The power of implementers: a three-level game model of compliance with EU policy and its application to cultural heritage

ABSTRACT

In this article we focus on compliance with European Union (EU) directives in the context of multilevel governance. Policies specified by EU directives move through different decision-making stages before they are implemented. We integrate these decision-making stages and actors in a game-theoretical model aiming to explain implementation in the EU setting. Practice and organizational literature findings already indicate that in this setting formal and informal policies can differ considerably. Our model shows that such a divergence is the result of the interactions of three sets of actors at different levels. We illustrate the main findings of our model with cases of transposition and implementation of the EU rules regarding cultural heritage in the European Union. Based on our analysis, we suggest that EU policy implementation is best understood as a patchwork of domestic processes in which implementing actors affect outcomes within limits set by national and European decision-makers.

KEYWORDS:

• Bureaucracy
• compliance
• cultural heritage policy
• European policy implementation
• multilevel games

1. Introduction

How do European policies work when national politicians are divided? Ever since Lipsky’s (1980) ground-breaking book, scholars of public policy have acknowledged that implementation of policies depends on the views and capacity of street-level bureaucrats as much as policy-makers. Looking at the European Union (EU) context, Lipsky’s arguments suggest an additional challenge for implementation. To be effective for the EU as a whole, many (if not all) EU policies require alignment between politics (in transposition) and bureaucracy (in implementation) at different levels of governance. In this article we present a three-level model of implementation and a set of empirical cases in the area of cultural heritage policy that show alignment between implemented policies of member states is quite difficult to achieve. As the European Commission found in its implementation reports, national administrations were unaware ‘ … of the existence – and the need for protection and defence – of goods which are part of the cultural heritage of other member states’ (European Commission 2000: 7).

The literature on compliance with EU law comprises more than three generations of studies using different models and approaches to transposition and policy implementation in the EU (Mastenbroek 2005; Toshkov et al 2010; Treib 2014).¹ Given the need for extensive data gathering at the micro level to study the actual implementation of policies, transposition studies still dominate this literature, yet the potential gap between transposition and implementation is widely acknowledged (Falkner and Treib 2008; Zhelyazkova et al. 2016). Theoretically, different lines of inquiry have roughly followed the division between management and enforcement approaches (Tallberg 2002; Mastenbroek 2005). While there has been a steady stream of work showing the importance of administrative capacity, the capacity approach cannot easily explain differences between sectors in the same member state or even between issues in the same policy area (Haverland et al. 2011). Therefore, we expect actor-driven explanations, including political and administrative actors involved in transposition and implementation at different levels, to provide better insights into existing variation in implementation.

Aiming to explain situations when compliance varies between sectors in the same country or even in the same policy sector, we explore a three-level model of political–administrative interactions. Previous work on cultural heritage policy in Bulgaria (Dimitrova and Steunenberg 2013) proposed a model explaining the existence of different outcomes from the adoption of one and the same formal European policy as a result of the interplay of actors at the EU, national political and national administrative levels. This article departs from this model and develops it further as a three-level interaction, including Commission oversight. We explore outcomes under a wide range of domestic preferences and different regimes of European supervision. Empirically, we present research from the implementation of the EU policy on moveable cultural heritage policy. We examine the state of the policy in four EU member states. The selection of the cases of France, Lithuania, The Netherlands and the United Kingdom (UK) aims to represent different types of policy: conservationists (France, Lithuania) versus market oriented countries (The Netherlands, the UK). The countries selected also vary in size and centralization and represent old and new(er) member states.

The case descriptions are based on various sources of data including government documents, evaluations and parliamentary debates. In addition, the analysis draws on a student capstone project involving studying cultural heritage policy in depth and also using secondary sources supplemented with interviews with senior administrators or experts.²

The EU has recently revised the key piece of legislation in the area of movable cultural goods, Directive 93/7/EEC, recast by Directive 2014/60/EU. One of the main reasons to do so has been poor implementation: in its fourth report on the implementation of the legislation, the European Commission characterized ‘the illegal trafficking of cultural objects’ as ‘a scourge affecting the European Union’ (European Commission 2013: 9). The Commission pointed out the need to ensure the return of illegally removed national treasures by improving administrative co-operation and consultation between the authorities ‘which carry out the tasks provided for in the directive’ (European Commission 2013: 9). The clear importance of the administrative layer of governance for achieving the EU-wide objectives of the policy makes EU moveable cultural heritage a suitable case for illustrating our three-level model of interaction.

2. A framework for understanding domestic policy implementation

The study of transposition of EU directives in the national legislations of the member states has become part of a very extensive literature (Mastenbroek 2005; Toshkov et al. 2010; Treib 2014). A long-standing weakness of this literature has been the predominance of transposition studies taking the adoption of (directives) into national legislation as a measure for compliance. Nevertheless, implementation studies have also grown in the last decade, often focused on a country or group of countries or alternatively on a policy area in several countries. Many of the existing case studies find poor implementation is rooted in low administrative capacity (Toshkov et al. 2010). Yet general levels of administrative capacity and government effectiveness do not account for existing variation in both transposition and implementation between sectors in the same country (Haverland et al. 2011; Zhelyazkova et al. 2016).

Some recent studies build on previous actor-driven models (Steunenberg 2007), but also incorporate capacity as a factor for implementation. Investigating the transposition and implementation of the cross-border health care directive, Vasev and Vrangbæk (2014) combine actor- and capacity-driven approaches by focusing on the role of bureaucrats and the available policy resources. They stress the independent role played by domestic administrative actors, as shown earlier in Dimitrova and Steunenberg (2013), but also find that bureaucrats take into account resource availability in choosing transposition and implementation strategies. In addition, a broad comparative study by Zhelyazkova et al. (2016) investigating implementation in four policy areas has shown that policy area characteristics matter more than country characteristics for implementation.

While recognizing the importance of structural factors, both administrative (Börzel et al. 2010) and cultural or political (Falkner et al. 2005), we seek to explain implementation variation occurring within the same country and even within the same sector. To do this, we develop a theoretical model of the interaction between actors at three different levels – the European, national and administrative levels – in one policy sector.

Lipsky’s (1980) insights that street-level bureaucrats may have discretion to change policies are difficult to integrate systematically in implementation studies. A different theoretical stream, the principal–agent literature, views bureaucratic discretion as a result from expertise or structural features of the decision-making process. While the classical literature sees discretion as stemming from information asymmetry in favour of the agent, later studies highlight the structural difficulties of multiple principals to agree on a course of actions concerning agent performance (see, for instance, Eskridge and Ferejohn 1992).

Inspired by this work, we model the interactions between domestic political and administrative actors involved in transposing and implementing European policy as a game. We define the implementation of an EU policy as a sequence of stages starting with the adoption of a domestic formal policy (legal transposition) by national political (parliaments) or administrative actors (ministries). After transposition, actual implementation takes place. Administrative actors with a responsibility for implementation shape the newly adopted formal policy so that it fits to practices on the ground.³ These administrative actors are mid-level state officials and civil servants who, similarly to veto players, are in an organizational position which makes them key figures for policy implementation.

Views and understandings emerge in the public domain as discourses, which we conceptualize here as coherent sets of statements or expressions of worldviews defining game forms as well as preferences of actors. Preferences are expressions of desirability; in our case, the kinds of policies actors prefer. To describe preferences, we rely on the tools of positive political analysis (see Shepsle 2010) and assume that each player – a policy-maker or an implementing authority in our framework – has a most preferred position on policy, which is called an ideal point. Moreover, a player’s preference for an alternative policy depends on the distance from their ideal point. The farther away an alternative policy is from a player’s ideal point, the less preferred this policy. Finally, we model the policy area as a one-dimensional outcome space, representing its key feature of sectoral/issue orientation.

Our proposed framework takes into account European supervision of the domestic implementation process, reflecting the multilevelness of the European policy process. We include the European Commission as an enforcer; that is, an external agency with a certain capacity to impose the European policy on domestic actors.⁴ When enforcing a European policy, the external agency faces transaction costs. These costs are a function of possible ambiguities in the interpretation of European law, information asymmetries, or capacity limitations. As a consequence, we expect that the enforcer will not challenge every deviation from a European policy, but only those deviations that go beyond the limits set by these costs. These values define a set of sustainable proposals, policies that will not be challenged by the enforcer. We label this set the enforcer’s costs-induced indifference set. Furthermore, we assume that the enforcer aims to ‘protect’ the European legislative status quo and therefore has an ideal position that equals the policy embedded in European law.⁵

Domestic policy-makers decide on policy in anticipation of the enforcer’s response. We make a distinction between the legal or formal policy and the actually implemented or informal policy (Dimitrova and Steunenberg 2013: 250), which is relevant with regard to the role of the enforcer. The European Commission is quite well informed about the legal transposition of a directive from the obligatory notifications of domestic legislative measures. This is not necessarily the case with the actual implementation of policy. In this article, we assume variation in the extent to which the enforcer is able to observe policy and monitor implementation. There are two possibilities, which we label as ‘strong’ and ‘weak’ supervision. In the case of ‘strong’ supervision, we assume that enforcer is aware of both the formal and informal policy. However, there are many reasons, first and foremost its own capacity limitations, to expect that the Commission is not aware of informal policy practices. To expect the Commission would be able to monitor all policies in all twenty-eight member states at the level of actual implementation is unrealistic. It is, however, quite plausible that the Commission is aware of problem member states or policy areas where non-compliance is brought to its attention by non-state actors. In other words, there are possibilities for both strong and weak supervision depending on the policy field and member state in question. Under ‘weak’ supervision we assume that the Commission only notes the formal policy, which is set during transposition, but remains unaware of the informal, implementing policy.

We model the decision-making process as a game with two main stages, which include decision-making on the domestic policy and subsequently the evaluation of this policy by domestic and European actors. The first stage includes the well-known steps of transposition, leading to the setting of a formal policy, and subsequently implementation.

In the first stage, domestic policy-makers and the implementing actor(s) decide on the domestic policy. The domestic policy-makers set a legal or formal policy, which transposes a European directive into the national legal order. In making this decision, each player needs to approve the policy, as represented by the veto power of each participant. Furthermore, one of these players acts as an agenda-setter making proposals to the others. The agenda-setter can be a ministry that has the lead in the transposition process, or another actor that has special responsibility for the domestic transposition of a directive. Subsequently, the implementing player (‘implementer’) decides how the formal policy will be implemented. Since the implementing player may have rather different views on policy, he/she may change the policy during implementation. This difference between the legal or formal policy and the implemented or informal policy is also known as ‘bureaucratic drift’ (McCubbins et al. 1987).

In the second stage, the Commission and the domestic policy-makers review domestic policies (evaluation stage). Based on this review, these actors can decide to reverse the policy either to the formal policy (domestic policy-makers) or the European policy (Commission). The domestic policy-makers can change the informal policy by issuing more specific domestic legislation, replacing the leadership of the implementing organization, or changing the institutional arrangement under which the implementing organization makes its decisions (for instance, by incorporating an oversight body). The Commission can use ‘naming and shaming’ strategies (Börzel 2003) or, ultimately, an infringement procedure to change the formal as well as the informal policy (Steunenberg 2010). The Commission can also initiate new, amended legislation. Importantly, these possibilities constrain the domestic actors in making their policy choice. In this way the domestic decision-making process on policy is ‘nested’ in a broader, European compliance game.

3. Political–administrative conditions affecting compliance

According to this model, implementation can take different routes depending on the European policy (e.g., status quo), enforcement costs faced by the Commission (as reflected by its costs-induced indifference set to the status quo), and the preference of implementing player vis-à-vis the European policy and the domestic player with agenda-setting power, respectively. To summarize the preferences of the domestic policy-makers we use the concept of unanimity set: these are all points domestic players cannot change by unanimity (e.g., not casting a veto). The points that are elements of this set are located between the extreme members of these groups; that is, the left-most and right-most domestic veto players along a single dimension.

The policy states resulting from various combinations of European status quo points, enforcement costs and preferences are summarized in Table 1, while the equilibriums supporting the policy outcomes are described in the Online Appendix. In this table we denote the European policy as q (e.g., status quo), the formal policy as p_F, and the informal policy as p_I. Finally, in the model and in the table we distinguish between ‘strong’ and ‘weak’ oversight by the Commission.

Table 1.  Outcomes of the implementation games.

Location of the European policy:	In domestic unanimity set		Not in domestic unanimity set
Location of the European enforcer’s indifference set:	Overlaps or is (partially) included in domestic unanimity set		Overlaps domestic unanimity set			Does not overlap domestic unanimity set
Implementer’s preference vis-à-vis:	European policy		Domestic agenda setter
	Coinciding preferences	Non-coinciding preferences	Coinciding preferences: Centric	Coinciding preferences: non-centric	Non-coinciding preferences
Policy state:	1	2	3a	3b	3c	4
	Incidental domestic compliance	Formal domestic compliance	Incidental domestic adaptation	Constrained domestic adaptation	Domestic adaptation	Imposed formal compliance
‘Strong’ oversight	pF = pI = q	pF = q and pF ≠ pI	pF = pI	pF = pI	pF ≠ q and pF ≠ pI	pF = pI
‘Weak’ oversight	pF = pI = q	pF = q and pF ≠ pI	pF = pI	pF ≠ pI	pF ≠ q and pF ≠ pI	pF ≠ pI

Note: See the Online Appendix for further details on the various policy states and the conditions under which they occur (including Table A-1); the grey cells for the game with ‘weak’ European oversight (last row) indicates outcomes that differ from those found for the model with ‘strong’ oversight. These differences are more choice options for the implementing actor (light grey) or new choice options for this player (darker grey).

Based on the model’s analysis of the interplay of the three levels of actors, we can identify four different main types of domestic responses to a European policy. These responses vary between full alignment with the European policy to decoupling, implementing a different domestic informal policy while symbolically adhering to the European formal policy. More specifically, the main types are:

1. Incidental domestic compliance (see also Dimitrova and Steunenberg 2013: 251): This policy state is an undisputed and straightforward implementation of the European policy, which according to our model only occurs under rather strict and specific conditions. Only when the implementer likes the European policy and this policy falls in the domestic unanimity set, it will be precisely implemented.

2. Formal domestic compliance: This policy state is similar to the previous one, but the implementing player prefers a different informal policy than the European one. While the domestic players cannot move the formal policy away from the European one and are forced to transpose literally the European directive (Steunenberg 2007: 30), the implementing player chooses a deviating informal policy better fitting to his/her preferences. This informal policy is located in the domestic unanimity set to avoid domestic challenge during the evaluation stage. In case of ‘strong’ oversight, the policy is also located in the (partially) overlapping indifference set of the Commission. In this policy state, the deviation between the formal and informal policy is the result of inertia of the domestic veto players.

3. Domestic adaptation, and incidental and constrained domestic adaptation: When the European policy is not in the domestic unanimity set, the domestic players have an incentive to shift the European policy closer to their preferences. If the Commission’s costs-induced indifference set overlaps with the domestic unanimity set, such possibilities exist. In that case, domestic adaption is the cause of differences between the domestic and European policy. How far the policy is shifted depends on the enforcer’s indifference set, since the domestic policy-makers do not want to be challenged by the Commission. Furthermore, the formal policy is shaped by the preferences of the domestic player with agenda-setting power, who will set a policy as close as possible to their own ideal position. Differences in preferences between the agenda-setter and the implementer – and the fact that the domestic players cannot reverse the implementer’s choice if the policy is in the unanimity set – cause differences between the formal and informal policy.

   Domestic adaptation includes two special cases. These occur: (a) when the agenda-setter and the implementer share the same, rather centric preferences (e.g., preferences in the overlapping part of the unanimity and indifference sets), leading to the same formal and informal policies (incidental domestic adaptation); and, (b) when the agenda-setter and the implementer have both more conservative, non-centric preferences (e.g., preferences that do not fall into the overlapping sets), they select the same policies, at least under ‘strong’ Commission supervision (constrained domestic adaptation). These specific cases are basically a ‘lucky’ incidence of coincident preferences and a corner solution with regard to Commission oversight.

4. Imposed formal compliance: In this last policy state, the European policy (and the Commission’s interpretation of it) differs substantially from the preferences of the domestic players. The Commission’s indifference set does not overlap the domestic unanimity set. Consequently, and in order to avoid being challenged, the domestic players have to adapt their formal policy towards the European one. Their best choice is a formal policy equal to the lower limit of the Commission’s indifference set (e.g. p_F = t_l). The implementing player is forced to implement the same policy under ‘strong’ supervision, otherwise it risks being challenged. Under ‘weak’ supervision, however, the implementing player has more discretion, as the Commission does not monitor its policy. In that case, the implementer can choose any policy from the domestic unanimity set, as domestic players will not be able to reverse this choice, creating a substantial difference between the informal and formal policy.

This analysis has several major implications. To start with, it shows that an implementing policy that reflects the formal policy as well as the European one (e.g. p_F = p_I = q) is just one possible outcome among several others. This outcome of having the same European, formal and implementing policy is found only when preferences are aligned (see Table 1).

Another important insight following from the analysis is that the discretion of administrative actors – our implementers – varies according to the level of EU supervision and the size of the domestic unanimity set. Domestically, the unanimity set is important, since it defines the extent to which domestic policymakers can affect the implementing policy carried out by the administration. The more the views of politicians diverge, the wider this set is, allowing the administration more discretion in setting the actual (informal) policy. Conversely, the more similar the preferences of politicians, the smaller the unanimity set, leading to a more restrictive informal policy choice. In the latter case, the administration cannot diverge much from the national political leadership. Finding that administrative actors have more discretion when political players are strongly divided corresponds with the conclusions of analyses of deadlocked political systems (see Crozier 1964; Kesselman 1970).

To explore and illustrate these findings, we research the implementation of moveable cultural heritage policy in the EU. Having been recently recast by the EU, the main policy measures and their implementation have been subject of considerable attention by the Commission, member states and European Parliament, and present ample opportunities to explore the differences between formal policy and implementation and the divergence between political and administrative actors.

4. European cultural heritage policy

European Union policies on cultural heritage were initially defined in relation to the freedoms at the core of the EU’s Internal Market, until a revision in 2013–2014 which shifted the emphasis to national cultural heritage and its economic importance. Article 36 of the Treaty on the Functioning of the European Union (TFEU) provides the possibility of member states to restrict the free movement of goods in order to protect national treasures of artistic, historical or archaeological value. After the abolition of border controls in 1993, the policy needed to be fleshed out to ensure the protection of national cultural treasures.

The policy was shaped by secondary legislation, in the form of Directive 93/7/EEC and Regulation 3911/92 (later repealed by Regulation 116/2009) and their Annexes specifying arrangements for the return of national treasures unlawfully removed from a member state. The policy adopted in 1992–1993 was restrictive, detailed and specific, as member states feared being overwhelmed by complains by individuals seeking the return of cultural artefacts (European Parliament 2014: 22). The key elements of the policy rested on the specification of the type of cultural objects that could be classified as national treasures (in terms of age and monetary value) and administrative co-operation to ensure the return of unlawfully exported national treasures.

The transposition of Directive 93/7/EEC by member states was late and only completed by 1999. According to the Commission, a number of member states transposed incorrectly or incompletely. For example, a number of national laws transposing Directive 93/7/EEC failed to mention the obligation to provide information and the general principle of co-operation between the member states (European Commission 2000: 13). As stipulated in Directive 93/7/EEC, the Commission and the member states created a system for evaluating the application of the regulation and the directive under which four reports have been published, covering implementation problems respectively for the periods 1994–1999, 1999–2003, 2004–2007 and 2008–2011. The implementation reports draw on member states’ responses to detailed questionnaires and their own assessments. They highlight considerable differences in implementation between the member states.

From our perspective, the most important finding from the Commission reports and from the European Parliament report (2014) is that member states and their administrations had very diverse preferences, growing further apart with the 1996 and 2004/2007 enlargements (European Commission 2000, 2005, 2009; European Parliament 2014). In contrast to the UK and France, which had, by the time of negotiation, well-developed, if different, policies of evaluating cultural artefacts and establishing market value, a number of member states that acceded in 1996 and 2004 did not have such legislation at all (European Commission 2000, 2009). Formal domestic policies on EU cultural heritage policy range from policies to facilitate trade with cultural goods to policies primarily targeting the preservation of national heritage.

Taking these differences into account, based on the Commission reports we distinguish between different groups of member states, which we label the ‘traders’ and the ‘conservationists’. The traders have substantial art markets and handle cultural artefacts from the whole of the EU and further afield. Member states from this group, like UK and the Netherlands, had pre-existing legislation and policy aiming to facilitate art trade (European Commission 2000: 11–12). The conservationists are primarily concerned with the preservation of their own cultural heritage. Countries belonging to this group had more restrictive legislation, aiming to control exports. Countries in this group include France and Lithuania, examined below, but also Italy, Greece and, after 2007, Bulgaria (Dimitrova and Steunenberg 2013).

In its policy evaluation, the Commission did not choose between the different groups of countries and their positions, but it ultimately went along with some of the conservationists in revising Directive 93/7/EEC. Cultural heritage does not appear to be an area in which the Commission proactively seeks to make an impact, given that the policy is not given priority in the political programme of the Commission (Juncker 2014). Despite the reported differences in implementation, there are no infringement cases seeking change in domestic policy (except for non-transposition). Therefore, we label the Commission’s oversight as ‘weak’ based on the definition in our model.

The following paragraphs examine policy preferences and implementing actors in each of our four cases.

4.1. The United Kingdom

The UK is an important centre for the international art market and British policy aims to strike a balance between the protection of national heritage, the rights of owners selling goods and of purchasers, and the reputation of the UK (Netten 2014: 37–8).

The UK transposed Directive 93/7/EEC very early with the return of cultural objects regulations from 1994. Similarities between the Directive and the UK legislation suggest the country was able to upload its approach to the EU level to a certain extent. The UK law is almost identical to the directive and has had only two amendments, in 1997 and 2001, following EU-level amendments. However, the UK had a formal policy in this area long before the EU legislation. The ‘Export Control Act 2000′ from 2002 covered all exports from the UK and not only exports to the EU. Therefore, the law is supplemented by an Order of the Secretary of State on the ‘export of objects of cultural interest’ in 2003. The ‘European license’ issued under the Council regulation 3911/92 is only one of four different export licenses mentioned in the Order.

The institutions carrying out implementation are the Export Licensing Unit at the Arts Council and the Reviewing Committee on the Export of Works of Art and Objects of Cultural Interests (RCEWA). When deciding whether a cultural object is a national treasure, the Art Council consults experts that refer to the criteria developed by the Waverly committee in 1952. The Waverly criteria define historical, aesthetic and scientific reasons why a cultural object might be a national treasure. While the initial export license applications are handled by experts hired by the Arts Council, the Reviewing Committee mediates in case of disagreement and balances the interests between owners, national heritage considerations and the market (Netten 2014).

The UK provides clear information to all involved of the relevant criteria for defining a cultural good as a national treasure. The UK also maintains a register of objects with a ‘temporary export deferral’. The key implementing actor, the Reviewing Committee produces clear and exhaustive annuals reports explaining policy practice. It has examined a total of 207 cases between 2003 and 2013. The UK issues by far the largest number of export licenses in the EU under Regulation 3911/92, a total of 38,445 licenses for the period 1993–1998, far more than the second largest issuer, France, with 8,338 licenses (European Commission 2000: 19).

The examination of administrative procedures and transparency in information provision suggests UK informal policy does not deviate from the formal or European policy, except that the UK aims to cover export authorizations for the whole world and not only the EU. National legislation neatly fits to the requirements of the European legislation, accommodating the needs of the international art trade. We can classify the policy state in the UK as incidental domestic compliance.

4.2. The Netherlands

Before the adoption of the European framework, cultural heritage protection in the Netherlands was based on a law adopted in 1984. Based on this law, a short list of items, considered ‘essential’ and ‘irreplaceable’ for Dutch cultural heritage, was composed. This list is short, partly because cultural objects owned by government (e.g., museums) are not included, but it also illustrates the liberal stance of the Netherlands with regard to cultural goods.

During the preparation of the European policy, the Netherlands, like the UK, was able to ‘upload’ part of their approach (Houwing 2015). Directive 93/7 was transposed in a rather direct manner without additional conditions on the return of cultural objects. The Dutch did not add any further licensing of cultural goods next to the European rules. Only when cultural goods fall within the categories defined by European law is a permit needed.

Analysing the positions of the main political parties, especially those in government, shows that they supported the existing rules. There was not much of a debate on the introduction of the new legislation. The media published only one critical article in which the Dutch Association for Art Trade expressed concerns that the new rules might hinder the art trade (Houwing 2015: 47).

Implementation was delegated to the Cultural Heritage Inspectorate, which is an agency directly under the responsibility of the minister of education, culture and science. Similar to other countries, Customs play an important role as the front office in checking and detecting illegal export and import of objects. Owing to the centralization of the licensing process, in which also the minister may object to the export of an important work of art, the variation in preferences is rather limited. Based on this preference configuration, we classify this case as incidental domestic compliance (Houwing 2015: 49).

4.3. France

In France there was already a cultural heritage policy in place before the discussion and adoption of the European policy in 1992. This policy was based on a law adopted in 1941 regulating the protection of works of art in France. This law resembles the EU policy, especially with regard to the grounds to refuse export. Not surprisingly, the French managed to swiftly transpose Directive 93/7 and embedded the implementation of the European legislative framework in the already existing structure under the Ministry of Culture and Communication.

While most of the requirements were transposed along the lines of the directive, the French adopted a slightly broader definition of cultural goods. In addition, they introduced by decree a detailed categorization of various goods to the law, a continuation of their national policy since 1993. Finally, the French authorities added various classifications and procedures to their policy, including the requirement for cultural goods to have a certificate when they leave France (Weidema 2014: 38–41).

The European policy was not contested in France, as it merged easily with the existing French policy. The national formal policy can be characterized as rather protective, rooted in the strong emphasis in French politics on presenting and preserving cultural heritage (Loosely 2003: 228). For implementation, France employs a centralized system of licensing. This policy fits with the centralized political system and allows for minimal differences between the main stakeholders in their view on cultural heritage. Furthermore, the centralization of implementation, in which the assessments of the committee of experts on the exportability of cultural goods is merely consultative, limits the possibility of deviation (Weidema 2014: 49). Based on this preference configuration, we would classify the French implementation as a case of incidental domestic adaptation.

4.4. Lithuania

Lithuania had prepared to adopt the EU policy framework before its accession to the EU, in 1996, by adopting the Law on the Protection of Moveable Cultural Property. This law defines cultural objects and the institutions involved in their assessment and protection. The definition of cultural objects is more extensive than in the EU policy, including, for example, ‘all moveable creations created 50 years ago and earlier’ (Klein 2014: 31). The broad categories of the law suggest a more restrictive policy than the European law. In the year of accession, Parliament adopted Resolutions 1107 and 1424 defining export procedures. These resolutions refer to both the EU Council Regulation and the definitions adopted in the 1996 domestic law.

The preferences of Lithuanian policy-makers can be established based on party manifestos of the main political parties entering parliament in the 2012 elections. While not all parties mention culture in their manifestos, the Social Democratic Party specifically referred to the need to modernize heritage management programmes with support from the EUs Structural Funds. They also stated the need to protect cultural heritage from ‘excessively liberalized and chaotic administration’, which can be read as opposition to the existing EU policy (Klein 2014: 40). One of the main opposition parties, the Homeland Union, also pointed at the need for more funds and argued that the policy should be redesigned and updated. By contrast, another opposition party, the Liberal Movement, the fourth biggest after the 2012 elections, declared to seek a balance between heritage protection and business interests, placing itself on the more liberal, market-oriented part of the spectrum.

The administrative actor responsible for implementation is central government’s Cultural Property Export Control Division. This division, which is part of the Department of Cultural Heritage Protection within the Ministry of Culture, issues export permits and investigates cultural objects found by the police, customs or others (Arulienė 2001). Interestingly, the division’s work falls under the National Cultural Heritage Commission. This broad expert committee, which is in charge of the Lithuania’s culture policy, reports directly to parliament and the executive.

The implementers’ preferences appear to be focused on the 1996 law, defining broad protection of Lithuanian cultural goods. The EU policy is assessed as appropriate, but seems to be regarded as a supplement to the already used national criteria and licensing. The most heated debates surrounded the question of whether the Department of Cultural Heritage Protection should be responsible not only for registering but also for maintaining cultural objects, and especially those defined as national treasures. A church fire, which destroyed an unknown number of objects in 2012, brought this issue into focus (Klein 2014: 39–40). The main subject of controversy in the debate was whether the Church or the state, represented by the Department of Cultural Heritage Protection, were responsible for registering, maintaining and protecting cultural objects located on church property. The interviewed expert stated that for the Department of Cultural Heritage the problem of maintenance of registered cultural objects was primarily one of resources (Klein 2014: 45).

Even based only on the positions of the parties that expressed a clear preference, the unanimity set in Lithuania is quite broad and preferences do not coincide with the European policy. With an implementing agency following a more protective stance, we classify Lithuania’s preference configuration as domestic adaptation.

4.5. Comparison

Table 2 provides an overview of the domestic positions and the predicted policy based on our model. The table also presents the actual policies in the four cases: the formal, legislative policy and the informal, implemented policy. The implemented policies in the UK and the Netherlands are very much in line with the European policy. As traders, both countries have more market-oriented procedures and fewer restrictive mechanisms for export, fitting to the European policy. France and Lithuania, however, have more protective national policies. While in France the domestic unanimity set is rather small, including an implementing department within a line-ministry, it is wide in Lithuania, where the implementing division is closely connected to an expert council protecting cultural heritage. Both France and Lithuania use additional criteria, but in Lithuania implementation follows primarily national rather than European criteria (Klein 2014: 42). The national policies seem to correspond with predictions we made based on our model. The most important factor that influences implementation appears to be the size of the national unanimity set. In combination with weak Commission monitoring, this provides opportunities for different implementation responses.

Table 2.  Positions of the different member states and prediction under ‘weak’ Commission oversight.

	Unanimity set versus European policy	Implementing agency	Predicted policy	Formal (legislative) policy (pF)	Informal (implemented) policy (pI)
UK	Small set favouring trade	Decentralized (e.g., experts versus RCEWA): established assessment criteria (coinciding national preferences)	(1) Incidental domestic compliance pF = pI = q	More liberal national policy	Aligned with national formal policy (‘Waverly criteria’) and EU rules
The Netherlands	Small set favouring trade	Centralized (e.g., ministry): established criteria within the EU framework (coinciding national preferences)	(1) Incidental domestic compliance pF = pI = q	More liberal national policy	Aligned with EU law
France	Small set favouring preservation	Centralized (e.g., ministry): protective policy (coinciding national preferences)	(3a) Incidental domestic adaptation pF = pI	More protective national policy	Aligned with national formal policy
Lithuania	Wide set favouring trade and preservation	Centralized (e.g., division in ministry): protective policy (non-coinciding preferences)	(3c) Domestic adaptation pF ≠ q and pF ≠ pI	More protective national policy	EU policy embedded in national priorities

5. Conclusions

The model and analyses presented here aim to contribute to the literature dealing with actual policy implementation (Börzel et al. 2010; Di Lucia and Kronsell 2010; Falkner et al. 2005; Martinsen and Vollaard 2014; Vasev and Vrangbæk 2014; Versluis 2007; Zhelyazkova et al. 2016). The model focuses on single policy sectors, and its approach fits well with recent findings focusing on the importance of state bureaucrats as implementers of EU policy (Vasev and Vrangbaek 2014).

We see compliance or lack thereof as a consequence of the diverse preferences of a broad group of political and administrative actors and their complex and iterative actions at several different levels. Political-level agreement is important to institute a formal policy, but to have it implemented, administrative authorities across the EU need to be aware of each others’ roles and to co-ordinate their actions. Even then, their own preferences may lead them to make a different policy on the ground when domestic politicians are divided.

Our model shows that for informal policy to be equivalent to the domestic formal and European policy, rather specific circumstances are required. If political and administrative actors at the European and domestic level have different views, other policy states are more likely. In these policy states, the domestic policy may differ from the European policy, owing to the limitations of Commission enforcement.

The model’s other substantial innovation is to differentiate between domestic implementer’s behaviour under weak and strong enforcement. There are member states and policy areas where the Commission may focus its attention on the informal policy, because of policy priorities or expectation of problems. Such member states would not be in a position to depart from the European policy, even if domestic actors would have different preferences, leading to a policy state of imposed formal compliance. Other member states function under weaker control allowing for domestic adaptation of the European policy.

In our empirical analysis we have found rather different domestic policies in several member states. The distance between the EU policy and the preferences of national actors – political and administrative – was sometimes quite large. This diversity in national implementation has an impact on the European policy as a whole. Only when national implementers co-ordinate their efforts in satisfying the main policy requirements (e.g., maintaining a register for moveable cultural goods of national significance and a proper procedure identifying these goods), implementation can be successful. This points to an implementation problem that is not related to the individual country performance, but to the joined performance of national administrations. Owing to the lack of sufficient policy co-ordination – the different national implementing policies – implementation was unsuccessful at the EU level and the actual effects from the existing European policy were minimal. The low number of return proceedings led to a general realization that Directive 93/7/EEC did not achieve its goal and subsequent recasting of the Directive. The new instrument, Directive 2014/60/EU, amended previous legislation in a substantial way to drop financial thresholds for cultural goods classified as national treasures. The new Directive and the Commission Communication (2014a) and Mapping report (Commission 2014b) indicate the way to meaningful implementation is through co-operation of the national administrative authorities. This formal and explicit recognition of the role of administrative actors is in line with the theoretical arguments advanced in this article.

Additional factors that emerged as important for implementation based on the case study analyses were the capacity of specific departments tasked with implementation in terms of manpower and expertise (in the UK) and the resources available for the implementation of the policy (in Lithuania). This finding suggests that actor-driven models should take resources into account, as Vasev and Vrangbaek (2014) have done. However, it is still unclear based on the examined cases whether the resources that matter are the means assigned to a policy area based on government ideology or the resources available to the administration in general. Some light on this is shed by Zhelyazkova et al.’s (2016) findings suggesting that good implementation is much less likely when the specificity of the policy area (part of the welfare state or not) means it requires substantial investment. In the case of cultural heritage policy, we find that a level of harmonization between member states required some investment in a register of moveable cultural objects, but not of a comparable magnitude with social policy.

Based on differences in national policy preferences and European supervision, our model predicts that a diverse and fragmentized pattern of policy implementation is quite likely. With regard to moveable cultural heritage and the return of stolen national treasures, the EU is trying to address this problem by building common knowledge and sharing implementing practices. Awareness and co-ordination between administrative actors, which the EU has called for and embedded in its recast Directive, however, may not be enough. This is especially relevant when powerful domestic discourses – for example linked to national history and identity – shape the preferences of implementing actors. Therefore, we can expect common implementation practices to remain elusive, at least until co-ordination of administrative actions is supported by EU-wide debates and shared discourses about the role of national cultural heritage in a European context.

Acknowledgements

We would like to thank Asta Ivanauskaite, Robert Goldmann, Jorrit Klein, Felix Netten and Marloes Weidema for research assistance with the country cases of France, Germany, Lithuania, the Netherlands and the UK. We also thank Klaudijus Maniokas, Maria Sturm and a number of country experts for their advice and expert opinions, as well as participants at the 2014 Annual Conference of the American Political Science Association in Washington and this journal’s referees for their comments.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes on contributors

Antoaneta Dimitrova is associate professor at the Institute of Public Administration, Leiden University, The Hague, The Netherlands.

Bernard Steunenberg is professor at the Institute of Public Administration, Leiden University, The Hague, The Netherlands.

Notes

1. Compliance is defined here as the extent to which domestic actors – policy-makers or administrators – produce the same policy outcomes as formulated in European policy.

2. The results of the individual capstone projects are reported in Houwing (2015), Klein (2014), Netten (2014) and Weidema (2014). As part of these projects and in addition to documentary sources, six expert interviews have been conducted: two in the Netherlands, Lithuania and the UK.

3. If we broadly define a policy as a set of (formal and informal) rules and practices aiming to achieve a certain objective with regard to a particular issue or a sector, the formal policy is defined by (primary and secondary) legislation and is enforceable by third parties. The informal policy consists of the rules that are used in practice and the way actors actually apply them.

4. The Commission sometimes needs to involve other actors, chiefly the European Court of Justice, to achieve compliance. Notwithstanding the Court's importance, we disregard these interactions, since we focus on the domestic implementation process.

5. Of course, this assumption can be relaxed. Steunenberg (2010) has shown that a Commission supporting higher levels of European regulation than domestic actors will act as ‘guardian of the treaties'. In case of a preference for lower levels than domestic actors, the Commission will support deviation by not acting against it ('silent witness'). In the area of cultural heritage, as we will discuss later, it seems that the Commission preferred a higher level of harmonization, which it had to water down during the revision of Directive 93/7/EEC.