Governing Europe ’ s Recovery and Resilience Facility: Between Discipline and Discretion

The EU ’ s Recovery and Resilience Facility (RRF) represents a bold integrationist step in European economic governance. Besides the size of the fiscal envelope, the novelty also lies in the new governance. Member states prepare integrated investment and reform plans and need to fulfil milestones and targets to access funding. This article assesses the balance of power in negotiating the plans and the effect on domestic policymaking. Based on five case studies, we show that the RRF has enhanced the steering capacity of the European Commission on reforms and investments, while member states remain ultimately in charge of the plans. Second, we argue that, while the RRF enhances the efficiency of the policymaking process and allows the fast-forwarding of reforms, it has also led to a contractualization of the relationship with the EU and a centralization of decision-making processes within member states. This latter aspect may hamper ownership and legitimacy in policy implementation.

known as frugals, have been mostly opposed to common borrowing, but have shown themselves open to the idea of further strengthening the Union's oversight of member states' commitment to structural reforms (Verdun 2022). Over the years, this debate has brought to the table a series of policy proposals, ranging from the Competitiveness and Convergence Instrument (European Commission 2013) to the Reform Support Programme (European Commission 2018) and the Budgetary Instrument for Competitiveness and Convergence (Eurogroup 2020). In each of these, the search has been to balance solidarity with principles of (financial) solidity. The RRF embodies both. It has been hailed by some as a historic turning point in European solidarity that allows much-needed investment for the recovery via common borrowing (Macron 2020), and reluctantly defended by others as a disciplining device for much-needed structural reform to ensure member states can stand on their own two feet if another crisis comes (Rutte 2020).
A key innovative feature of the RRF lies in its governance, which makes access to grants and loans conditional on the presentation of forward-looking reform and investment plans aimed at enhancing the Union's economic, social and territorial cohesion and convergence, and by supporting the green and digital transitions. These plans should be in line with the European Semester's country-specific recommendations (CSRs). The Commission then assesses the relevance and approves the plan based on the degree of alignment with the subset of challenges identified in the CSRs. The RRF further introduces a new performance-based approach, which links the disbursement of the financial support to the compliance with an 'operational agreement', signed between the Commission and the national government. This agreement reflects a Council Implementing Decision (CID) on the plans and defines the specific qualitative milestones and quantitative targets to be achieved within a fixed timeframe. With this governance, the RRF de facto changes the status of the Semester's recommendations as well as the monitoring role of the Commission within European economic governance. A positive way of interpreting this change is that member states are now rewarded for their actions and can receive fiscal resources to achieve their goals. However, the opposite interpretation is equally valid. Member states have given an external body the possibility of withholding funds in the event that they fail to deliver as proposed in EU recommendations.
Against this backdrop, the purpose of this article is to investigate the effect (if any) of the Recovery and Resilience Facility on the steering capacity of the Commission on CSR implementation in member states. To this end, we examine the drafting phase and approval of the national Recovery and Resilience Plans (RRPs), illuminating the interplay between the Commission and national governments. A second objective is to zoom in on the national governance of the RRPs and explorein a preliminary fashionwhether and how the introduction of the plans leads to a reconfiguration of the national decision-making process based on the performance logic of the RRF.
The article proceeds as follows. In the next section, we set out the theoretical angles from which we approach these questions and present our hypotheses. The third section sets out the case studies and the methodology. There are then three sections with empirical discussions on the process of drafting and negotiating the RRPs and the consequences of this particular governance setup and the style of governing involved. The last section concludes.
Situating the research questions in the debate on EU economic governance The European Semester, introduced after the euro crisis, has driven the nature of the EU polity away from classic integration by law, towards a more hybrid model, where the stringency, or effects, of policy recommendations cannot simply be discerned from its legal status and changes over time. This has resulted in variegated readings and appreciations of the nature of these instruments, in terms first of legitimacy and then of effectiveness. A heuristic concept to study their difference is in the appreciation of hierarchy in governance. Hierarchy here is meant as a governing style based on rule-setting, steering from the top in the most predictable ways, minimizing negotiations and exceptions to the rules (Dunlop and Radaelli 2016). This section will set out the various interpretations of hierarchy in economic governance over time and what these may imply for the RRF.
The EU's turn towards New Modes of Governance (NMG), such as the Open Method of Coordination, in the early 2000s has made hierarchy an important subject of study among integration scholars. The supposedly 'soft' character of nonbinding coordination raised suspicion as to whether they would need the threat of 'harder' legislative or executive instruments to be effective (Héritier and Lehmkuhl 2008). Opposed to this hardening-for-effectiveness approach stands the school of experimentalist governance scholars, who see the non-hierarchical nature of governance as an important condition for its success. This school stresses the importance of learning from diversity, regional autonomy in implementation of centralized policy goals and a governance design that is conducive to iterative policymaking based on reflexivity (Sabel and Zeitlin 2007). In practice, European governance has long been portrayed as a combination of harder and softer elements (Tömmel and Verdun 2009).
The transformation of the OMC into the Semester has widely been interpreted as a move towards more hierarchical steering skewing power away from domestic democratic fora towards executive and technocratic power through governing by rules and numbers (Crum 2018;Dawson 2015;Schmidt 2020). The legitimacy concerns about an overly meddlesome Commission were especially alarming in the first years of the Semester and should also be considered against the austerity paradigm of that time. In response to the criticisms of its hierarchical steering, the Commission in later years adopted a more flexible approach to the Semester, focused on national ownership. This approach also allowed for understanding the nature of economic governance more as the Commission having cognitive and agenda-setting influence on national reforms rather than imposing them and being more or less successful depending on domestic political circumstances (Bokhorst 2022a). The search for ownership shifted the debate away from the starker legitimacy concerns towards critics who argued that EU coordination mechanisms have in fact been too soft to be effective. Especially among economists, the dominant image emerged of the Semester's CSRs failing to gain much traction in Europe's capitals (for an overview see: D'Erman and Verdun 2021); proposals followed that implied a much stronger oversight role of the Commission in national reforms (Bénassy-Quéré et al. 2018).
Against these different approaches, the introduction of the RRF can, at least in theory, be interpreted as a potential reinforcement of the hierarchical nature of EU economic governance. With the RRF, the Commission can ex-ante push member states to be ambitious and reject their plans in the event these insufficiently address the CSRs. The plans are indeed written in tight consultation and negotiation with the Commission, therefore allowing the Commission to discuss and potentially influence policy goals, deliverables and their level of ambition before they are formally adopted by governments and parliaments. In addition to this, the RRF performance-based approach, which links payments to milestones and target achievements, leads to an increase in the contractualization of the relationship between member states and the Commission. The plans are not a statement of intent by the government, but a commitment to deliver results by the state. All parts of the policy machinery, including parliaments, regions and social partners, need to deliver in order for the milestones and targets to be achieved. A failure in the fulfilment of such objectives may lead to a reduction or a delay in payments.
While the RRF has the potential for more hierarchy, there are also good reasons not to assume a full move to hierarchical steering. Despite the need to align reforms and investments to the CSRs, the RRF milestones and targets are not 'imposed' by the Commission to the national governments but are proposed by the latter. In terms of governing logic, it should also be noted that Europe has gone through a period of reflection and learning, and elites haveto varying degreeschanged their understanding of how to respond to crises (Ladi and Tsarouhas 2020). Evaluations, especially of the member states under the hierarchical Memoranda of Understanding (MoUs), have stressed the need for national ownership to ensure that reforms produce lasting effects (Buti 2021;Manasse and Katsikas 2018). Accordingly, we may expect a balancing act between EU discipline and (national) discretion in terms of the steering capacity, rather than attempts for direct imposition from the European Commission.
The RRF governance structure does not just shape the relation between the EU and member states but may also affect the structuring of the policymaking process of member states internally. For the Semester, the norms, targets, deadlines and reporting requirements can also be actively used by governments as effective internal discipline, to ensure accountability and alluding to the responsibility function of government (Ferrera 2016). Catherine Moury et al. (2021) find that administrations governing under the MoUs also actively made use of the constraint imposed on them to pursue unpopular policies that they deemed necessary. Such a strategic usage of hierarchical constraint, known as a vincolo esterno, has been commonly associated with the structural reforms that were pushed through to meet accession criteria for euro adoption in Italy (Ferrera and Gualmini 2004).
Here it should also be stressed that there are marked differences from the MoUs, as the RRF plans typically involve an investment logic, rather than the tough austerity found during the euro crisis.
The new performance-based approach of the RRF, characterized by the rigid ex-ante definition of milestones and targets with low flexibility for ex-post adjustments, may potentially reduce political space for other actors involved in policymaking and thus reinforce centralization dynamics. Indeed, the first empirical studies on the RRF have noted the difficulty of involvement of social partners and regions in formulating the plans as consultations remained insufficient (Carrosio et al. 2022;Schelkle et al. 2022;Vanhercke and Verdun 2021).
While, in theory, the RRF governance may allow for hierarchy within policymaking processes, it should be noted that evaluations of performance-based management in the past have mostly warned against such developments. Eulalia Rubio (2022), for example, points out here that evaluations of performance financing for International Monetary Fund (IMF) programmes or development aid have mostly concluded that even if there is ownership by national governments, this may not imply legislative buy-in or social acceptance. Others have primarily pointed at the overreliance on ex-ante knowledge in a world of uncertainty (Moynihan et al. 2011;Pollitt and Dan 2013). With performance-based financing, there is the risk that important elements of good policy that are hard to measure become inferior to measurable outcomes. Ex-ante targets must be set in a meaningful way, but due to inherent uncertainty and rapidly changing conditions in policymaking, this is no easy task. Modern investment practice therefore often relies on smaller milestones and regular joint review to allow for revision, in line with the experimentalist logic (Gilson et al. 2009). The extent to which these concerns played a role in the national decision-making process in the drafting phase of the RRF plans remains to be empirically investigated.

Case selection, data collection and methodological approach
To understand the impact of the RRF on the steering capacity of the Commission and on national decision-making processes, we focus on the analysis of five countries: Italy, Spain, Croatia, Germany and the Netherlands. The selected countries represent two samples of high and low beneficiaries of the RRF. Spain, Italy and Croatia received an envelope equal to 5.8% (€69.5 billion), 4.1% (€68.9 billion) and 12.2% (€6.3 billion), respectively, of the national GDP, while Germany and the Netherlands received 0.8% (€25.6 billion) and 0.7% (€5.9 billion), respectively, of their GDP. Clearly, the larger the financial envelope under the RRF, the higher also the number of measures included in the plans and the higher the number of milestones and targets to be complied with within the timeframe of the RRF. The Croatian plan includes 222 investments and reforms, the German plan 40. Against this backdrop, we expect the RRF to have a larger impact on national policymaking processes and we expect deeper scrutiny and steering by the Commission in Italy, Spain and Croatia. By contrast, Germany and the Netherlands, which entered the pandemic with fewer macroeconomic imbalances and received a much lower amount of funding, are test cases to check the generalizability of our findings. The selected countries represent also a mixed sample of beneficiaries of the Cohesion policy funds, with Spain and Croatia being net beneficiaries, Germany and the Netherlands and Italy being net contributors. Finally, the five selected countries show different levels of public debt, with Italy, Spain and Croatia showing high debt constraints, while Germany and the Netherlands have low levels of indebtedness ( Table 1 in  Methodologically, this article builds on a multimethod approach of qualitative content analysis and data-driven elite interviews. As a first step, we carried out a content analysis of the RRPs. First, we systematically collected information on the reforms and investments. The main documents used for the identification and classification of the measures included the annexes to the Council Implementing Decisions on the Approval of the Assessment of the Recovery and Resilience Plans and the Commission Staff Working Document, which accompany the proposals for the Council implementing decisions. The data collected from the Commission and Council documents were then completed with information directly retrieved from the RRPs submitted by member states. As a second step, to understand which factors explain the (difference in) national choices and to identify the main containing and driving factors in which policy choices are embedded, we carried out 21 interviews with national public officials in charge of the drafting of the recovery plans at the ministerial level (prime ministers' cabinets, ministry of finance, and/or social and labour affairs), and EU Commission representatives in the Directorate-General for Economic and Financial Affairs (DG ECFIN) and the Secretariat-General RRF Task Force (SECGEN RECOVER). The Commission is the central actor of focus here, as previous studies have already concluded that it is in the driving seat of the daily operation of the RRF, even if the Council eventually is the actor to sign off on all payments (Corti and Núñez Ferrer 2021). During the interviews, we identified issues of contention and insufficiency in implementation from the qualitative analysis that we could probe, to better bring out the phenomena of steering capacity and mechanisms such as pressure. The purpose of the interviews was to understand how the Commission used its enhanced possibilities for pressure for including or removing specific measures and to understand who was setting the terms in negotiations. Second, we used the interviews to try to understand the mechanisms of change and how the RRF contributed in practice to changing national decisionmaking and agenda-setting. A list of our interviews and reference codes to interviews (e.g. EC1 for Commission officials, HR1 or IT2 for member state officials) is provided in Annex 2 in the Supplementary Material.
From theory to practice: the preparation of the RRPs While the RRF regulation was formally adopted in February 2021, preparation of the national plans already started during the summer of 2020, immediately after the Council agreement on the NGEU package. Contrary to the traditional European Structural and Investment (ESI) funds, the RRF introduces a new central management system that no longer relies on national and regional managing and audit authorities to manage the programmes, implement them by selecting projects, control and assess them. Instead, the ministries of finance usually have either overall responsibility for the plan, as in the case of Germany and the Netherlands, or share responsibility together with the prime minister's office or other ad hoc central coordination bodies, as in the case of Italy, Spain and Croatia. To ensure effective monitoring and implementation of the RRPs, including the envisaged timetable, milestones and targets, and the related indicators, member states either used established governance/control structures or set up new ones. Italy, for instance, set up a completely new governance structure with six new bodies, as did Spain and Croatia. Germany and the Netherlands set up coordinating bodies within already existing structures. Since the governance and control systems were not fully in place, Italy, Croatia and Spain, like other countries, included several milestones in their national plans, committing to set up the implementation and monitoring structures quickly. The need to set up new governance systems clearly took time and pushed the first meeting with the Commission to October 2020, even though the first deadline to submit a draft of the plans' headlines was September. Yet the countries with the systems already in place also started the meetings with the Commission after the initial September deadline. The Netherlands began much later with the negotiations, due to a prolonged government formation process and the political decision to hold off on detailing the plans until a formal government was in place.
From the preparation phase onwards, exchanges between the Commission and the countries were constant and progressively intensified the closer the April 2021 deadline to present the plans approached. In the case of Croatia and Spain, more than 85 meetings were held with the Commission. In the case of the Netherlands around 50. In August 2020, the Commission established a RECOVER task force under the Secretary General that was responsible for steering the RRF's implementation, for coordinating it with the European Semester and also responsible for involving the other policy DGs at all stages of the process as necessary. For this purpose, SECGEN RECOVER and DG ECFIN held weekly coordination meetings and regularly engaged with other DGs through 'country teams'. These country teams were built on existing European Semester teams, but also included specialists from all relevant DGs to provide the required expertise on the different policy areas covered by RRPs. Interactions with the member states were managed by 27 negotiating teams. These teams, led by a chief negotiator, were responsible for analysing (draft) RRPs, preparing assessments and implementing acts related to the plans and payment requests, and now are in charge of monitoring implementation progress. Finally, horizontal teams were set up to deal with green (including the Do No Significant Harm (DNSH) principle), digital, costing and milestone/target issues. All in all, once the Commission received the final versions of the plans to start its official task of assessing these based on the criteria, every page and policy action had long been discussed and negotiated, so that final assessment was in part a formality. The plans were then approved by the Council, where only few questions were raised and no changes were introduced (EC1).

Negotiating investments: the channelling role of the Commission
The negotiations with member states took place on two levels. Investments were mostly discussed at a technical level and earlier in the drafting process, while reforms were subjected to a more political debate later in the process and centred on alignment with the CSRs. The key subject of discussions on the investment side was related to compliance with the eligibility criteria established in the RRF regulations. Here the Commission acted as a kind of gatekeeper in ensuring that the coefficients for green and digital spending would be respected and the targets met, and to ensure compliance with the more difficult DNSH principle. Since the investments' projects could have been financed as of February 2020, member states were allowed to include projects under the RRF that had already been programmed. If we zoom in on social spending, Francesco Corti et al. (2022) notice that Germany used 80% of its social expenditure under the RRF to cover already budgeted projects that were planned to be implemented. The German plan was indeed designed on the basis of the 'corona fiscal package' adopted in June 2020 (DE1). Similarly, in the Netherlands, it was decided early on that only existing projects that had already been fully covered with national funds could now be replaced by RRF funding. As such, departments were initially even reluctant to join the RRP since the RRF comes with additional reporting requirements and the money had already been budgeted (NL2). By contrast, Italy and Spain used the largest share of the social expenditure -64% and 60%, respectivelyto finance new projects, which without the RRF would have likely not been implemented (Corti et al. 2022). Similarly, Croatia also included new measures that were not budgeted before and that with the RRF found a window of opportunity.
In its gatekeeper role, the Commission had to rely on what the regulation allowed them to do, which meant that as long as eligibility criteria were met, the Commission mostly did not interfere with national priorities for investments as set by the governments. Given the legal possibility to have backdated projects, there was little the Commission could do to force member states to also bring something new to the table. Even for future investments, member states often had good reason to propose projects that were already in the pipeline, given the fact that all payment requests have to be made by 2026 and that there was little time to develop completely new ideas. In this respect, it was considered logical that there were synergies between the RRF and other already planned domestic interventions (DE1). As observed by our interviewees, the investments were rarely a novelty: 'we were not in the position of inventing something new. Instead, we focus on something which we were already developing' (IT1).
This said, even though all countries included investments already in their pipeline and in some ministerial drawers, the key added value of the RRF consists in the fact that for some countries without the RRF such investments would have likely not been implemented, while in others the measures included in the RRP were already budgeted. This emerges in a clear way from the interviews. In Germany and the Netherlands, the RRF did not represent a breakthrough for the adoption of new initiatives. As confirmed by a German representative, 'the national budget was enough to cover the national needs, that is why the initiatives were (partially) left aside in the RRP' (DE2). By contrast, in Croatia, Italy and Spainthat is, in those countries with less fiscal margin to intervenethe RRF made the difference, providing the financial resources needed to adopt reforms and initiatives. As acknowledged by our interviews, the RRF was considered an important opportunity, a 'trigger' and a tool for the implementation of social reforms that would either not have been adopted because of a lack of adequate funding or would have been further delayed (HR2; IT2). Overall, 'the RRF was a push to do what had been postponed or delayed for certain reasons' (HR2) and can be used by domestic actors to speed things up (HR1; IT2; ES2).
At the same time, we should not imagine the negotiations between the Commission and the member states on the investment projects in the national RRF plans as a 'box-ticking exercise'. For the Commission, it was important to understand the rationale and strategy behind the investment choice rather than having simply a list of projects (IT3). Instead, the debate on the investments should be seen as a comprehensive back-and-forth process in which the ownership of the intervention was at the national level and the Commission guaranteed the compliance with the RRF criteria but also tried to help fine-tune and channel the national proposals and make milestones concrete.
According to our interviewees, the Commission did use persuasion to try and convince member states either to be more ambitious or to be realistic in terms of what could be achieved within the given timeframe. Both Spain and Croatia had initially introduced a vast number of milestones and investment projects, which were later reduced in consultation with the Commission, yet without decreasing the level of ambition nor ownership. Much of the debate in this case focused on technical details such as cost justification. This is important because member states need to make accurate predictions of what can be achieved within the estimated budget. Unlike with ESI funds, the ambition cannot be lowered later in the process for the same amount of money if costs turn out to be higher. In the event that costs turn out to be higher than anticipated to meet the pre-agreed ambition (even if this is due to inflation), the member state itself has to cover the difference.
Negotiating reforms: the steering role of the Commission Turning to reforms, here is where the political negotiations between the Commission and the member states were focused and where interaction resembled real bargaining. According to the RRF Regulation, the plans should contribute to effectively address all or a significant subset of challenges identified in the relevant CSRs of the 2019 and 2020 Semester. And whereas this link between funds and CSRs had already been present in the ESI funds (Ciffolilli et al. 2018), the latter did not include the contractualized commitment and operationalization to detail that is now found in the RRPs. In terms of negotiations, the Commission's interpretation of its task was to try simply to get the maximum possible implementation of CSRs, knowing full well that it could push harder in some cases than in others. When entering the pandemic crisis, countries had different socioeconomic vulnerabilities, which were reflected not only in a larger envelope under the RRF but also in a higher number of CSRs to address. This also enhanced the leverage position of the Commission on these countries with the largest net benefit from the RRF. In this respect, both the Commission and the member states were aware of the countryspecific challenges to be addressed in the national plans, as well as the relative RRF financial evelopes to achieve these objectives.
The assessment of the plans did, however, not constitute a situation of 'two weights, two measures', as the Commission tried to be persuasive in all cases. While interviews showed that the Commission's steering capacity was lower in the case of both the Netherlands and Germany, the RRF still contributed to accelerating the adoption of reforms that would otherwise not have been adopted in such fashion. In the case of Germany, for instance, the Commission had long recommended removing barriers to public investment, especially at the local level, including personnel and financing and regulatory constraints, as well as a poorly digitized, complex and decentralized legal system for public procurement hampering private investments. As put by our interviewees, 'We would have not implemented this reform if there was not a pressure from the RRF' (DE2). Nonetheless, when reading the Commission assessment for Germany, it becomes clear that for many of the other CSRs, the RRF did not represent a breakthrough. Major areas of the CSRs, for instance on the sustainability of pensions, disincentives to work, tax on labour or regulated professions, are either not addressed or addressed only to a limited extent (European Commission 2021). For example, on long-term sustainability and adequacy of the pension system, the German plan does not go beyond the introduction of an information portal. For the Netherlands, one of the key contentious issues that has long been a topic of debate between the Commission and the Dutch, namely the fiscal incentives on mortgages, has not been addressed in the plans. For the Dutch government, the Commission was already very happy with the level of ambition on reforms to tackle aggressive taxation practices, which was seen as a priority. Here the Dutch deliver much more ambitiously than some of the others, such as Ireland and Luxembourg, where the plans are rather meagre in addressing this CSR (Bokhorst 2022b). Nonetheless, the Dutch have been eager to include a wide range of reforms, including labour market reforms, pension reform and administrative reform, in their RRP to prove that they are reform-minded. As such, for low-beneficiary countries the RRF might have enhanced the influence of the Commission, but it has not represented a breakthrough in terms of CSR implementation.
For the high-beneficiary countries, this picture looks quite different. For one of the member states, the ambition in negotiations was to go for the full set of CSRs, whereas the first version of the plan was seen as insufficiently ambitious in this respect (HR3). This was perceived similarly by officials of that member state: 'We knew that we would be under special monitoring because of the importance of the RRF for us. They [the Commission] were in a position to push us in the terms of the adequate level of ambition of certain reforms' (HR2). In other cases, there was also an element of political appreciation in the negotiations, as it might not be realistic to see all CSRs fully addressed as final outcome (IT3). Overall, on the side of the Commission there is clearly pride in what has been achieved in terms of CSR implementation in high-beneficiary countries. The RRF is depicted as 'gamechanger', thanks to which CSRs that stayed on paper for a long time are now finally implemented (HR3; IT3). This notwithstanding, even in high-beneficiary countries, not all the CSRs were translated into the content of the RRPs, and some countries included several reforms and initiatives that were non-related to the CSRs.
The negotiations with the Commission in high-beneficiary countries involved qualitative discussions on the overall level of ambition of reforms, but also on issues of policy design, reform direction, coherence and timing. On timing, milestones and targets were used to frontload the structural reforms, especially in those countries, such as Italy, Spain and Croatia, which included politically sensitive reforms in the plans, like pensions, labour market, but also justice and public administration, which require broad political support. On substance, for example in Croatia, the Commission used the RRF's requirement of ensuring territorial cohesion in persuading the Croatians to ensure that active labour market policies would reach the poorest regions. Interestingly, for both Italy and Croatia the Commission even received and commented on draft laws of projects and reforms already in the pipeline. Spain did not send draft laws to the Commission before publishing them, but the Commission participated in the public consultations held on initiatives that fell under the RRF. As such, the Commission wears multiple hats in the RRF: it is the one to push until plans are sufficiently ambitious to get access to the funds (although de jure the Council is the official body to decide), and it also engages in discussions over the content of reforms. As such, their role goes beyond the indirect and cognitive influence on member state reforms observed in the pre-COVID Semester (Bokhorst 2022a).
This does not mean that the level of ambition nor the content of a reform is unilaterally decided by the Commission. As stressed above, the approach is rather constructive and can be framed in terms of a dialogue between the Commission and national governments. Two examples are the pension reforms in Spain and Croatia. In its RRP, Spain proposed relinking pensions to the consumer price index on a permanent basis and dissociating initial pension levels from changes in life expectancy. At the same time, Spain took a number of measures to boost the effective retirement age that partly offset this further increase in expenditure. As such, the key reform goes against the challenge of fiscal sustainability, as overall pension expenditure is expected to increase further, despite ageing in Spain causing a very substantial dependency ratio. Interviewees would admit that indeed this was the most sensitive point of discussion but still the government brought convincing arguments to the table to justify the choice to reintroduce indexation (ES3, ES4). For Croatia, the government convinced the Commission that linking pension age to life expectancy now would lead to problems with social partners, who had blocked such proposals just two years earlier. Instead, the Croatian government, in consultation with the Commission, searched for less rigorous measures that would still leave open the possibility for more comprehensive reform in the future.
To sum up, while the enhanced influence of the Commission can hardly be denied, we did not encounter elements that could clearly be identified as hierarchical steering. In the end, it was the member states who proposed lists of interventions in line with the CSRs, even when in close consultation with the Commission. Member states were aware of what was expected of them, and it should also be stressed that they had a strong drive to deliver themselves (ES2; IT1; ES2; DE1; HR2). Unlike the austerity context of the MoUs or the early days of the Semester, negotiations and consultations with the Commission were generally seen by member state officials as constructive, a dialogue about what constitutes good policy to ensure growth, while the role of the Commission was primarily described as 'helpful' and 'healthy' (ES1). While we use the word 'negotiations' to describe the preparation of the plans, the interaction was also iterative: proposals would be submitted, the Commission would comment and send them back, which would lead to new discussions and adaptations. The need to ensure ex-ante operationalization on details also led to learning, both on the side of member state officials and on the Commission, where officials admitted that their knowledge of domestic policies and policy challenges is now much deeper.

From negotiations with Brussels to internal steering
Turning to the impact of the RRF on the national policymaking process, the RRF attaches additional leverage for administrations at the domestic level, notably in the high-beneficiary countries. In this respect, our interviewees in Croatia, Italy and Spain considered the RRF as an important opportunity, a 'trigger' and a tool for the implementation of many reforms that were already in the pipeline of national governments, but that either would not have been adopted because of a lack of adequate funding or would have been further delayed because of political obstacles. Overall, the RRF has been used by domestic actors, notably national governments, to speed things up (HR1; IT2; ES2). According to a Croatian official, 'the RRF was a push to do what had been postponed or delayed for certain reasons' (HR2). Indeed, the requirement to set detailed milestones and targets was used not only by the Commission to ask governments for more details and ambition, but also by the governments to force and accelerate national decision-making processes. Overall, administrations have been eager to include a wide range of reforms in their plans and thus allow for external pressure to deliver. This is particularly true in those countries, such as Italy, Spain and Croatia, where the financial envelope is significant and where the risk is higher of losing out on EU funds due to noncompliance with milestones and targets.
The performance-based approach contributed to developing mechanisms of internal disciplining and efficiency. As put by one of our interviewees, 'When we say to somebody, you have a deadline, you have a reform, you have a law, you have to implement it, nobody would object. We decided to ask either all the amount or nothing, and when you have that kind of positive pressure, nobody wants to be pointed out for being last' (HR2). As put by an Italian official, commenting on the new performance-based approach, 'we are able to tell them to realize the selected project because we are committed to it. This performance-based approach is something that is helping us to overcome resistance' (IT2).
Overall, senior-level bureaucrats in Spain, Croatia and Italy positively welcomed the contractualization that is derived from the new performance-based approach as a positive cultural shift in public policymaking. They stress that despite the heavy administrative burden of the detailed operationalization, this new approach forced the administrations to identify expected outputs and outcomes and introduced an efficiency dimension in public policymaking (ES3; IT2). At the same time, they stress that the contractualization of the performance-based approach reduces leeway for deviation and increases common responsibility to meet the agreed objectives within the agreed timeline. By contrast, in countries such as Germany where senior public officials do not feel responsible or pressed to meet the targets and milestones included in the plan, 'The main problem of the implementation system is that there is no sanctioning system if the other ministries don't implement the measures they promise to implement. They know we have the financial resources to cover the costs. The problem in this case is for the Ministry of Finance that will have to report the unmet milestones and targets' (DE2). In the Netherlands, the performancebased approach is not expected to alter domestic policymaking seriously, as it can be easily integrated into existing administrative practices, but the Dutch are very positive about the use of performance-based financing in EU funding (NL2).
The ex-ante formulation of expected goals is seen to enhance deliberation about the usefulness of policy instruments and gives clear metrics to evaluate success. Beyond increasing national ownership, the selection of reforms and investments based on expected output and outcomes pushed member states to 'think about reforms and investments in parallel and this is a positive element because it forces having a coherent approach' (IT2). In this sense, the new performance-based approach represents a 'big experiment' (IT2) not only since it forced countries to go faster, but also because it 'armed' reform proposals with specific financial and political resources to be 'used' by domestic political actors to build strategies for modernizing their economies (IT2; HR1; ES2; ES1).
Importantly, we note that the positive associations with a more contractualized nature of the relation in the implementation phase are primarily held by those officials directly involved in the formulation of the plans and in more senior positions in administrations. In line with our theoretical section, we may also point here to a number of risks and challenges that this centralization and contractualization may have when it comes to implementation.
First, the plans have been formulated under high time pressure and in a centralized manner. To speed up the processes, the milestones and targets were in some cases only quickly discussed within the line of ministries before being agreed upon and negotiated with the Commission by the ministries of finance, mostly with very limited consultation. As put by one of our interviewees, 'in mid-August 2020 I was asked to submit projects in four or five days. We had to rush in August. This was a first problem that you will find in the plan because the project had to be prepared in a rush, to prepare it in a month' (IT1). While there has been extensive deliberation over the existing projects, there has not been much space for new ones. Similarly, there have been worries about the milestones and targets becoming goals in themselves with governing overly focused on the attainment of quantitative and measurable output rather than a means to better policy outcomes. Some interviewees experienced 'box-ticking' practices already in the drafting phase. As put by an Italian official, 'they only asked for huge Excel files. But nobody asked you about the content of the project or an explanation of the budget' (IT1).
A second obstacle is linked to the high degree of ex-ante knowledge found in the plans that may not correspond to local needs and, in combination with the contractualized nature, moves away from iterative and reflexive elements of policymaking. In this sense the level of detail in milestones and targets is surprising. Some targets specify precisely how many thousands of civil servants will be trained under a specific programme and which percentage should succeed in the training. Or, in the case of bureaucratic simplification in Italy, their milestone is only achieved when all relevant secondary legislation is adopted by parliament and implemented, including delegated acts and ministerial decrees relating to 200 critical bureaucratic procedures. Targets also relate to the quantity of childcare centres needed per region, or the precise number of people to be hired in public employment centres.
The necessity to define a clear and transparent set of milestones and targets ex-ante and within a very constrained timeframe may materialize in a stiffening of the implementation process whereby little space is left for context-specific interpretation and adaptation. The established governing principle in economic governance of comply-or-explain (because explanations may matter) has been changed in the RRF implementation to comply-or-do not-get-paid, implying that explanations do not matter, only implementation does. In this sense, the governance does not necessarily combine well with diffusion of ownership throughout the state apparatus, whereas implementation crucially depends on the ability of regional and local actors and parliaments to deliver for milestones to be achieved. The Commission has some wiggle room in interpreting whether milestones and targets are 'satisfactorily fulfilled', but in principle they should hold member states to their promises as precisely as possible.
Conclusion: capturing the changing dynamics of the RRF The EU response to the COVID-19 crisis marked without a doubt a path-breaking innovation in the European economic integration process. While many commentators focused on the magnitude of the fiscal effort financed through commonly financed debt and the rediscovery of public investment as an engine of the postpandemic recovery, this article has zoomed in on the novelties introduced by the RRF in terms of governance and governing. By linking financial assistance to the presentation of long-term investment and reform plans in line with the Semester's CSRs, the RRF indeed puts meat on the bones of the latter and introduces a new positive conditionality in European economic governance. At the same time, placing the Commission in the driving seat of the new plans empowers it to decide on compliance of the plans with the requirements of the regulation. The RRF, in theory, has the potential to strengthen top-down hierarchical steering mechanisms by introducing a new comply-or-do-not-get-paid approach, based on the compliance with ex-ante defined, detailed milestones and targets. At the same time, the plans are not imposed by the Commission but rather discussed with member states, which also gives good reasons not to assume a full move to hierarchical steering.
What emerges from our empirical reconstruction is a balanced scenario. Far from introducing a new hierarchical approach in the European economic governance, the RRF is accompanied by a new understanding of the interaction between the European Commission and member states. Moving away from the ruling-by-numbers approach characterizing the post-Great Recession Semester, the RRF leaves national governments in the driving seat of their Recovery and Resilience Plans, with the Commission balancing between discipline and discretion. Clearly, such a balancing act largely depends on the financial envelope received by the member states under the RRF. As predicted, the higher the financial support, the higher the number of milestones and targets to comply with and the greater the pressure from the Commission in the drafting of the plans. Yet, such pressure should not be negatively characterized; quite the contrary, it resulted in a constructive policy dialogue between national governments and the European Commission.
We further showed that such pressure should be differently understood according to whether we are speaking about investments or reforms. In the case of the former, we should understand the role of the Commission as more of a gatekeeper, channelling investments to make sure they meet the criteria set in the Regulation. By contrast, on reforms, the Commission plays a more proactive role in ensuring realism and sometimes persuading and pushing countries to be more ambitious. In both cases, and especially for large beneficiary countries, the negotiations between the Commission and the member states were driven by the idea of preparing plans with a long-term growth strategy rather than merely complying with eligibility criteria. The role of the Commission has changed from that of an accountant, whoput in simplistic termscomes in once a year to check the annual accounts on the basis of opaque SGP metrics, to that of an investor, who is less concerned with daily operation and budget but wants to discuss the growth strategy and set targets in exchange for funds. The investor, at least in theory, is not interested in receipts (as is still the practice with ESI funds), but in results.
Given the primacy of the growth strategy, it is important to ensure both delivery and ownership across all layers. In this article we have pointed out that while the formulation of the RRPs was in part also marked by deliberation and iteration, the involvement of stakeholders was left at the discretion of member states. In this respect, while actors, both European and national, have been aware of the need to ensure inclusivity in the process, we note that in practice the RRF includes a high degree of centralization that may not always combine well with diffusion of ownership throughout the polity (including parliament and civil society as well as the state apparatus). Against this background we preliminarily identify two risks.
First, as posited by Vincent Della Sala (1997), increased EU integration and economic interdependence may lead to the 'hardening of the shell' of the state. 1 This hardening of the state has resulted in a centralization of decision-making processes that has become less permeable to penetration from (local) interests and societal demands. A similar trend can be observed in the management of the RRF, whose performance-based approach gives priority to system effectiveness over the promotion of democratization. Put differently, the RRF plans may strengthen commitment to reform by the state, but also limit space for politics by contractualizing important elements of a country's growth strategy.
Second, the centralization of decision making in a tight timeframe during the plans' drafting goes against the principles of 'flexible experimentation' (Vanhercke and Verdun 2021), which would instead require reflexivity and iteration, allowing adaptation of plans based on new information and unforeseen developments, while inducing lower-level actors to engage in problem-solving and learning. Therefore, the risk is that the actual deployment of the performancebased approach of the RRF will paradoxically lead to a lack of ownership among domestic actors, whose contribution in implementing the reforms and investments remains crucial. Similarly, in implementation, the performance-based approach risks favouring fast implementation of reforms and channelling a large amount of funds in a short period of time as political mission to ensure success in terms of output, over a focus on quality of policies and their implementation in terms of outcome. Whether these concerns materialize in practice may again depend on the balance between discipline and discretionary flexibility that is chosen by the Commission in governing the implementation of the national plans.