Does institutional misfit trigger customisation instead of non-compliance?

Abstract This article analyses the role of institutional misfit in why member states customise European Union (EU) renewable energy (RE) policies when implementing them. Institutional misfit theory posits that member states only adjust to EU policies when the adaptation pressure remains moderate and national actors’ policy preferences align. Conversely, this article tests the argument that member states manage institutional misfit by adjusting – customising – EU policies, that is, through vertical EU policy change rather than domestic change. Using Qualitative Comparative Analysis, the article compares the customisation of EU Directive 2009/28/EC in Austria, France, Germany, the Netherlands, Sweden and the United Kingdom. Surprisingly, results suggest that institutional misfit is not a necessary condition for customised implementation. Instead, when high institutional fit meets high salience, member states may issue substantively more ambitious policies than the EU requires. Conversely, when high institutional fit meets low salience, member states have no impetus to customise EU rules.


Introduction
This article is the first analysis of how the institutional misfit hypothesis assists in explaining member states' customisation of European Union (EU) renewable energy (RE) policies during implementation.Determining the conditions under which EU member states change national policies and institutions to comply with EU legislation is key to understanding the EU's problem-solving capacity (Börzel and Risse 2003;Thomann and Sager 2017;Treib 2014).A prominent approach to analysing these conditions is the institutional misfit hypothesis, which posits that for successful implementation, there needs to be at least a moderate level of institutional fit between the EU policy and the national status quo (e.g.Duina 1997;Knill and Lehmkuhl 2002;Knill and Lenschow 1998).Conversely, high institutional misfit would lead to non-compliance, with national implementation not conforming to the intentions and provisions of EU law.Originally, this hypothesis did not account for actor-related variables (Duina 2007;Mastenbroek 2005;Treib 2014).Knill and Lehmkuhl (2002) suggest that an EU policy must pass two levels: the institutional level and the actor or preference level.Whether this model accurately predicts EU compliance is disputed and has been tested elsewhere (Brendler 2022).
Previous research on the role of misfit in member states' implementation of EU policies has come to somewhat conflicting conclusions.While some studies indicate that institutional misfit poses a serious problem, leading to non-adaptation and non-compliance (e.g.Duina 1997Duina , 1999;;Knill and Lehmkuhl 2002;Knill and Lenschow 1998;van Waarden 1995), some degree of misfit is also emphasised as a necessary precondition to domestic change (Börzel andRisse 2000, 2003).In this article, we argue that institutional misfit might plausibly drive another type of implementation outcome, namely members states' customisation strategies.We thus move beyond compliance and focus on the role of institutional misfit in explaining how member states customise EU rules (Thomann 2019).Customisation measures not compliance with EU rules, but how member states interpret and change EU rules regarding their restrictiveness and density during legal transposition (e.g.Thomann 2015Thomann , 2019;;Zhelyazkova and Thomann 2022).Research indicates that customisation patterns capture more fine-grained variations in member state implementation, differ empirically from compliance patterns, and may also have different theoretical explanations (Thomann 2019;Thomann and Zhelyazkova 2017).
Our core expectation is that member states would use customisation to actively manage adaptation pressure in situations of intermediate or high misfit.Rather than changing domestic arrangements and agendas or opting for non-compliance, depending on the prevailing actor preferences, member states might change, i.e. customise, EU policies to make them fit domestic contexts.In testing this argument, our study is the first to use the institutional misfit hypothesis to specifically explain customisation outcomes.Our analysis thus contributes to the conceptualisation and empirical analysis of empirical varieties of differentiated EU policy implementation (DPI) (Zhelyazkova et al. 2023; see also Mancheva et al. 2023;Pircher et al. 2023).We study whether customisation, as one manifestation of DPI, can be explained with a classical yet contested compliance theory.Addressing this question is relevant because while goodness of fit scholarship has emphasised the question of domestic change, EU policies themselves also change during the Europeanisation process (Bocquillon and Evrard 2017;Featherstone and Radaelli 2003).However, little systematic theorising thus far has addressed potential explanations for such vertical EU policy change (Thomann 2019).
In our empirical analysis, we study how institutional misfit relates to member states' customised implementation of EU Renewable Energy Directive 2009/28/EC (RED I).Getting member states to rely on more sustainable energy was an important goal of the EU's Energy 2020 strategy.As member state's natural resources, energy mixes, and energy policy preferences differed, so did their reliance on renewable energy sources (RES) and their approaches in supporting an expansion of renewables.We study national regulatory reactions to RED I provisions, including mandatory targets as well as discretionary requirements.We compare Austria, France, Germany, the Netherlands, Sweden and the United Kingdom as six countries that vary regarding 'institutions' and 'interests' .Combining qualitative case studies with multi-value Qualitative Comparative Analysis (mvQCA), we identify the combinations of conditions under which we observe different degrees of customisation of EU provisions.
Unexpectedly, we do not find intermediate or high institutional misfit to systematically explain customisation.Instead, customisation was more prevalent under the condition of fit.As the EU's institutional guidelines emphasised active state intervention and legalism, member states with a high level of fit were also prone to calibrating the implementation of EU law, proactively going further than the EU minimally requiredthereby resulting in customised outcomes.However, we also find situations of high fit resulting in (almost) literal transposition -when there was either no need, or no pressure and no direction for customisation.Thus, institutional fit played an important, yet unexpected role in explaining customisation outcomes, interacting with issue salience.

Institutional misfit and customisation
Several versions of the misfit argument have been proposed within the existing literature.In their definition of institutional misfit, Knill and Lehmkuhl (2002) highlight a potential mismatch between European regulatory models and national regulatory styles as a crucial implementation hurdle.National regulatory styles are based on conceptions of statehood, state-society relations, and legal systems, all reaching back to the nascence of nation-states, manifesting in specific regulatory preferences, intervention styles, and problem-solving approaches, as specified below (van Waarden 1995).Other authors have referred to policy misfit, legal misfit or normative misfit, which at times were also labelled as 'institutional misfit' , making it difficult to compare empirical findings and gauge the overall explanatory power of the misfit variable (Toshkov 2010;Toshkov et al. 2010).Knill and Lehmkuhl (2002) posit that whether member states adapt to EU legislation relies on two conditions being met.First, the level of institutional fit between the EU policy and national institutional arrangements needs to be at least at a moderate level.Second, an advantageous national interest constellation vis-à-vis the EU policy's objectives and provisions is required.Thus, the domestic response to EU policies also depends on the preferences or beliefs held by domestic, political, administrative and social actors (Börzel 2000;Mastenbroek and Kaeding 2006).
While providing a parsimonious and compelling argument, in previous research, it often turned out that 'goodness-of-fit propositions provide relatively poor predictive capacity with respect to adaptation responses' (Mendez et al. 2008: 294; see also Falkner et al. 2007;Haverland 2000;Mastenbroek and Kaeding 2006;Mastenbroek and van Keulen 2006;Steunenberg and Toshkov 2009;Treib 2014; for a reply, see Duina 2007).At times, scholars had to make significant adjustments to explain legal compliance using the misfit model (e.g.Börzel 2000;Falkner et al. 2007;Knill and Lenschow 1998).
We suggest the adoption of a novel perspective on the outcomes produced by institutional misfit.We argue that member states' reactions to EU policies must be understood not only against measures of compliance or degrees of domestic adaptation (Börzel and Risse 2003), but also in terms of how EU policies are interpreted and adjusted by member states during implementation (Thomann 2019;Zhelyazkova and Thomann 2022).Rather than responding to high adaptational pressure with outright non-compliance, member states might actively manage and thus mitigate that pressure, by customising EU rules to better fit domestic institutions and interests.Importantly, this adjustment may or may not result in compliant outcomes, i.e. member states' actual conformity with (the original) EU law.Therefore, the misfit proposition might better explain customisation, rather than (legal) compliance.
Member states' EU compliance generally relates to national observance of EU law in a two-step process, consisting of (1) the legal transposition of EU law into national law (compliant when national law is in accordance with EU law) and (2) the practical application of that law (laws are actually enforced 'on the ground') (Falkner et al. 2005).We focus on member states' legal transposition of EU law, specifically, how European requirements are customised during transposition.We do not, however, aim to assess the level of (legal) compliance, instead, we focus on customisation levels and types. 1

Outcome: varieties of customisation
The customisation of rules contained in EU Directives is a wide-spread phenomenon that occurs even when member states comply with EU policies (e.g.Zhelyazkova and Thomann 2022).Customising the restrictiveness of an EU provision refers to changes in content, i.e. raising or lowering a regulatory standard, while customised density measures the quantity or amount of regulatory measures.Both restrictiveness and density can move in different directions.For example, a member state could add density by introducing a variety of new rules, but simultaneously remove restrictiveness, making measures substantively less binding or lowering levels of regulatory standards.On the other hand, a member state could integrate previously fragmented regulation into one central legislative act, thereby removing density, but without changing restrictiveness, i.e. the qualitative level of regulation.
We distinguish literal implementation and a low, intermediate or high degree of customisation, depending on which regulatory dimensions are changed (Table 1).Decisive for the degree of customisation is regulatory restrictiveness, as it captures a regulation's qualitative content.Furthermore, customised restrictiveness is a form of problem-solving and can enhance practical EU compliance (Zhelyazkova and Thomann 2022).Therefore, we seek to explain the difference between customised implementation (CUST), when the restrictiveness of EU rules is changed, and limited customisation (∼CUST, with the tilde sign ∼ reading as 'absence of '), when at most the density of rules is adjusted, but not their content.
While the extent of transposition, i.e. 'the degree to which the original directive is actually translated into national law' , including the 'number of modifications introduced' has been connected to misfit previously (Duina 1997: 156), it was essentially framed as a manifestation of non-compliance (Duina 1997(Duina , 1999)).Importantly, customisation is different from limited, non-compliant transposition.In fact, it can also manifest in member states exceeding EU requirements, indicating especially eager problem-solving.In addition to that, selectively choosing between certain options is sometimes even intended by EU law, i.e. a compliant form of implementation (Zhelyazkova and Thomann 2022).The different degrees of customisation neither follow an order in terms of their presumed regulatory effectiveness, nor do they reflect degrees of EU compliance.Instead, they allow us to analyse (1) the types of regulatory reactions to EU adaptation pressure and (2) the respective conditions explaining these reactions.

Making the misfit proposition fit for analysing customisation
We build on the model of Europeanisation mechanisms (Knill and Lehmkuhl 2002) in order to derive expectations about how different levels of institutional misfit could explain customised implementation outcomes.We conceptualise four explanatory factors as conditions, and the explanandum customisation as 'outcome' .
In order to assess institutional misfit, we first identified the regulatory model assumed by EU legislation by analysing the EU Directive in question as well as the European Commission's supplementary binding template on National Renewable Energy Action Plans (NREAPs).While the EU model gives member states discretion as to specific RE support policies and other complementary measures, it requires them to adopt a prescribed regulatory approach.Hence, the potential conflict (misfit) does not necessarily arise between a certain EU provision and a (divergent) national policy.Instead, the relevant type of misfit relates more fundamentally to member states' regulatory traditions, that is, the preferred problem-solving approaches and regulatory paradigms reflected in the national policy status quo ante.
In contrast, an analysis of the content-level misfit would not be appropriate with regards to renewable energy policy.First, a quantifiable content-level misfit would only apply to a singular rule, making a cross-comparison of content-level misfit between rules and its relevance as an explanatory factor impossible.Second, since the Renewable Energy Directive was aimed at significantly increasing member states' renewable energy ambitions, it charted a development that necessarily would have needed to be higher than that which the member states would have generated on their own, i.e. without EU interference; therefore, we would expect a content-level misfit to be the norm across member states, thus limiting the variation needed for empirical study.
Active state intervention refers to the degree in which a policy field or development is actively shaped and/or steered by the state.With regards to renewable energy, an active intervention style manifests in the state's proactive regulation of the energy sector or energy mix, with the objective to increase renewable energy production and/or consumption.With a reactive intervention style, there is very little state-led governance, i.e. the state regulates only in response to strong societal demands and/ or international obligations.
Legalism refers to the legal specification of policies, e.g.support mechanisms for renewable energy.With a legalistic regulatory style, support mechanisms are legally formalised with high specificity and detail, with the aim of universal rule application.Pragmatic regulatory styles rely on a more general framework, leaving significant room for discretion and flexibility.Administrative actors and (local) rule addressees are accustomed to continually finding case-specific solutions and/or compromise.
Accordingly, depending on where the national policy status quo ante was located on the spectrum of active vs. reactive intervention and legalist vs. pragmatic regulation respectively, the degree of institutional misfit vis-à-vis the EU model was higher or lower.Whenever national (renewable) energy policy exhibited both active state intervention and a tradition of legalism, i.e. both dimensions matched the EU model, we find low levels of institutional misfit (MISFIT = 0); if only one of the two regulatory dimensions was present, we find intermediate levels of institutional misfit (MISFIT = 1), and, finally, if none of the regulatory dimensions was present, the level of institutional misfit was high (MISFIT = 2).We do not separately distinguish policy fit, as our conception of institutional fit already refers to the respective policy (sub)field and its regulation ex ante.We traced member states' regulation over time, starting with the beginning stages of national RE policy, analysing which regulatory style was dominant in the respective policy (sub)fields and to what extent it remained stable up to the adaptation requirements posed by the EU Directive (Knill and Lenschow 1998).
Next, we distinguish three central actor-related and/or contextual conditions.First, we consider the presence of a stable socio-political consensus on the issue (CONS).Second, we consider whether influential and proactive state or non-state policy actors act as a driving force in regulation (ACT).We draw on policy-specific literature on national energy and renewable energy policy (i.e.Brand-Schock 2010;Dagger 2009;Laumanns 2005;Reiche 2005) to identify three categories of actors potentially relevant to national energy policy making: the national government and the ministries in charge, political parties in parliament, and societal actors, particularly industry actors (e.g.manufacturing industry, energy industry, agriculture).The final condition is a high economic relevance of the policy issue (ECON).
Applying the misfit hypothesis to customisation, we expect member states to use customisation as a strategy to (a) mitigate institutional misfit and (b) adapt EU policies to national policy preferences.The outcome of customised implementation does not necessarily equal non-compliance, but can also manifest, for instance, in member states exceeding EU requirements in a compliant manner.

Expectation E 1 : misfit expectation
First, similarly to how misfit is conceptualised as a necessary condition for domestic change in Europeanisation literature (Börzel and Risse 2003;Knill and Lehmkuhl 2002), we expect an intermediate to high level of institutional misfit (MISFIT > 0) to be required (necessary) for an intermediate to high level of customisation (CUST).Otherwise, member states would lack the necessary impetus to substantially alter EU provisions.

Expectation E 2 : accommodation mechanism
Second, while we expect misfit to be a necessary condition for customisation, misfit alone may not be sufficient.Knill and Lehmkuhl (2002) extend the misfit hypothesis to posit that an advantageous actor constellation is also necessary for member states to abide by European policies.Translating this proposition into our research setting, our second expectation is that customisation is utilised to (also) accommodate domestic interests.This can relate to the absence of a stable socio-political consensus (Thomann 2019), or situations when policy issues are particularly salient (Spendzharova and Versluis 2013).Policy issues are salient when proactive and influential policy actors serve as a driving force in the policymaking process and the economic relevance of a policy further spurs customisation in light of misfit (e.g.Gan et al. 2007;Reiche and Bechberger 2005;Sung and Park 2018).Thus, we expect customisation to occur only if, in addition to institutional misfit, another triggering factor that relates to the accommodation of domestic interests comes into play.This can be either a lack of consensus or high salience.

Expectation E 3 : minimal consensus
So far, we have focussed on the combinations of conditions that would produce customisation.There might also be constellations in which we would expect literal implementation, for instance, in situations of a double deadlock, on the institutional and the actor level.In this situation, domestic actors would not be able to find an appropriate winning set to introduce a new, customised policy (Steunenberg 2007; on domestic conflict during implementation, see also Pircher 2017;Treib 2008).Therefore, the EU policy would equal the lowest common denominator, given a high level of institutional misfit (MISFIT = 2) and the absence of consensus.Ergo, a lack of consensus might both drive (E 2 ) and prevent (E 3 ) customisation, depending on the context theoretically (Table 2).

Data and methods
We empirically analyse customised implementation of EU Directive 2009/28/EC in six member states.We now discuss our case selection, the method of mvQCA employed, and how we measure and calibrate the outcome and the conditions.

Case selection
We study customised implementation in the area of energy transition.The national implementation of EU Directive (2009/28/EC) on the promotion of renewable energies was the first EU policy to target the energy sector as a whole, with the goal of transitioning to renewables.Being the central EU policy on energy transition (until an update in 2018), this EU policy is crucial and exemplary in the light of large-scale techno-social transformation processes related to finite natural resources, energy security concerns and the fight against climate change.These policy issues and the related EU policy responses put member states under a great deal of pressure.With the European Green Deal, REPowerEU and other future EU policies, many more similar adaptation requirements are yet to come.Therefore, studying national reactions to the first comprehensive EU renewable energy policy seems particularly relevant and might to some degree be predictive vis-à-vis upcoming challenges as well.Aside from its socio-political relevance, EU renewable energy policy is also an interesting case to study the institutional fit of previous domestic arrangements with new EU rules, since this varies  3).
We analyse the regulatory reactions to these EU provisions in member states which vary in terms of conditions, while also controlling for possible intervening variables.We exclude Eastern and Southern European countries to control for different administrative and/or judicial capacities (Falkner 2010;Falkner and Treib 2008) and/or a differentiated history of EU membership.To vary institutional fit, we select member states with varying regulatory styles, building on van Waarden's (1995) categorisation.In terms of national interest constellations and economic relevance, we base our selection on policy field-specific data, i.e. member states' energy production mixes and shares of renewably energy, including RE leaders, RE laggards and 'midfielders' (Eurostat 2016(Eurostat , 2022)).Our selection criteria yield Austria, France, Germany, the Netherlands, Sweden and, as the EU Directive in question was implemented pre-Brexit, the United Kingdom, see Table A2 (online appendix 2 ).

Qualitative Comparative Analysis
In order to assess our expectations, we use a comparative method of analysis that captures the core assumptions of the institutional misfit framework, where the interplay of misfit and actor constellations creates necessary and (potentially) sufficient conditions for domestic implementation outcomes.Qualitative Comparative Analysis (QCA) (Oana et al. 2021;Ragin 2014;Schneider and Wagemann 2012) is a set-theoretic method, based on Boolean logic, that models these assumptions.Moreover, QCA allows us to model that the same outcome might have several, substitutable explanations ('pathways'), and that high levels of customisation might have a different explanation than low levels of customisation.Additionally, QCA allows us to combine a theory-evaluating with an exploratory logic in our analysis.
Fuzzy-set QCA conceptualises the phenomena of interest -such as customised implementation -as sets in which cases can have full membership, no membership or partial membership.For instance, some cases are members of the set of countries that customised EU rules (CUST), others are not (∼CUST).Our analysis integrates misfit as a multi-value set with different categories that can produce qualitatively different outcomes (Haesebrouck 2016).We integrate different types of sets into one QCA using the R packages QCA and SetMethods (Duşa 2019;Oana and Schneider 2018;Thiem 2014).
In a first step, if no single condition proves necessary for the outcome, conditions are added until necessity is obtained.To assess sufficient conditions, all logically possible combinations of conditions are displayed in the rows of a 'truth table' .Each case displays one of these configurations.Some truth table rows are not observed empirically ('logical remainders').In a next step, we assess whether enough cases in a given truth table row display the outcome: the configuration is sufficient for the outcome.During the process of logical minimisation, the QCA algorithm derives the shortest possible logical description of those configurations that are sufficient for the outcome.If, for instance, we observe both CONS*ACT*ECON → CUST, with the multiplication sign * indicating ' AND' and the arrow → signifying 'is sufficient for' , as well as ∼ CONS*ACT*ECON → CUST, then whether CONS is present or absent does not matter for customised implementation to occur and can be eliminated to yield a shorter expression: ACT*ECON → CUST.The Boolean plus sign + reads as 'OR'; the backwards arrow ← reads as 'is necessary for' .
All QCA results must pass the tests of empirical consistency, empirical importance, and substantive importance (Oana et al. 2021).The first two are based on the consistency, Proportional Reduction of Inconsistency (PRI), coverage, and Relevance of Necessity (RoN) parameters, which all range from 0 to 1.To account for limited empirical diversity, we make theoretically informed assumptions about the sufficiency of logical remainders and exclude illogical assumptions from the analysis (Schneider and Wagemann 2012).We complement our analysis with targeted case studies of typical and deviant cases (Oana et al. 2021).See the online appendix for raw datasets, descriptive statistics, truth tables, full results, justification of thresholds and assumptions made on logical remainders, and the R replication file.

Measurement and calibration
We now specify how we measure the outcome and conditions and how we calibrate them into sets for the QCA.Data sources include the mandatory National Renewable Energy Action Plans (NREAPs), national legislation, official documents pertaining to policymaking and the legislative process (e.g.governmental white papers, documentation on committee meetings, plenary debates, etc.), party manifestos, policy statements, and press releases by political parties and interest organisations, complemented by selective media coverage and accounts from secondary literature.

Customisation
With regards to national RE targets, we operationalise regulatory restrictiveness as the target level and regulatory density as the amount of targets, referring to sectoral and/or temporal targets.For RES-E support, in light of unsuccessful harmonisation efforts, the EU Directive does not set out a common approach; instead, the Directive urges member states to intensify support mechanisms.Therefore, we must evaluate member states' customisation against the backdrop of their previous support policies, using the domestic status quo as baseline.We operationalise regulatory restrictiveness as the level of support granted and regulatory density as the amount of support mechanisms/policies in place.The same logic applies for biofuels support (online appendix, Table A3 and A4).As the EU Directive leaves member states a considerable amount of discretion in designing their support policies, while also providing an overall regulatory model for the renewable energy sector with the template for mandatory NREAPs, this gives us an excellent empirical opportunity to study how institutional misfit might drive national customisation strategies and under which conditions member states opt for high or low levels of customisation.Customisation levels are calibrated into the four-value fuzzy outcome set 'customised implementation' (CUST): fully in (1) equals high customisation, partly but not fully in (0.67) equals intermediate customisation, partly but not fully out (0.33) equals low customisation, fully out (0) equals literal implementation.

Conditions
Table 4 summarises how we operationalise our four conditions.In Table A5 of the online appendix we list and justify all of our calibration thresholds in detail.The condition of institutional misfit is calibrated as a multi-value set, because we expect low, intermediate and high levels of misfit to produce qualitatively different outcomes.

Results
We first discuss the overall customisation patterns observed among the member states in our study, before then turning to the conditions that have produced customisation.

Patterns of customisation
In line with findings from other policy sectors (Thomann and Zhelyazkova 2017), we find that the literal transposition of EU provisions (Type 1) is a rare scenario in our cases (Figure 1).'Extreme' customisation, where member states add restrictiveness and density to EU requirements (Type 9), did not occur.Two customisation scenarios were particularly prevalent, the first being unchanged restrictiveness, but removed density (Type 2).We call this a 'nucleus implementation' , where member states aim for EU compliance by adopting the core requirements (restrictiveness), while retaining some leeway on specific details.For example, Austria adopted the EU mandated renewables target of 34% as the basis of its energy policy (Bundesministerium für Wirtschaft, Familie und Jugend (BMWFJ) and Bundesministerium für Land-und Forstwirtschaft, Umwelt und Wasserwirtschaft (BMLFUW) 2010: 7-8), but stressed in its NREAP that calculated sectoral targets were only estimations, with future changes to be expected (Bundesministerium für Wirtschaft, Familie und Jugend (BMWFJ) 2010: 1).The density dimension can thus serve as a way to mitigate adaptation pressure, e.g. when political priorities momentarily shift (Kriechbaumer 2016), while signalling compliance on the restrictiveness dimension.Even more often, however, member states did not change regulatory density, but added restrictiveness, implementing more ambitious policies than the EU has prescribed (Type 5).For example, Germany effectively instituted a 12% target for biofuels until 2020 (Bundesministerium für Umwelt, Naturschutz und Reaktorsicherheit (BMU) and Bundesministerium für Ernährung, Landwirtschaft und Verbraucherschutz (BMELV) 2009: 6), while the EU Directive required only a share of 10% for the transport sector as a whole.This finding illustrates how (some) member states go beyond EU legislation, particularly with environmental standards (Thomann and Zhelyazkova 2017).
We can also see some idiosyncrasies in how countries customised EU rules.The UK followed its well-known habit of avoiding 'gold-plating' (Thomann 2019), customising EU provisions only in terms of density.Regarding national RE targets, the UK adopted the mandatory 15%, but did not opt to integrate sectoral targets into its regulatory framework.Instead, the UK highlighted that sectoral calculations were estimations, for the sake of greater flexibility, the UK would not set fixed sectoral targets (Department of Energy and Climate Change (DECC) 2009: 8; United Kingdom (UK) 2010: 5-6, 11-12).Germany, Sweden and Austria were more extensive customisers, adding restrictiveness to at least two rules.France did not customise an EU rule on more than one dimension.

Necessary and sufficient conditions for customisation
We now analyse how the misfit framework can explain these customisation patterns -first by analysing necessary conditions and then turning to sufficient conditions.We discuss typical and deviant cases and evaluate the corresponding expectations.

Institutional misfit is not necessary for customisation
Beginning with necessary conditions, our analysis does not yield any empirically consistent and relevant necessary conditions for customisation (online appendix, Table A10).According to our misfit expectation (E 1 ), an intermediate to high level of institutional misfit (MISFIT > 0) would have been necessary for customised implementation.Our results refute this expectation.Instead, we find there mostly had to be at least an intermediate or high level of institutional fit for customisation to occur (online appendix, Table A11).Yet, as our sample is skewed towards at least some level of fit (i.e.MISFIT = 0 or 1), it is difficult to interpret this finding.Still, the results indicate that fit, not misfit, might enable the customisation of EU policies.
With regards to literal transposition or low customisation, we find low salience to be a necessary condition in our cases (i.e.no influential, proactive policy actors and/or low economic issue relevance).Only the implementation of biofuels support requirements in France, where salience was high, seems to contradict this finding.Upon further analysis, we find this to be a case of 'previous customisation' (online appendix, Table A8).Hence, this case does not substantively contradict low salience being a prerequisite for limited customisation.This observation in turn lends credence to our expectation that high salience might drive customisation (E 2 ).

Three customisation scenarios: agreement among strong actors, high fit and high salience or making up for delays
We find three scenarios suggesting that member states may customise EU policies (1) when strong actors agree on a direction, (2) high fit and high salience enable more ambitious policies or (3) to make up for delays (Table 5).
Scenario 1: consensus among strong actors.In the first scenario, similarly to successful national policymaking, influential, proactive policy actors build on a stable socio-political consensus to customise EU policies.Contrary to our accommodation expectation (E 2 ), the level of institutional misfit was irrelevant.Furthermore, it was not a lack of consensus that drove customisation, but the combination of proactive actors and a stable consensus.An example for this is the implementation of RE targets in Germany, where the national government was a central driver for an ambitious climate and energy policies.At the European level, Chancellor Angela Merkel had been a crucial figure in achieving agreement among member states, specifically on instituting binding national targets.At the national level, the coalition government formulated an Integrated Climate and Action Plan to expand existing national RE policies and simultaneously implement EU objectives (Bundesregierung 2007).Going beyond EU requirements, Germany's policy output included one law per sector, electricity (EEG), heating (EEWärmeG) and biofuels (BioKraftQuG), each containing a quantitative target (added density).Yet, there was no precedent custom of equally formalising an overarching RE target, so the binding nature of the EU's 18% target was not apparent in German law.This removed restrictiveness served to manage a very specific instance of legal misfit (Steunenberg and Toshkov 2009), notwithstanding a broader institutional fit.Hence, this finding illustrates how different variants of misfit may manifest during implementation.While 'consensus among strong actors' might be enough to drive customisation, in some cases, there might be more granular misfit at play.
Scenario 2: high fit and high salience.Second, institutional fit (MISFIT = 0), paired with proactive, influential policy actors, driving regulatory activity, as well as a high economic relevance-led member states to exceed EU minimal requirements with more ambitious national rules.We find this combination of both high fit and high salience in five cases: RES-E support in Germany, Sweden and Austria, and biofuels support in Austria and Germany.In all of these five cases, restrictiveness was added, while density remained unchanged (Type 5, see Figure 1 and online appendix, Table A1).The EU policy had spurred ambitious national policymaking in favour of RES-E and/or biofuels expansion that went beyond what was strictly necessary from a legal compliance point of view (Thomann 2019).As expected, we find that high salience drove customisation.However, it was institutional fit, not misfit, that supported this, and hence our accommodation expectation (E 2 ) is again not supported.
Scenario 3: making up for delays.In a third scenario, pertaining only to one provision implemented by France, we find customisation under the condition of intermediate misfit (MISFIT = 1), a lack of consensus and a lack of proactive policy actors.France reacted to EU adaptation pressure by adjusting RES-E policy to a higher level of support (added restrictiveness).This was surprising, as the electricity sector was still dominated by pro-nuclear policy actors and conflicts between RE proponents and status quo advocates were ongoing (Evrard 2012;Jouzel et al. 2007).Still, France chose to raise planned RES-E capacities for solar and wind energy in its 2009 Programmation Pluriannuelle des Investissements (PPI).To explain why France chose to 'lean into' EU requirements, we need to consider the temporal component, France had neglected previous EU provisions, with French RES-E shares stagnating around 14 to 15% (Eurostat 2022), despite the indicative target of 21% RES-E until 2010, set out by EU Directive 2001/77/EC.Preparing the 2007 EU climate and energy package's implementation, growing awareness of France's status as RE laggard figured in the national debate (Bocquillon and Evrard 2017: 170;Jouzel et al. 2007: 61; République française (RF) 2007: 2).In adapting national RES-E plans, France effectively made up for a delay in previous EU adaptation. 3Hence, customisation served as a catch-up mechanism.
The solution is plotted in Figure 2.There is no contradictory case (lower right corner).These scenarios together explain all but two cases of customised implementation (upper left corner), which we discuss below.

Three scenarios for literal transposition or limited customisation: no pressure and no direction for customisation, no need for customisation, or deadlock
We see three scenarios suggesting that literal transposition or limited customisation of EU provisions may take place when (1) there is either no pressure and no direction for customisation, (2) there is no need for customisation or when (3) the EU policy provides a minimal consensus (Table 6).
Scenario 1: no pressure and no direction for customisation.Scenarios with high institutional fit (MISFIT = 0), no consensus, and low economic relevance, indicating low salience, resulted in only limited customisation.There was no apparent pressure towards extensive adaptations, due to high fit and low salience, and a lack of consensus on the direction of customised restrictiveness, exemplified by the implementation of RE targets in France and RES-E support in the UK.The UK did not adjust support levels for RES-E production (unchanged restrictiveness).However, due to a continuing interparty push in Parliament, the government agreed to introduce feed-in tariffs for small-scale RES-E generation, thereby complementing the updated Renewables Obligation and thus adding density (UK Energy Act 2008).This limited customisation was sufficient to accommodate relevant interests.No further adaptations were supported by a larger coalition.As expected, a lack of consensus limited domestic customisation options.However, contrary to our minimal consensus expectation (E 3 ), the expected institutional deadlock (high misfit) was absent.
Scenario 2: no need for customisation.Another constellation that produced only limited customisation was the combination of high institutional fit (MISFIT = 0), a stable socio-political consensus, but simultaneously, an absence of influential and proactive policy actors, indicating low salience.We find this scenario in the UK's biofuels support policy and for Austria's RE targets.In the UK, the European requirement of installing a biofuels support policy and anchoring it to a specific quantitative target level did not pose a challenge to existing arrangements, as the government already employed a quota model with yearly obligation levels for fuel suppliers, embedded in a certificate trading scheme (Renewable Transport Fuel Obligation).With its high flexibility (i.e.buyouts) and in the absence of an agricultural biofuels industry, the policy was generally accepted.Hence, there was no institutional or interest-based impetus for change.
Scenario 3: minimal consensus with active actors.We find some evidence for the minimal consensus expectation (E 3 ), i.e. high institutional misfit (MISFIT = 2), together with an absence of consensus, creating a double-deadlock and resulting in an almost literal transposition (non-customisation).However, this only happened under two additional conditions, i.e. the involvement of proactive, influential policy actors and low economic relevance, and only in one case, non-customisation of RE targets in the Netherlands.Initially, the production from renewables had been irrelevant for the national energy mix and there was little political interest in RE expansion.Still, the new coalition government seemed eager to tackle climate change and introduced ambitious climate and energy objectives, e.g.20% renewables until 2020 (Christen-Democratisch Appèl (CDA) et al. 2007: 20;Ministerie van Economische Zaken (EZ) 2007: 34).However, a discrepancy between rhetoric and policy output proved problematic (Energy Research Centre of the Netherlands (ECN) 2007).Eventually, the unrealistic level of ambition decreased to a more appropriate target of 14%, in line with the EU Directive (Rijksoverheid 2010: 21), and in the end, restrictiveness remained unchanged.Density was, however, removed to preserve the market-oriented and pragmatic regulatory style, i.e. manage high institutional misfit.
As Figure 3 illustrates, the solution for the absence of customisation is also highly consistent but fails to explain two cases (upper left quadrant).

Interplay of the two customisation dimensions to manage misfit
To complement our analysis, we discuss 'most deviant' cases (Oana et al. 2021), which our models could not explain, in Table A8 (online appendix).Here, we focus on the deviant case of RE targets in Sweden, which has implications for our misfit expectation.This case would principally fit the high fit and high salience scenario, were it not for partial misfit due to a generally active, but also pragmatic regulatory tradition.While continuous state intervention in the energy sector (for example, energy taxation) had been well-established since the 1950s, there was a simultaneous preference for policy solutions that would provide substantial leeway to market actors.In 2009, the government proposed an overall RE target of 50% until 2020 (Regeringskansliet 2009: 1-2), while the EU required only 49%.Still, the opposition demanded an even more ambitious commitment (Sveriges Riksdag 2009: 48).Both the government and the opposition were actively pursuing ambitious national RE targets (albeit at different levels), resulting in added restrictiveness.Simultaneously, target density was removed, as Sweden did not employ sectoral targets (Regeringskansliet 2010).This reflected the preference for market-oriented, pragmatist solutions.So, while proactive policy actors drove implementation towards added restrictiveness, customised density was a way to preserve national regulatory traditions and increase institutional fit.This case suggests that a closer look at the interplay of the two dimensions of customisation helps us understand how customisation is used to address misfit (Zhelyazkova and Thomann 2022).

Conclusion: institutional fit, not misfit, drives customisation patterns
Applying the institutional misfit hypothesis to explain customised implementation outcomes, our core expectation was that in situations of misfit, member states might customise EU policies rather than adapting domestic arrangements or engaging in outright non-compliance (Börzel 2000;Duina 1997;Knill and Lehmkuhl 2002).However, we find that, while customisation or the lack thereof can be a way to manage intermediate or high misfit in some cases (see also Thomann 2015), customisation may more often be a way to accommodate for other aspects, e.g.national policy preferences (i.e.Treib 2003Treib , 2008)), when at least some level of fit is given.To conclude, we highlight five main results and discuss their implications.
First, contrary to our misfit expectation, customisation does not require high misfit.In fact, we observed customisation mostly under conditions of low or intermediate misfit.It follows that customisation did not primarily function as a strategy for member states to resolve high levels of institutional misfit.Instead, customisation occurred mostly under the condition of high fit, paired with high salience, in these cases.We also find institutional misfit to be irrelevant in situations of 'consensus among strong actors' , which already proved sufficient for customisation.In the majority of cases, levels of fit, not misfit, helped us explain whether or not member states choose customisation.Thus, institutional fit nonetheless proves relevant for customisation, albeit in partially unexpected ways.
Second, customisation was most prevalent in cases with both high fit and high salience.Based on the EU policy under study, we measured institutional misfit on two dimensions, active state intervention and legalism (van Waarden 1995).Member states with high fit, indicating a tradition of active state intervention and legalism in the renewable energy sector regulation, tended to also proactively (re-)regulate and (re-)calibrate salient EU provisions during implementation, in order to pursue more ambitious policies.This way, we find national regulatory styles to influence the differentiated interpretation and adjustment of EU rules during implementation.This finding adds another layer to the relevance of national regulatory styles, in addition to being foundational for national policy-making (e.g.Feick and Jann 1988;Hayward 1974;Richardson 2013Richardson [1982]]) and/or posing a possible impediment for EU implementation (e.g.Knill and Lehmkuhl 2002;van Waarden 1995).An alternative interpretation of the high fit and high salience constellation is that varying sectoral policy capacities, in terms of skills and resources of state and interest-group actors, have influenced national policy design (Daugbjerg 2022;Wu et al. 2018).Future research should specify and separate the influence of sectoral fit, salience, and capacities on EU implementation and customisation outcomes.
Third, salience matters.In line with previous research, we find that, in customising EU policies, member states can act as especially eager problem-solvers, pursuing a higher level of regulation than the EU requires (Thomann 2019;Zhelyazkova and Thomann 2022).In our cases, this happened specifically under the condition of high salience or in situations of 'consensus among strong actors' .In turn, we found that low salience can be a prerequisite for non-customisation: to remain unchanged during implementation, EU policies needed to stay out of the spotlight.Otherwise, they might elicit regulation attempts, either by proactive, influential national policy actors or simply due to their economic relevance.These findings partially align with previous research where low salience has been discussed as an explanatory factor for non-implementation (e.g.Cerych and Sabatier 1986;Versluis 2004Versluis , 2007)), indicating that low salience might lead to either transposition neglect or literal transposition.Further research on the link between salience and customisation might help clarify the effects of salience on EU compliance (Dimitrova and Toshkov 2009;Spendzharova and Versluis 2013;Vasev and Vrangbaek 2014), as well as its role in differentiated implementation.
Fourth, as suggested by our accommodation expectation, customisation can still be a way to simultaneously manage misfit and accommodate domestic interests, particularly through the interplay between the two customisation dimensions, restrictiveness and density.Removing density from EU rules can increase their fit with highly pragmatic and market-oriented domestic regulatory styles (van Waarden 1995), as observed with target-setting in the Netherlands, Sweden and the UK.Conversely, through added restrictiveness, higher political ambitions can be accommodated.As suggested by Zhelyazkova and Thomann (2022), the two customisation dimensions can serve different purposes.Therefore, their interplay deserves further attention.
Fifth, we find a dynamic over time.Previously higher customisation levels can foster literal implementation later, when the EU 'catches up' to the domestic level of ambition or a country may compensate for a previous delay in EU adaptation through customised implementation.This finding is relevant for misfit theory as it indicates that customisation can be used to accommodate different dynamics and speeds of domestic change (for example, national energy transitions) and Europeanisation (see also Solorio and Jörgens 2017).
Does institutional misfit help us explain customised implementation outcomes?For the case of EU renewable energy policy, our answer is mixed, but ultimately, yes.High levels of institutional fit appear crucial in explaining why member states choose whether or not to customise EU policies; and salience plays a crucial role.Our results suggest that when high institutional fit meets high salience, we may observe customised implementation.In such a scenario, an active and legalist regulatory style, indicating high institutional fit with the EU Directive, might drive member states to issue substantively more ambitious policies than the EU requires.Conversely, when high institutional fit meets low salience, member states often have no impetus to customise EU rules.
Our analysis is based on a set of cases that vary in regulatory styles, degrees of misfit, actor constellations, and economic relevance of renewable energy.This diverse set of cases is minimally representative of the possible variation (but not distribution) of our core variable of interest, namely different degrees of institutional misfit (Gerring 2008).Therefore, our results may be relatively robust regarding the role of misfit in customisation.However, the scope of our results is confined to EU rules with a certain political and economic relevance, in Western European countries with good administrative and judicial capacities.Particularly the role of salience in customisation may differ and interact more with administrative capacities in other contexts.Future research should further test the presented hypotheses and specifically expand on the role of institutional misfit in member states' customisation of EU policies.Our analyses suggest that DPI, here in the form of customised implementation, is a distinct phenomenon that requires fresh theorising (Mancheva et al. 2023;Pircher et al. 2023;Zhelyazkova et al. 2023).Testing the usefulness of EU implementation theories in explaining a diversity of implementation outcomes will help us gain a broader and more detailed understanding of EU implementation beyond legal compliance.

Notes
1. Our sample does not include cases of outright legal non-compliance.
However, the European Commission initiated infringement proceedings against member states that referred to the non-notification of national implementing acts as well as more detailed, sector-specific provisions like electricity grid regulation; moreover, practical compliance with mandatory RE targets for 2020 was quite varied among member states (Brendler 2022).2. For the online appendix and replication data, see https://doi.org/10.7910/DVN/BEOJSA.3.For another example of the temporal component playing a decisive role in customized implementation (but working in the opposite direction), see Table A2 (online appendix).

Figure 2 .
Figure 2. sufficient conditions for customised implementation.

Table 2 .
conditions and directional expectations.

Table 5 .
intermediate solution for cust.