Edinburgh Research Explorer Talkin’ bout a revolution? Institutional change in the EU Recovery and Resilience Facility

The European Union (EU) post-COVID-19 investment and reform programme, the Recovery and Resilience Facility (RRF), has been hailed as novel and ambitious, both as a ﬁ scal instrument and as a lever for accelerating progress towards EU climate commitments. Yet, its design also exhibits strong path dependency, drawing on existing processes and commitments. Adapting theories of institutional change and models of hard/soft governance, we argue that the RRF is an example of signi ﬁ cant yet gradual change – of evolution rather than revolution – taking place via layering and conversion of existing frameworks, and alteration of their logics of action. We show how the RRF repurposes the European Semester and track continuity and change in climate policy, a key priority area. Our ﬁ ndings suggest that the literature on institutional change should give greater consideration to the interplay between layering and conversion as a mechanism of gradual yet transformative evolution.


Introduction
Moments of crisis, where the pre-existing institutional order is destabilised whilst political actors vie to shape an exit, are often regarded as potential windows of opportunity or critical junctures for large-scale institutional change (Capoccia and Kelemen, 2007;Gourevitch, 1986;Lipscy, 2020). It has often been argued that European Union (EU) integration, in particular, advances through crises as the failure of existing institutional arrangements prompts new steps towards further integration, however incomplete (Jones et al., 2016). The EU's response to COVID-19 seemingly reflects this pattern, having set into motion a process of variegated institutional change (Schmidt, 2020).
At the core of the EU's response is the Recovery and Resilience Facility (RRF), a novel, albeit temporary, financial instrument. The RRF is part of NextGenerationEU (NGEU), an economic support package speedily adopted in July 2020. It enables the European Commission to raise funds on financial markets on behalf of the EU and make them available to support member states' responses to the pandemic. The RRF stands out, first, because of its unprecedented size (€724bn) that expands significantly the EU's spending capacity as defined in its Multiannual Financial Framework (MFF) and, second, because it consists not only of loans (€385.8bn) but also of grants (€338bn), confronting the historical taboo about EU-level borrowing (Dimitrakopoulos and Lalis, 2021;Schelkle, 2021). With the objective of 'building back better', the RRF targets two of the von der Leyen Commission's priority areas by requiring that 37% of funds be allocated to climate-related spending (€268bn) and 20% to digital spending. The RRF entails significant, although time limited, transfers of powers to the EU, in the form of fiscal transfers to member states and the creation of own resources raised and administered by the Commission to support borrowing. Its governance is based on (apparently) strict conditionality. Access to funding is tied to policy objectivesspending targetsand stricter monitoring and enforcement by the Commission, with iterative disbursement linked to implementation progress.
The RRF has been hailed as transformative, with commentators identifying a window of opportunity skilfully exploited by policy entrepreneurs, chiefly the Commission, to advance a radical new governance framework as a response to COVID-19 (Armingeon et al., 2022;Dimitrakopoulos and Lalis, 2021;Ladi and Tsarouhas, 2020;Vesan et al., 2021). However, as the development and implementation of the RRF has progressed, important elements of continuity and path dependency have been identified (Bokhorst, 2022;D'Erman and Verdun, 2022;Vanhercke and Verdun, 2022). The RRF's planning and reporting mechanisms use the existing framework of the European Semester. In terms of content, the objectives of the national plans are to be aligned with pre-existing policies. Furthermore, whilst the RRF introduces seemingly novel conditionality to funding, 1 based on previous experience, it remains to be seen if it will be enforced in practice. This raises the question of whether the RRF represents a case of gradual or radical institutional change.
This article bridges these understandings. Focusing on the design and early operation of the RRF, it addresses two research questions: (1) What kind of institutional change does the RRF represent? And (2) how and to what extent does it alter pre-existing institutional arrangements? Addressing these questions together is important because when incremental changes accumulate and alter the logic of action of the pre-existing structures, they can result in large-scale and even radical change (Streeck and Thelen, 2005). To conceptualise the type of change at play in the RRF, we draw on Mahoney and Thelen's (2010) typology. To assess to what extent the RRF has reshaped pre-existing institutional arrangements, we use the literature on hard and soft governance Bekker, 2021a;Saurugger and Terpan, 2021), focusing on the degree of centralisation, obligation and enforcement introduced by the RRF. This body of literature is particularly well suited to capture the implications of the most salient element of the RRFits use of conditionality in the disbursement of new fundsas well as less obvious changes via, for instance, reporting and monitoring. Together, they enable us to characterise the type and nature of institutional change introduced by the RRF.
Empirically, we analyse the RRF's design and operation during its first two years -2021 and 2022. In addition to examining the overarching framework and processes, we focus on the case of climate policy within it (see Appendix S1 for further details on case selection). A priority of the recovery package, as illustrated by the 37% spending target, this case enables us to track how the RRF draws on and interacts with existing policy content, goals and instruments in a well-established and politically salient policy area. The empirical aspect draws on the systematic analysis of policy documents pertaining to the RRF itself, the European Semester and other relevant governance frameworks on which it builds (see Appendix S1 for more details). We complemented and triangulated the documentary analysis with 23 elite interviews with officials from the Commission, member states and non-governmental organisations (see Tables S1 and S4).
The article makes three core contributions. First, it extends the historical institutional analysis of the RRF, which identifies path-dependent logics in the utilisation of the Semester as the RRF's vehicle for implementation. We characterise the type of gradual change introduced by the RRF as a combination of layering, where rules are added to existing institutions, and conversion, where existing rules are modified and repurposed, highlighting their interplay as a core mechanism of change. Showing how the introduction of the RRF, as a new institutional layer, has (in most cases) hardened existing rules and governance frameworks enables us to assess its transformative effects. The second contribution is to move beyond the aggregate level that has dominated the existing literature and explore the concrete implications of the RRF for institutional change in a specific area: climate policy. A well-established policy area and a priority of the von der Leyen Commission as part of its 'Green Deal', climate policy represents an important case to assess the transformative effect of the RRF on pre-existing institutions. We find that whilst the RRF has contributed to hardening rules on climate spending, its transformative effect might be less evident in practice. Finally, the article's findings enlighten wider debates about European integration and institutional change. Whilst layering has been much discussed as a mechanism of institutional change (Van der Heijden, 2011), we draw attention to processes of conversion as a crucial yet often neglected source of institutional change in European governance and highlight the interplay between layering and conversion, where new institutional layers are not only added to pre-existing institutional structures but also transform them including through hardening (or softening). In turn, 'successful' conversion can be used as justification for further layering, as suggested by the grafting of the newly introduced 'REPowerEU' plans onto the RRF, as part of the effort to achieve energy independence from Russia.
The article first reviews the literature, focusing on how it has characterised the RRF. The second section presents our theoretical approach, combining theories of institutional change with a framework to assess the hardening/softening of governance. The empirical section is in two parts; we look first at the type of institutional change underway, before turning to the implications of this change for existing institutions and governance. The final section draws out the implications of our analysis for debates on the future of the RRF, European integration and institutional change more generally.

I. The RRF: From Critical Juncture to Gradual Change
The adoption of the NGEU and creation of the RRF are widely characterised as exceptional and significant. Reporting after the marathon European Council summit, which secured the terms of the package in July 2020, media outlets framed the deal as 'groundbreaking' (Politico, 2020a) and a 'landmark agreement' (Financial Times, 2020). French President Macron praised a 'historic day for Europe' and then European Parliament President David Sassoli hailed an 'unprecedented agreement' (Politico, 2020b). It became popular to discuss whether the situation constituted the EU's 'Hamiltonian Institutional change in the EU recovery and resilience facility: The case of climate policy moment' (Kaletsky, 2020). 2 In the academic literature, the agreement on the NGEU and RRF is understood as 'a major breakthrough' (Pisani-Ferry, 2020), 'a sea change' (Creel et al., 2021) and 'a turning point' (Bokhorst, 2022). Fabbrini (2022, pp. 186-187) describes a 'paradigm change', an 'unprecedented transfer of fiscal power to the EU level' but also an accompanying transfer of political power that 'constitute[s] a watershed in the process of European integration'. The literature to date has analysed the role of national actors in influencing the design of the recovery package (Bekker, 2021b), the inter-institutional dynamics that have driven its adoption (De la Porte and Dagnis Jensen, 2021; D'Erman and Smeets and Beach, 2021), the practical operation and rules of the RRF (Fabbrini, 2022;Pisani-Ferry, 2020), and the configuration of 'winners and losers' in the new, combined Semester and RRF (Vanhercke and Verdun, 2022).
These assessments generally identify four particularly important innovations (D'Erman and . The first is the provision by which the EU will raise funds of its own on capital markets, breaking with member states' traditional reluctance here. Although the EU has borrowed on capital markets before, the scale and structure of the current provisions is unprecedented; the Commission will borrow up to €800bn between 2021 and 2026 and, to repay loans, can generate its 'own resources' via new taxation instruments (European Commission, 2021a). The creation of new EU resources, often considered a taboo for member states given the transfers of power they imply, represents a second major innovation. The third novelty is the inclusion of grants along with loans in the package. The former represent most of the funds, a significant shift from the approach taken in response to the sovereign debt crisis of 2009(Pisani-Ferry, 2020Schelkle, 2021). Finally, the disbursement of the money is conditional upon the vetting of national plans for investments and reformsthe National Recovery and Resilience Plans (NRRPs)by the Commission to ensure that they contribute to the overall objectives of the NGEU.
Despite the innovative nature of the RRF, the extent of its 'path dependence' has become clearer over time. Though characterising the provisions as novel, Schelkle (2021, p. 51) acknowledges that they are '… incremental insofar [as] they are reversible'. The RRF is temporary, in place only until 2026. Its tax-and-spend provisions are explicitly linked to addressing the pandemic and its redistributive nature means that wealthier states will have an interest in revisiting its provisions (Schelkle, 2021, pp. 51-52). As for the RRF's governance, Vanhercke and Verdun (2022) find that this is based on pre-existing structures, namely, the European Semester. The Semester was chosen as the vehicle for implementing the RRF because it strikes 'a balance between providing sufficient constraints, while leaving considerable leeway to the member states' (Vanhercke and Verdun, 2022, p. 208). They identify further elements of path dependency in the decision to repurpose the Semester, pointing to earlier discussions of a budgetary instrument for convergence and competitiveness (BICC), which normalised the idea of using funds from the EU budget for national investments and reforms, and pre-existing patterns of actor engagement under the Semester, which have been fostered under the RRF.
The decision to manage the RRF via the Semester raises the potential for a fundamental change in the character of the latter, 'from being a non-binding structure for policy coordination to a vehicle for the allocation of a major economic impetus which is to have more teeth' (Vanhercke and Verdun, 2022, pp. 217-218). The 'teeth' are provided by the conditionality implied in the RRF's requirement that member states demonstrate progress towards their Country Specific Recommendations (CSRs)the main output of the Semesterto receive funds (see also Bekker, 2021b). 3 Closer scrutiny, however, raises questions about the assumed impact of such conditionality. The scale of the funding and the explicit policy linkage between the RRF and the CSRs have the potential, on paper, to harden the Semester and increase compliance (Nguyen and Redeker, 2022;Schelkle, 2021, p. 51). Yet, although the RRF provides more 'implementation power to the Commission' (Bongardt and Torres, 2022, p. 13), it is not clear that the latter has, in practice, the capacity to engage in rigorous scrutiny of the plans and monitoring of implementation (Vanhercke and Verdun, 2022, p. 218). Given the need for rapid implementation, the Commission will likely be under intense pressure to approve disbursements and may lack the legitimacy to apply strict conditionality (Bokhorst, 2022, p. 13). In short, as the RRF enters early-stage implementation, it is increasingly understood as a case of path-dependent, gradual evolution. The potential of novel elements, such as conditionality, to transform more fundamentally the governing logic of the Semester depends on their implementation.

II. Theorising Institutional Change
Building on the insights of the most recent literature, we draw on theories of institutional change to assess the type of change at play with the RRF, and on models of hard and soft governance to explore the implications of these changes for governing logics, in both the overarching RRF and climate policy.

Types of Institutional Change
Historical Institutionalism (HI) is concerned with how existing decisions and institutions affect the choices available to policy-makers and the development of new institutions. It is built on two core conceptual pillars: critical junctures and path dependence (Pierson, 2000). Critical junctures represent the 'brief periods of institutional flux … during which more dramatic change [than that usually feasible] is possible' (Capoccia and Kelemen, 2007, p. 341). In between these periods of flux, HI posits that institutional development is characterised by long periods of institutional stability, path dependence and gradual change.
Work in the HI tradition has shown, however, that small adjustments need not always be path dependent or conducive to inertia; instead, they can be 'cumulatively transformative' (Palier, 2005, p. 130). Seeking to move away from work based on punctuated equilibria and stable orders, Streeck and Thelen (2005;also, Mahoney and Thelen, 2010) explore how small, incremental changes dramatically shape institutional structures and processes over time, thus shifting attention from short-term events as historical break-points to longer term change. Mahoney and Thelen (2010) identify four kinds of gradual institutional change: displacement, layering, drift and conversion, which can be identified by observing changes in existing and new rules or norms (see Table 1). They also identify four observable dynamics that determine the type of institutional change at play: the removal of old rules, the neglect of old rules, an intentional change to the impact or enactment of old rules, and the introduction of new rules.
Where existing rules are removed or neglected, institutional change is characterised as either displacement or drift. In displacement, new rules replace the old rules, whilst in cases of drift, the old rules are ignored, side-lined or not updated to account for developments in technology or policy environment; in both cases, the old rules become irrelevant. Where old rules remain relevant, the institutional change is a case of either layering or conversion. Layering occurs when new rules are added to existing ones. It involves 'active sponsorship of amendments, additions, or revisions to an existing set of institutions' that might later crowd out the existing rules or structures (Streeck and Thelen, 2005, p. 24). By contrast, in cases of conversion 'institutions are redirected to new goals, functions, or purposes' (Streeck and Thelen, 2005, p. 26). New rules are not needed. Instead, old rules are enacted in new ways or their impact on behaviours change due to new policy conditions. Their inherent ambiguities are exploited towards new objectives or to serve new actors' interests.
In contrast to HI accounts that emphasise adaptation and continuity, Streeck and Thelen (2005, pp. 8-9) posit that gradual changes can accumulate and slowly transform an institution, by altering its fundamental 'logic of action'. This can be understood as an institution's 'stable core', including its mode of governing and/or underpinning ideology. Examples include the introduction of private alongside public pension systems as embodying different logics and the shift from centralised to decentralised healthcare systems (Faletti, 2009). In this article, we consider not only the type of change but also its extent or intensity. A key innovation of the RRF is the large fiscal resources available and associated conditionality. To capture the alterations that such innovations entail in terms of 'logic of action', we draw on the literature on hard and soft governance.

Hardening and Softening Governance
There has been a recent growth in the literature that explores the 'hardening of soft governance' (Bekker, 2021a(Bekker, , 2021bBocquillon et al., 2020;Saurugger and Terpan, 2021), reflecting an observed 'blurring between hard and soft' forms of decision-making (Graziano and Halpern, 2016, p. 5). This literature starts from the observation of a continuum running from non-legal norms (such as declarations), through soft law (such as recommendations or guidelines), to hard law (such as regulations and treaty provisions) (Saurugger and Terpan, 2021, p. 4). An influential set of studies assesses the 'degree of hardness' observed based on three criteria: the extent of obligation that it puts upon actors, the degree of precision that it contains and the extent to which authority is delegated to a third party . Adapting these criteria, Terpan (2015) uses the strictness of enforcement, rather than the degree of precision, to assess EU soft law, whilst Knodt et al. (2020) retain precision but add the requirement to justify, mechanisms for blaming and shaming, and the strength of pressure from third-party actors in their assessment of hardness. Degrees of hardness are not fixed, however (Saurugger and Terpan, 2021). Building on existing models and applying them specifically to the Semester as amended by the RRF, Bekker (2021b) develops an analytical framework for assessing hardening and softening trends, which we adapt here (Table 2). Crucially, this framework draws on de la Porte and Heins (2015) to incorporate the key innovation of post-euro crisis EU governance: conditionality.
Bekker (2021b) structures the assessment of hardening and softening around three core traits: centralisation, obligation and enforcement. Within the context of the EU's division of competences, provisions, which give a larger role to national actors and priorities, suggest decentralisation and a softening of EU involvement and influence. By contrast, highly centralised provisions, which reduce the role of national actors and impose EU-level objectives, suggest a hardening dynamic. Attaching a particular policy provision to a non-binding or binding rule will, respectively, soften or harden it. Describing a provision in vague terms or with generalised standards will soften it, whilst specifying in detail and thus reducing margin for discretion will harden the provision. Finally, enforcement is softened by the lower frequency of monitoring activities and weaker (or absent) mechanisms to redress infringements and hardened by increased monitoring frequency and stronger corrective mechanisms that punish failure to comply (i.e., sanctions). Increased financial conditionalityi.e., adding requirements that must be satisfied in order for funds to be releasedhardens the enforcement dimension of a policy provision; reducing conditionality softens it. We use this frameworkapplying it in a manual review of the policy Institutional change in the EU recovery and resilience facility: The case of climate policy documents and triangulating with interview data (see Appendix S1)to assess the implications of the changes introduced by the RRF and whether they constitute a shift or transformation in the logic of action, that is, the hardness or softness of governance.

III. Gradual Institutional Change in the RRF: Between Layering and Conversion
Designed to increase the EU's oversight of national economic and fiscal policy in the aftermath of the 2008 financial crisis, the Semester is an iterative policy process. Common EU priorities are identified [and published in the Annual Sustainable Growth Survey (ASGS)], governments integrate these into national plans, known as the National Reform Programmes (NRPs), and the Commission evaluates the plans against their contribution to the priorities via ongoing monitoring and reporting, culminating in the adoption of CSRs by the EU.

Repurposing the Semester
To support the implementation of the RRF, these existing structures have been repurposed (conversion) and several new rules 4 added (layering) ( Table 3). The RRF sits underneath the Semester, in this sense, and is quite different in nature. Whereas the Semester provides broad priorities and steering, the RRF provides detailed and prescriptive direction. The RRF is the latest of several policy frameworks to be linked to the Semester's horizontal architecture (others include the structural and investment funds, and support from the European Central Bank) and, as such, its provisions are based on the Semester's priorities. The RRF outlines six policy pillars to guide member states' reform and investment plans, introduces 20% and 37% spending benchmarks for digital and climate, respectively, creates a new Recovery and Resilience Task Force (RECOVER) within the Commission to oversee both the RRF and the Semester and provides for an 'emergency brake'controlled by the Councilto delay funding disbursements if a member state feels that the Commission has been too lenient in its assessment of another state's progress. Each constitutes an example of a new layer of rules added onto those of the Semester. In addition, existing rules have been converted. In the 2021 cycle, the ASGS, which launches the Semester each November, was published early and reoriented to provide strategic guidance for the drafting of the NRRPs, the one-off plans that outline how member states will spend their portion of the RRF funds (European Commission, 2021d). The annual NRPs continue but now serve as one of the two annual reporting requirements for the RRF, in addition to their original reporting capacity within the Semester (European Commission, 2022a). For the 2021 cycle, NRPs and NRRPs were combined and submitted together. From 2022, the Country Reportspublished by the Commission to offer a more in-depth analysiswill include an assessment of weaknesses in the implementation of the NRRPs, flagging instances where these fail to address issues raised in the CSRs.
None of the changes have involved the removal of existing provisions and there is little evidence of their neglect (Table 1). The debt and deficit rules contained within the Stability and Growth Pact (SGP) have been temporarily suspended but their monitoring continues. The preventive and corrective arms of the SGP remain in place and though, as Vanhercke and Verdun (2022, p. 212) note, some elements of the Semester were put 'on hold' to accommodate the RRF in 2021, this does not seem to be part of a strategic (in)action, so much as a pragmatic necessity. This is unsurprising given that the RRF is a distinct institution under the umbrella of the Semester and cannot displace it. It is possible that particular Semester rules will be displaced or replaced as part of the ongoing revision of the economic governance framework; the outcomes of this are yet to be determined but 'a simple return to the pre-RRF Semester in 2026 can be virtually ruled out' (Nguyen and Redeker, 2022).

Climate Policy Within the RRF
Similar trends are seen in climate policy. The RRF does not displace existing climate objectives or rules but adds some new ones (layering) and repurposes multiple existing policies and instruments (conversion).
As specified in the RRF Regulation, the objectives of the NRRPs are to be aligned with those of the Energy Union Governance Regulation (EUGR) and its National Energy and Climate Plans (NECPs) and be coherent with other pre-existing policies. Looking at headline climate targets, the RRF introduces a new rule, in the form of a binding target for national spending on climate investments of 37% as part of the NRRP. 5 This builds on the pre-existing soft target to commit 30% of Union-level expenditure in the MFF (previously 20%) to climate action (Rietig, 2021). Moreover, space has been created within the Semester to introduce new climate reforms. Whilst energy policy reforms have found a place in the Semester, particularly since 2020, climate reforms had not been systematically discussed (see Table S2). In including climate as a specific priority, the RRF has enabled related commitments in new areas, such as biodiversity, environmental taxation and public transport infrastructures (Interviews 12 and 13). 6 By far, the dominant dynamic of climate policy under the RRF, however, is conversion. Most interviewees confirm that a significant proportion of the climate-related  Table S4 provides examples of illustrative quotes supporting the discussion in this empirical section of the paper.
investments and reforms proposed in the NRRPs are based on pre-existing plans and policies. This is largely explained by the speed at which the NRRPs were drafted: NRRP reforms and investments were drawn from the Country Reports and CSRs, 7 preparations for the next MFF and Cohesion Fund provisions, and policy-specific documents such as the NECPs (Interviews 11, 12, 14-16, 18 and 23). These were mined for inspiration and ideas, particularly as drafting was happening rapidly (Interviews 15 and 18). This drove member states 'to bring forward what could be brought forward' (Interview 16). As an NGO interviewee highlighted, Member states had to come very quickly with proposals for a huge amount of money […] in general, governments put plans and projects that were already in the drawers […] which is logical because you cannot invent a brand new thing. (Interview 12) There is evidence (see Table S3) of policies being replicated across the Country Reports, the CSRs and the NRRPs. For instance, in the case of Malta, the 2020 Country Report describes detailed energy efficiency policies for buildings, which are then discussed in the CSR recital and in more general terms in a Recommendation, underpinning the inclusion in the NRRP of specific measures as part of a €78m budget request for RRF funding.
Whilst the timeline required the utilisation of existing plans, time pressure was also useful for novel developments. In the words of an interviewee from a national representation: 'So we are moving fast … 50% were in the pipeline … [and] 50% of new things […] This could be challenging, but I believe it also brings good things because they need to happen fast, they need to happen in the following three years' (Interview 18). This estimation that about half of NRRP proposed investments and reforms proposed in their country's NRRP are pre-existing projects was also shared by another member state representative (Interview 23). Such repurposing was facilitated by a provision in the RRF Regulation back-dating eligibility to cover all projects adopted since 1 February 2021 (Regulation EU 2021/241, paragraph 32). It also reflects findings in other sectors, such as social policy, where 'the RRF has contributed to fasten-forward the implementation of welfare reforms and initiatives which would have […] remained on the paper [sic]' (Corti and Vesan, 2023, p. 513).
The RRF has also converted some elements of the EUGR. Like the Semester, the EUGR has a bottom-up, recursive design where national plansthe NECPsare drafted by member states, guided by EU headline targets, reviewed by the Commission and revised based on those recommendations (Bocquillon et al., 2020). The NRRPs have, to varying extents, been informed by the NECPs. This was particularly the case where 'granularity' was required at short notice (Interview 11). In some cases, such as Belgium and Cyprus, the 2020 NECPs are widely referred to in the 2020 Country Reports and 2020 CSRs and are also cited as a key source for NRRP policies in 2021 (Table S3). For example, in Belgium, there is a strong focus on building stock renovation across the different plans (Table S3), and the NRRP includes this as the largest spending item (€1bn out of a total of €6bn for reforms and investments) (European Commission, 2021e). In Cyprus, a recurrent referral is made to sustainable and green mobility, and in the NRRP, it is a 'Key Measure' with €89m of €1.2bn total investment planned to target this (European Commission, 2021f). In other cases, the NECPs have been used in a more rhetorical than substantive manner, owing in part to the broader pitch and, in some cases, the outdated nature of their provisions (Interviews 11-14).
The conversion of policies and instruments under the RRF was not unqualified. The Commission reviewed the NRRPs critically, forcing changes to provisions not deemed compatible with the climate objectives (Interview 14)such as the repurposing of coal and gas power plants in Bulgaria (Interview 15), and projects on waste incineration and gas infrastructure promotion (Interview 18)on the basis of the 'do no significant harm' (DNSH) principle (see below). Whilst funding of previously planned projects was expected, under the additionality principle, which prohibits the use of RRF funds to cover costs that are addressed under other national or Union programmes, the Commission also rejected the inclusion of projects already funded by, for example, the Cohesion Funds (Interview 12). However, driven by the speed required and the desire to maintain coherence between policy frameworks, the dominant trend has been one of conversion and repurposing.

IV. Gradually Hardening Existing Governance Frameworks
Next, we turn to the question of whether the changes introduced by the RRF via layering and conversion of existing frameworks have altered their logic of action and amount to a hardening or softening of those frameworks. We focus on the evolution of centralisation, obligation and enforcement in the adapted Semester and climate policy.

Centralisation
Formally, the RRF Regulation gives national actors a stronger role in drafting the NRRPs than the Semester's NRPs or CSRs, not least because of time pressure. Plans should address an 'all or a significant subset of CSRs' (Article 18) and 'should be consistent with the relevant country-specific challenges and priorities identified in the context of the European Semester' as well as the NECPs (Article 17) but member states are not formally required to incorporate specific initiatives or recommendations [Regulation (EU) 2021/ 241 Article 18 (4b)]. This would seem to decentralise the drafting process -'the thrust was on [member states'] side' (Interview 11)resulting in substantial variations in the form and content of national plans (Interview 12).
However, in practice, the picture is more nuanced, suggesting a higher degree of centralisation. The NRRPs were created in 'mutual dialogue', with extensive and intensive bilateral co-operation during the drafting process (Bekker, 2021b;Interviews 15 and 18). Interviewees point to a difference between reforms and investments (Interviews 11, 12 and 18). The Commission had extensive prior experience with reforms as part of the Semester, overseeing the drafting of Country Reports and CSRs. Consequently, they were able to steer the content of the NRRPs more directly, proposing specific reforms, milestones and targets, with pressure on member states to compromise to get their national plan approved (Interview 18). In contrast, having previously lacked a significant fiscal capacity, the Commission was less experienced with guiding investments and the member states retained more control of these parts of the NRRPs. The speed at which they were expected to come up with many investment projects also meant the Commission had neither the expertise nor the capacity to get involved in the details of project selection, being confined to checking the 'project fiches' met the selection criteria and eligibility for funding, or asking for more concrete proposals (Interviews 14, 15, 18 and 23). This aligns with findings elsewhere in the recent literature that 'the Commission mostly did not interfere with national priorities for investments' but played a greater role in steering reforms (Bokhorst and Corti, 2023, p. 8).
In the RRF, leverage is linked to the scale of resources vis-à-vis national gross domestic product (GDP). When countries did not rely heavily on the RRF or were already relatively ambitious on climatesuch as Belgium, Denmark or the Netherlands, for instance the influence of the Commission was reduced (Interviews 11-14). In contrast, for countries that were more heavily reliant on RRF funds, such as Bulgaria or Italy, the leverage was enhanced. Bulgaria's NRRP offers one such example of this influence. The Commission had reviewed Bulgaria's draft NECP (Bulgarian Government, 2020, p. 68) in October 2020 and criticised the plan for its lack of coal phase-out plans including the conclusion that a revision 'will have to respect the principle of "do no harm"' (European Commission, 2021c, p. 3, p. 20;Interview 14). Bulgaria's NRRP ranks as one of the largest (after Croatia) as a share of grants in relation to GDP, at 10.2% of its 2019 GDP (European Parliament, 2023; see Figure S1). The leverage of the RRF's conditions and funding was considered significant (Interview 14) in shaping a revised plan, in February 2021, which included a reference to phasing out coal (Bulgarian Government, 2021a) and a plan for converting coal to gas power stations estimated to cost €244m, 70% funded by the RRF (Bulgarian Government, 2021b, Annex 13, p. 4). The Commission forced further revisions to unlock RRF funding through the incorporation of a concrete coal phase-out date (of 2038) in the fourth iteration of the plan in October 2021 (WWF, 2021). This example illustrates how the RRF has been used by the Commission as an instrument for more intrusive intervention into national policy choice and, ultimately, transformative change even in the face of resistance. As a member state interviewee puts it, you see 'magical things happening … when you give money on this scale' (Interview 16).

Obligation (Codification and Precision)
Codification: Compared with the Semester, the RRF is more codified. The RRF Regulation sets out guidance on how to write NRRPs, identifies six priorities that the plans must speak to, stipulates that plans should address recent CSRs, defines two specific spending targets and provides detailed rules on the eligibility of investments in the NRRP (Bekker, 2021b, p. 181). This is especially the case for climate. In contrast to the looser EU climate spending target in the MFF, the 37% spending target in the RRF is enshrined in EU law and directly binding on member states. In addition to giving the Court of Justice of the EU a role in enforcement, this implies that the Commission is accountable to the Court of Auditors (ECA) if it fails in its oversight of the target, increasing pressure on member states (Interview 11).
The headline target is also associated with a relatively detailed classification of climate spending (the so-called 'Annex 6' of the RRF Regulation), which reduces opportunities for 'greenwashing'. This 'methodology for climate tracking' includes a 9-page list of 143 different investments each categorised as making a partial or total (or no) contribution to climate adaptation or mitigation. For example, wind or solar investments are categorised as 'substantially' (100%) contributing. Member states are also obliged to present a 'qualitative explanation' for whether investment projects account for climate spending (RRF Regulation, Article 18). Investment and reforms must also comply with the DNSH principle as defined in Article 17 of the 'Taxonomy Regulation' [Regulation (EU) 2020/852] including the general prohibition of 'significantly harming' climate change mitigation. A Commission notice provides technical guidance on DNSH's application under the RRF, including a checklist to assess whether environmental objectives are met.
Precision: Overall, precision within the RRF has been enhanced. First, at a general level, the increased amount of data now at the Commission's disposal (from the Semester and other processes, such as the NECPs) has provided it with more knowledge of the situation within member states. This has enabled greater precision in the drafting of objectives in the NRRPs, facilitating implementation and enforcement (Interview 11). With regard to the 'climate tracking' methodology, some interviewees suggest that it has also provided the Commission with a 'more precise, more detailed' way to assess climate investments and projects in the NRRPs (Interview 11) and reduced subsequent tensions and negotiations with member states in assessing the plans (Interview 13). Others though, especially from the green NGO sector, describe the categories as too broad to be fully effective in directing investments and avoiding greenwashing (Interviews 12 and 15). With regard to the DNSH principle, it is generally seen as less precise and therefore more difficult to implement (Interview 15).
Progress is also measured against a set of objectives that are more specific and often quantified. The milestones and targets negotiated between the member states and the Commission for the purpose of monitoring and for the disbursement of additional funds are more numerous and precise than in pre-existing frameworks such as the Semester (Interview 11). They enable the Commission to assess progress more closely at key steps of the implementation process.

Enforcement (Monitoring and Sanctions)
Monitoring: Because of the funding at stake, reporting is frequent with multiple disbursement points and more detailed with the new milestones and targets (Interviews 11 and 22). Under the RRF, the Commission is empowered to adopt delegating acts setting out common indicators for use in reporting, monitoring and evaluation of progress towards the NRRPs' objectives, as well as the methodology for calculating these, and a new scoreboard of indicators to measure progress in implementing the NRRPs. 8 Emphasis is also put on national control systems that implement the Commission 'ambitious audit strategy' and governments will have to justify the continued appropriateness of these controls when submitting revised plans (European Commission, 2023). Monitoring within the context of the Semester will now cover not only the CSRs but also the detailed NRRPs, giving the EU oversight of specific national reforms and, for the first time, investment portfolios. The new RECOVER Task Force in the SECGEN will, jointly with DG ECFIN, co-ordinate the monitoring of the NRRPs, and an amended framework for assessing milestones and targets was published in February 2023 (European Commission, 2023, Annex 1).
This enhanced monitoring creates challenges in terms of member states' ability to rapidly deliver on their objectives and reporting requirements (Interview 15). 'Administrative burden' has increased beyond initial expectations, limiting flexibility for states with comparatively small resource allocations (Interviews 13, 16 and 17), for those with smaller administrative capacity (Interviews 18 and 19) and even for the Commission (Interviews 21-23). As the stimulus is time limited, the Commission has been keen to try and avoid absorption capacity issues common with Cohesion Funds, but this has proved challenging (Interview 14).
Sanctions: The most visible change made by the RRF is the introduction of conditionality as a tool of enforcement. Whilst the literature has explored a continuum of leverage in the Semester, where states in receipt of EU funds experience something akin to conditionality in their CSRs (Baeten and Vanhercke, 2017;Bauer and Becker, 2014;Di Mascio et al., 2020), the RRF introduces this formally, making receipt of RRF funds conditional upon performance: namely, the achievement of milestones and targets within the NRRPs and progress towards the CSRs. The heated debates that shaped the final RRF reflected the polarised position of member states on conditionality. What is novel in the RRF is that implementation is staggered and the disbursement of loans and grants is conditional upon meeting milestones and targets. Failure to comply could potentially result in delayed disbursements or partial payments, though at the end of 2022, member states were not clear how this would be interpreted and enforced (Interviews 20-22). 9 This represents a partial hardening, at least in principle, but the willingness of the Commission to flex its muscles by proposing to withhold (climate) funds remains to be seen. A new methodology for payment suspensions, published in February 2023, seeks to impose uniformity on these decisions but also indicates that flexibility around unavoidable implementation delays will be exercised (European Commission, 2023). Two factors that may serve to force the Commission's hand in sanctioning are the Council's rolevia the Economic and Financial Committeein applying the 'emergency brake' and the ECA's oversight of the Commission's implementation of the RRF.
Concerning climate policy, a plan that does not reach the 37% target, as assessed by the Commission, 'will not be accepted' (European Commission, 2021b). The Commission already had extensive opportunities to monitor progress towards energy and climate objectives as part of the NECP process and various (associated) provisions enshrined in energy and climate legislation; yet budgetary provisions remained limited. With the RRF, a high level of monitoring and the possibility of sanctions help with tracking NRRPs' implementation, ensure their climate integrity and strengthen enforcement. Indeed, all member states have met, or exceeded, the 37% threshold, with 40% of the plans allocated to climate on average with several member states over 50% (European Commission, 2022b). An interviewee from the Commission describes a sense that the ECA is 'on our neck' and this has translated into NRRPs that observe the target and, in some instances, go beyond it 'to be safe' (Interview 11). In practice, however, the stringency and accuracy of the tracking methodology remains contested. There may also be some remaining discretion due to the Commission's limited monitoring capacity, especially for investments at the project level, and the overriding objective to demonstrate that the RRF is a 'success'. As an interviewee puts it: 'the Commission will not be looking at all of the individual projects […] one of the goals of the Commission here is to show that the recovery is working and the money is being spent […] to show that this instrument is working' (Interview 14) (see Table 4 for a summary of hardening and softening dynamics).

Conclusions
Whilst crises often appear initially as moments of radical change, a longer view can reveal a more nuanced picture. Hailed by many as an institutional revolution at its inception, the RRFthe centrepiece of the EU's economic response to COVID-19increasingly appears as an evolution of the EU's economic governance rather than its revolution. It displays elements of continuity and change both at the framework level and in the specific policy area of climate.
Drawing on theories of institutional change, this article advances the academic debate on the novelty of the RRF. It argues that the RRF is best understood as an instance of gradual yet transformative change. Looking at the type of institutional change at play, we identify a mix of layering and conversion. The RRF has not displaced existing processes of economic governance such as the SGP or European Semester, nor has it replaced the well-established body of existing climate policy objectives and instruments. Rather, it has been temporarily added onto them and partially transformed them. In the case of climate, stimulated by the RRF, the Semester now includes moreand more precisereferences to energy and climate projects and reforms. Whilst the NECPs remain  unchanged, their revisions from 2023 will now integrate RRF funding and projects. Looking more closely at the nature of institutional change, the RRF has hardened existing processes and instruments, through the carrot of new funding and the stick of conditionality. As part of the Semester process, the RRF has increased the codification and precision of obligations included in national recovery plans and strengthened monitoring and sanctions for non-compliance. However, past experience with the SGP and Semester, and evidence from the early phases of the RRF, suggests that, in practice, conditionality might not be as stringent as expected and hardening more limited. The funding available has also accelerated the implementation of pre-existing climate and clean energy projects and objectives, even though speed and capacity issues have meant a mixed record in terms of the projects' novelty and climate integrity. Emerging evidence from other policy areas such as social policy (e.g., Corti and Vesan, 2023) suggests that these dynamics may not be limited to the climate case and could be widespread across other policy areas. Systematic comparative analysis across policy areas thus represents a promising avenue for further research.
The RRF itself is now evolving in the context of Russia's invasion of Ukraine and associated energy crisis. In May 2022, with the aim of rapidly phasing out EU dependence on Russian fossil fuels, the Commission published the REPowerEU Plan, a proposal amending the RRF Regulation to include an extra €72.3bn of financing to support member states in fostering energy independence. Governments are encouraged to amend their NRRPs by adding a chapter of relevant reforms and investments, subject to the same monitoring arrangements as other RRF measures, under the existing Semester framework. This initiative extends the RRF by adding a new layer onto it, thus solidifying its role as a central EU economic investment and reform framework. It also triggers further conversion, expanding the RRF to energy security and creating potential tensions with climate commitments, when achieving energy independence is framed as a justification for the development of fossil fuel infrastructures.
Whilst the RRF was originally designed as a temporary crisis response mechanism, it may therefore have initiated a gradual process of transformative change of EU economic and sectoral governance frameworks. The Eurozone crisis precedent suggests that it is not uncommon in EU integration for temporary institutions to endure and shape further institutional developments (Gocaj and Meunier, 2013;Schwarzer, 2012;Verdun, 2015). We find that the operation of the RRF may not be as radical a break as it first appeared, yet our analysis also highlights its transformational potential, as it alters the existing institutions that it builds on. This finding offers insights for other cases of integration through crisis, which would deserve further investigation.
The case of the RRF also sheds light on a neglected dimension of institutional change in the EU and beyond: the interplay between mechanisms of layering and conversion. Whilst the existing literature has tended to focus processes of layering (Van der Heijden, 2011; Van der Heijden and Kuhlmann, 2017), we suggest that conversiona 'hidden face' of institutional change (Hacker et al., 2015) is an equally important yet neglected mechanism of change (see also Thatcher and Coen, 2008). Although it can be tempting to look at different mechanisms or types of change in isolation or separately from one another, we find that their interplay (or 'sequencing'; see Van der Heijden and Kuhlmann, 2017, pp. 544-545) is equally important. In the case of the RRF, layering has fuelled conversion, in turn calling for further layering. Moving beyond static typologies, our analysis therefore calls for more attention to, and a better understanding of, how the two mechanisms interact in shaping gradual yet transformative institutional change.