain ground in 'evidence-based' negotiations over the CSRs, but also to revise the rules of the Semester game and influence the positions of their economic policy interlocutors on some key issues. 'Puzzling' and 'powering' (Heclo 1974) thus went hand-in-hand in socializing the European Semester.
In understanding how such apparently weaker actors can advance their substantive goals and decision-making influence despite the institutional asymmetries and structural power imbalances underlined by the Semester's social critics, 'actor-centred constructivism' and the 'usages of Europe' approach provide valuable conceptual resources. Complex governance processes like the European Semester, as these approaches emphasize, do not narrowly predetermine the outcome of political struggles, but typically offer multiple opportunities and resources for strategic agency by contending groups of actors, whose effective exploitation depends on the latter's ability and willingness to identify and act upon them. By seizing such opportunities and creatively adapting their own organization and practices, these actors may be able to reshape pre-existing power balances and procedural rules even where the overarching institutional framework remains formally unchanged. The socialization of the European Semester demonstrates in turn the power of these theoretical approaches to illuminate the evolution and dynamics of EU governance processes, along with the usage of such processes by national actors to which they have mainly been applied in previous research.
An outstanding empirical issue remains the domestic implementation and causal influence of the European Semester on national social and employment policies. This issue, as observed at the outset, requires a different research approach, based on comparative process tracing of national policy-making, similar to that developed in empirical studies of the OMC (Barcevičius et al. 2014). Here, too, 'actor-centred constructivism' and the 'usages of Europe' approach are of great potential value, by emphasizing the scope for creative appropriation of EU targets and recommendations, as initial research like that of Eihmanis (2017) on 'cherry-picking external constraints' and the implementation of social CSRs in Latvia illustrates. In the absence of more such empirical studies, however, scholars should not assume the existence of a direct relationship between the formal legal basis of the CSRs and their national influence, as much of the critical literature on the social dimension of the Semester discussed earlier appears to do, especially since no sanctions have yet been imposed on member states for non-compliance with either the MIP or the SGP. 11  3. Crespy and Verheuverzwijn suggest that this trend could stem from higher implementation rates of retrenchment-orientated recommendations, but do not provide supporting evidence for this interpretation. 4. References to this Appendix are indicated in the body of the text by bold numbers in square brackets. 5. AGSs, proposed CSRs, and CSRs as adopted by the Council are available for 2011-2014 at http://ec.europa.eu/europe2020/making-it-happen/index_en. htm. Countries covered by an Economic Adjustment Programme or Memorandum of Understanding do not receive CSRs under the European Semester. 6. For a country key to this and subsequent CSR analyses, see Online Appendix Section 2. Figures in the text refer to numbers of member states receiving recommendations on a particular topic, not to numbers of CSRs or recommended measures. The number of member states receiving CSRs varies from year to year, as countries subject to an Economic Adjustment Programme are excluded. 7. For a discussion of the 2012-2013 CSRs, along with a country key to socially orientated recommendations referred to in the text, see Online Appendix, Section 2. 8. Under the Lisbon Treaty, when the Council takes a decision not based on a Commission proposal, this must be supported by a 'reinforced' qualified majority comprising 72 per cent of member states accounting for 65 per cent of the Union's population (http://eur-lex.europa.eu/summary/glossary/ reinforced_qualified_majority.html). 9. For an analysis of 'hidden' social recommendations in the 2015 CSRs, see Online Appendix Section 3. 10. For reasons of space, we do not discuss in this contribution EU social NGOs and social partner organisations, who played a limited role in the evolution and dynamics of the European Semester during this period. For extended analyses, see Zeitlin and Vanhercke (2014: 54-5); Vanhercke et al. (2015: 15-17); Sabato et al. (2017). 11. Apart from a small fine levied on Spain for fraudulent fiscal reporting by one of its autonomous regions, exposed by the national statistical office itself (Savage and Howarth 2017).