Paper prepared for the 1rst International Conference on Public Policy (ICPP) June 2013, Grenoble, France

Transcription

1 Similar domestic solutions to transboundary problems? The customization of EU directives of veterinary drugs during transposition in four Western European countries Lic. phil. Eva Thomann Center of Competence for Public Management at the University of Berne Paper prepared for the 1rst International Conference on Public Policy (ICPP) June 2013, Grenoble, France Panel on International Organizations and Domestic Public Policy Chairs: Carsten Daugbjerg and Grace Skogstad -- WORK IN PROGRESS - COMMENTS ARE WELCOME Abstract This paper seeks to explain the degree to which member states use the leeway left by EU directives during transposition. It provides an unprecedented overview of the domestic settings and the transposition of veterinary drugs policies in Austria, France, Germany and the UK. The method of fuzzy-set Qualitative Comparative Analysis provides insights on the complex interaction of different policy and domestic factors with each other: rather than being mutually exclusive, the top-down and bottom up perspectives, the logics of consequence and of appropriateness complement each other. The responsiveness of the EU governance mode, as well as the salience of the issue, structure subsequent decisions at the national level. The resistance of domestic actors is an important reason why EU policies are customized, depending on the way in which the institutional structure effectively shelters the decision-making body from societal demands. By conceiving the closeness of domestic policies to EU norms as an outcome of the transposition process, this paper sheds a different light on the goodness of fit thesis: the countries reinterpret the EU norm if there is a cultural misfit with their local policy style.

2 1 Introduction In a world of globalised markets and increasingly transboundary challenges, the question of whether or not countries resolve their problems in similar ways is highly relevant. The European Union (EU) has the objective of providing common solutions to shared problems (Versluis 2007). Yet, the EU steers, but does not row: its directives only represent binding minimum standards which are subsequently transposed into domestic law. 1 They often leave considerable leeway for adaptation to the implementing countries, enabling the latter to be considerate of the needs and particularities prevailing at the domestic level. As a result, the decentralised, multi-level implementation structure of the EU leads to dissimilarities in the transposition of EU directives between countries (Falkner et al. 2005). EU compliance research has increasingly departed from the thesis that the institutional compatibility of European policies with domestic institutions and policies explains domestic responses. Instead, the importance of domestic factors for policy outcomes is stressed (Mastenbroek and Kaeding 2006). But in what ways does EU policy matter, as opposed to domestic factors? How do these factors interact with each other? What compliance research tends to neglect is the fact that the domestic regulations can differ notably even though compliance is given. In the light of these differences, I ask how far the influence of hard supranational policy instruments on domestic policy making actually reaches compared to domestic factors. Rather than addressing the often-asked question of compliance, this paper seeks to explain the degree to which countries customize EU directives during transposition. Thereby, the closeness of domestic regulations to EU templates is no longer an explanatory variable (Knill and Lehmkuhl 2002), but becomes the phenomenon which needs to be explained. At the turn of the millenium, several food scandals related to animal diseases have shed light on issues of food safety in the single market for food products. Due to their major role in securing food safety, the regulations of veterinary drugs for livestock have gained increased public attention and regulatory importance. Using qualitative data on veterinary drugs regulations in France, the United Kingdom (UK), Germany, and Austria (Sager et al. 2011), I 1 This distinguishes EU directives from EU regulations, which are also compulsory, but directly applicable at the national level; and from other, non-compulsory, soft modes of EU governance (Knill and Lenschow 2003).

3 compare the domestic regulations according to their closeness to the respective EU directive. A fuzzy-set Qualitative Comparative Analysis (fsqca) (Rihoux and Ragin 2009) is employed to assess whether particularities of the given policy aspect, on the one hand, or country characteristics, on the other hand, or combinations thereof, make countries stick more or less to the EU template. This paper is structured as follows: in the next section, the concept of customization is defined and confined (Sartori 1970). Subsequently, the findings of research on EU implementation and integration are discussed in order to derive conditions for customization. The empirical research design and the fsqca method are outlined in section four. Section five provides a brief description of the cases. The results are presented in section six, and section seven concludes. 2 Outcome: The customization of EU directives Even if there is disagreement about the sources of European integration, it is clearly taking place as such: national states increasingly transfer parts of their autonomy to a common institutional framework the EU - to allow for common rules or policies (Corbey 1995: ; Pierson 1996; Puchala 1999; Starke et al. 2008). One instance of this is the EU s activity as a regulatory state (Majone 1994): It promulgates regulations whilst pushing the costs of implementation onto the member states, with the aim of harmonising regulations intergovernmentally (Yeung 2010: 69-70). In this context, the transposition of EU directives can be conceived as a case of coercive policy transfer (Dolowitz and Marsh 2000). The outcome of the transposition process within the EU has so-far mainly been perceived in terms of the dichotomy between compliance or non-compliance (Mastenbroek 2005: 1114; for recent overviews see Benson and Jordan 2011; Falkner et al. 2005; Pülzl and Treib 2007). However, the remaining differences when EU directives are transposed have hitherto received little scholarly attention. One exception is Steunenberg (2007), who distinguishes the literal and non-literal interpretation of EU directives within the boundaries allowed by the European Commission (EC). In this line of reasoning, the concept of customization focuses on the closeness of domestic regulations to the EU template in the situation when a) an EU directive exists and b) 1

4 compliance is given. To what extent do countries use the leeway left by the binding EU minimal standards to adapt them to domestic particularities? The term customization refers to the degree to which the domestic regulations depart from or complement the EU directive, i.e., of direct and complete transfer at the stage of transposition ( copying ; Dolowitz and Marsh 2000: 13). Customization is measured by comparing the concrete regulations of EU directives with their domestic counterparts. The terms regulations or policies are used interchangeably to refer to what is commonly defined as public policy instrument, more specifically, a set of techniques by which public actors wield their power in attempting to ensure support and effect or prevent social change (Vedung 1998: 21). Since EU directives allow for more, but not less far-reaching domestic regulations (Falkner et al. 2005: 1), extensive customization often also implies a higher degree of regulation. But what does more far-reaching mean? Changes in policies can concern the regulatory style (density) or the policy content (intensity) (Knill and Lenschow 1998; Knill et al. 2012). Thus, it takes a distinction of, on the one hand, the density of the domestic regulations in terms of the amount to which the EU directive has been concretized by more rules (the level of detail of the national legislation according to Versluis [2003: 47]). On the other hand, the amount of domestic rules does not inform us about their actual restrictiveness relative to the EU template. Hence, a second dimension refers to the degree to which the action content of the domestic rule additionally restricts the freedom of action of the policy addressees. Customization differs from concepts such as emulation, lesson-drawing or diffusion since it refers to the non-voluntary transfer of policies. Processes, for instance the change of domestic institutions compared to their previous state (Radaelli 2003), are not analyzed; this would require a diachronic design (Heichel et al. 2005). Domestic regulations are not compared according to their convergence towards each other (Holzinger and Knill 2005), but to the respective EU template. It is, however, not correct to speak of degrees of compliance (Haverland 2000) since compliance is given as the starting point of the analysis. 3 Explanations for customization The impact of European policy-making differs largely across policies and countries and is contingent upon domestic constellations (Héritier et al. 2001). Consequently, the subsequent discussion follows the distinction of factors varying at the policy level and factors 2

5 varying at the national level (Mastenbroek 2005: 1113). It lends from existing findings of Europeanization research in order to derive the conditions for different degrees of customization. 3.1 Policy characteristics The misfit argument put forward by earlier waves of Europeanization research (e.g. Knill and Lehmkuhl 2002, Börzel 2002) has found little empirical support (Héritier et al. 2001; Mastenbroek 2005; Mastenbroek and Kaeding 2006). Yet, other arguments follow a similar top-down line of reasoning by assuming a direct causal link between European policies and domestic outcomes (Pülzl and Treib 2007) EU regulatory modes The prerequisite for high degrees of customization is the existence of leeway. An implementing actor has discretion whenever the effective limits on his power leave him free to make a choice among possible courses of action and inaction (Davis 1969: 4). Different modes of regulation have emerged in the EU, depending on the specific allocation of competencies within its multi-level governance system. However, insufficient attention has been paid to the ways in these governance modes are implemented at the memberstate level (Jordan, Wurzel and Zito 2005: 480). Knill and Lenschow (2003: 5-6) assume that different modes of regulation will results in different mechanisms of policy adaptation on the national level. Their typology, slightly adapted by Treib et al. (2007), is based on two criteria and results in four modes of intervention. First, the level of obligation refers to the degree to which the EU regulatory policy imposes legal authority on its addressees. Since EU directives are compulsory, the level of obligation is always high. However, second, these policies still differ according to the level of discretion they grant to decentralised actors in the implementation chain. We can distinguish regulatory standards which represent the most hierarchical mode of regulation, entailing obligatory and detailed rules in terms of legally binding standards. In turn, new instruments represent a more indirect approach to achieve behavioral change. They leave a broader leeway to comply within a relatively open regulatory space (framework regulation) 3

6 or to react to new procedural or incentive structures (economic and communicative instruments). Despite being legally binding, such a flexible approach to implementation can nevertheless entail soft law (e.g. by defining only broad goals to be specified by member states, or offering exemption and derogation possibilities or by allowing member states to choose from a variety of possible policy options) (Treib et al. 2007: 14-5). New instruments are more responsive to the context than regulatory standards: they grant implementers leeway in adjusting to regulatory requirements in light of different national and subnational problem constellations (Knill and Lenschow 2003: 9-10). Accordingly, I conceive new instruments as a more responsive regulatory mode under which extensive customization at the domestic level is more likely to occur, whereas unresponsive regulatory standards would imply no customization Issue salience Given that the number of EU directives is enormous, the amount of political attention received by an issue needs to be taken into consideration for understanding a country s transposition style. The assumption here is that the visibility of an issue conditions behaviour. Domestic actors decide how to distribute their attention, and less salient regulations are ignored (Knill and Lenschow 1998; Versluis 2003: 44ff., 227ff.). Issue salience refers to the visibility of and the importance attached to a topic. The main indicator of salience is public attention, e.g. via news coverage and discussions held in the public sphere. Two elements can crucially trigger attention: first, the perception of serious risks and their visibility in terms of their potential impact on a wider public. Second, focusing events, e.g. campaigns or dramatic events, push issues onto the agenda (Versluis 2004: 10ff.) The risk of non-enforcement is lower when an issue is salient (Versluis 2004). However, when it comes to transposition, and not enforcement, processes, Mastenbroek and Kaeding (2006: 341) point out that the actors concerned by a policy are more likely to bring their point of view if an issue is deemed more salient. Thus, a higher degree of customization can be expected with salient issues, especially if a policy is contested. 4

7 3.2 Domestic factors More recently, it has become obvious that the top-down approaches of EU implementation research laid too much emphasis on structural parameters (Pülzl and Treib 2007: 98) and are too static (Mastenbroek 2005). As a response, the question about the ways in which domestic politics and diverse national traditions and institutions shape the process of European integration has received increased scholarly attention. Some approaches assume a logic of consequence: the choices of the countries are determined by definitions of alternatives, consequences, preferences and strategic options which are strongly affected by the institutional context. Such approaches focus on institutional opportunity structures, actors and interests. Other explanations are based on the logic of appropriateness: action is taken to fulfil socially constructed and accepted identities, rules and practises. These approaches provide more cultural explanations (Mastenbroek and Kaeding 2006; March and Olsen 1996) Domestic politics The insight that we need to bring domestic politics back in to explain EU compliance (Falkner et al. 2005: 329) has led to a stronger theorization of and empirical support for the thesis that preferences and beliefs of domestic actors matter for implementation (Mastenbroek 2005; Treib 2003). The domestic players involved in EU implementation range from governmental departments, parliament, political parties and subnational entities to interest groups. EU policies may lead to political struggles between these actors. In a first line of reasoning based on rational choice institutionalism (Mastenbroek and Kaeding 2006), the outcome of transposition depends on the institutional arena in which decision-making takes place, and on the interests of the domestic actors involved. Policy-specific veto players can hamper the implementation if the directive is not in their interest. Empirically, domestic support (Knill and Lenschow 1998) or opposition (Falkner et al. 2005) is indeed a crucial explanatory factor for implementation in the EU. Thus, the opposition of domestic societal players should lead to high degrees of customization (Steunenberg 2007). Domestic resistance is given when the policy s main target group opposes the EU policy. It will be most effective when its principal 5

8 representatives have a significant potential to influence domestic policy-making Institutional veto points and players However, Haverland (2000) argues that it is not the amount of domestic opposition to EU policies which is decisive, but whether the policy-making body is effectively sheltered from societal demands. Historical institutionalist arguments (Hall and Taylor 1996) grant particular importance to variations in the national institutional opportunity structure for administrative, societal and political actors, which empower societal actors to exploit opportunities to pursue their interests (Mastenbroek 2005: 1111). Institutional veto points, defined as stages in the decision-making process on which agreement is legally required for a policy change (Tsebelis 1995), provide such opportunities. Crucial here is the dispersion of power (Mastenbroek and Kaeding 2006: 336): only in centralised states can the transposition be completed on the national level alone (Versluis 2003: 238). Consequently, the latter implement EU law more strictly than countries with a fragmented institutional structure. Policy-specific processes of consultation when the parliament is not involved are equally important (Steunenberg 2007). In systems with strong corporatist interest intermediation, more extensive customization can take place even in the absence of institutional veto points. Corporatist integration is defined by Siaroff (1999: 189) as a long-term co-operative pattern of shared economic management involving the social partners and existing at various levels such as ( ) joint shaping of national policies. Generally said, if the domestic opposition has effective veto points at which it can substantially delay the process or modify the outcome of the transposition process, modifications to its favour are likely to occur (Haverland 2000). A similar line of reasoning focuses on the capacity of institutional (formal veto power) and partisan veto players for policy change (Falkner et al. 2005: 296ff.; Treib 2003). A veto player is an individual or collective actor whose agreement is required for a policy decision, i.e. who can block the adoption of a policy (Tsebelis 1995: 293). As the number of veto players increases, decision-making processes become more difficult. Notwithstanding, formal veto players should only be relevant if the transposition process enters the formal legislative arena, which presupposes that it results in statutory law (Steunenberg 2007: 28); they may not always want to use their authority, depending on their goals, e.g. on the issue saliency (Mastenbroek and Kaeding 2006: 341). 6

9 3.2.3 National policy styles Arguments in the tradition of sociological institutionalism (Mastenbroek 2005) focus on preexisting norms and dominant belief systems in a country (Falkner et al ; Héritier et al. 2001). Based on the idea that the fit with domestic beliefs is more important than the fit with the status quo, local actors can reinterpret the overarching norm, so as to ensure that it fits their identities (Mastenbroek and Kaeding 2006: 345-7). Indeed, nations have distinct cultures of policy making (Haverland 2000: 84), compliance (Falkner et al. 2005) and enforcement (Versluis 2003). A large share of the transposition process is done via lower-level instruments (government decrees, ministerial orders) and never enters the formal political arena. Given this, the behaviour of administrative bodies as policy makers gains crucial importance, in contrast to political variables. The context of transposition is therefore more policy-specific than country-specific (Steunenberg 2007: 24ff.). While this points to the need of empirically founded typologies of culture such as Falkner et al. (2005) s worlds of compliance, these can be difficult to replicate (Toshkov 2007). The challenge for this stream of research is to carefully conceptualize different policy-making cultures, using analytical categories which can travel and empirically assessed (Mastenbroek 2005: 1112; Sartori 1970). Based on the assumption of a mutual reinforcement between political culture and institutional development, policy and administrative styles are deemed to influence policy implementation in multi-level governance systems (Richardson 1982; van Waarden 1987). Basic elements of policy styles are the scope and style of state intervention, modes of actor coordination and values of political culture (e.g. Feick 1990). The only element of policy styles which can easily be measured in a way allowing systematic cross-case comparison is the level of state intervention (Battaglini and Giraud 2003: 295). In this context, Vedung s (1998) tripartite classification of policy instruments offers a useful analytical framework. The policies are distinguished by the degree of authoritative force (coercion) exercised by the governour upon the target population. Sticks (regulations) are authoritative rules and directives limiting the choices available to the individuals within society. The key criterion is that the governee is obligated to do what the governor tells her to do. Carrots (economic or non-monetary (dis-)incentives) are economic policy instruments which make an action easier or more difficult by adduction or deprivation of 7

10 Country level Policy level material resources (money, time, effort, other valuables). The addressees are not obligated to take the measures involved. Finally, Sermons are voluntary appeals by means of information or exhortation, characterized by the absence of obligation (Vedung 1998). The degree of coerciveness is a formal feature of policy instruments, meaning that is independent of their action content. A stick is not necessarily more restrictive than a carrot in substantial terms. Since policy instruments usually come along in combinations, the degree of coercion of a policy style should be conceived as a mix of multiple policy instruments and be based on their relative importance in a country (Sager 2009: 541 ff). The criterion of the coerciveness of policy styles is comparable with the criterion of discretion in EU regulatory modes, which facilitates an assessment of the cultural misfit argument (Mastenbroek and Kaeding 2006): For instance, a responsive new instrument should be rendered more restrictive in a country with a coercive policy style, and such countries may generally customize the EU minimal standards more extensively in terms of restrictiveness, than countries which traditionally rely more on voluntary instruments. In contrast to this, nations with a less coercive policy style may try to customize EU regulatory standards by adding more detailed regulations which allow for certain relaxations, but they will hardly design domestic regulations which are more restrictive than the EU policy. Table 1 resumes the conditions for customization and their expected direction of influence. Table 1: conditions for customization Condition Expected direction of influence on extensive customization Responsive regulatory mode (RESPREG) + Salient issue (SALIENT) Domestic resistance (RESIST) +; interaction with domestic resistance +; interaction with veto points Many veto points (VETOPOINT) + Many veto players (VETOPLAYER) Coercive policy style (COERCIVE) +; interaction with salience No expectation; depends on fit with regulatory mode 8

11 4 Empirical research design The presentation of the research design begins by defining the cases and justifying their selection before presenting the method, the data bases and the measurement of concepts. 4.1 Case selection: The regulations of veterinary drugs Most research on veterinary drugs focuses on their authorisation (e.g. Krapohl 2007), where an underlying trade-related dimension fosters harmonization. FFigure 1 illustrates that dispensing and administration of veterinary drugs are more domestic processes which grant more regulatory leeway at the national level. Accordingly, I analyze the domestic regulations of the dispensing and administration of veterinary drugs for food-producing animals. The dispensing of veterinary drugs for livestock is defined as the process by which a veterinary drug reaches the end user who administers it to the animal. Administration refers to the process during which drugs are applied or given to the animal. These policies regulate the prescription and supply (dispensing), and use of veterinary drugs (administration; including the manufacture of medicated feedingstuffs by livestock holders). They concern the type, the amount, and the storage of drugs that may be given to animals, under what circumstances, and by whom, and the documentation requirements for these processes. Figure 1: The life cycle of a veterinary medicinal product Source: Authors own illustration. Boxes represent stations (actors) and arrows stand for stages (processes). Four western European countries are compared. Because livestock farming has a similar 9

12 relevance for agriculture, Austria, Germany, France and the UK face comparable regulatory requirements: the share of products of animal origin in agricultural value added ranges from one third to one half, and the contribution of the primary sector to the total gross value is below two per cent (Sager et al. 2011: 301). These countries have all been a member of the EU since the early nineties, when the first EU regulations concerning veterinary drugs were issued, which means that the historical evolving of the regulations was embedded into similar frame conditions. Furthermore, all four nations have experienced scandals or crises related to issues of animal diseases and food safety at the beginning of the millennium: The bluetongue disease in France (2007), the UK (2007, plus the severe BSE crisis and the foot and mouth disease) and Austria (with an additional veterinary drugs scandal in 2001). In Germany, the problem of resistances to antibiotics transferred from food products to humans has received considerable public attention. This case selection thus raises the question inhowfar countries customize EU directives when facing similar regulatory challenges, but low standardization pressures from the single market. 13 regulations of the dispensing of veterinary drugs and 6 policies for the process of administration were chosen according to a) their direct comparability and clear distinguishability from other processes in the life cycle of a veterinary drug, b) the existence of an EU directive regulating this specific aspect, c) the possibility of customizing the policy, and d) the availability of exhaustive information on the domestic regulations of these aspects. This results in 76 domestic regulations as cases. Table 2 provides an overview of the policies, their content, the legal basis at the EU level, and their main target groups. I compare most similar cases (Lijphart 1975: 165) with varying outcomes (Mill s [1843] method of difference). Limiting the number of potential intervening factors (Lijphart 1971) is a prerequisite for the application of the method of Qualitative Comparative Analysis (QCA) (Schneider and Wagemann 2010: 6), to which I turn next. 10

13 administration Dispensing Table 2: Policies, EU legal basis, and target groups Policy id Policy content EU directive EU regulatory mode clinicalex Requirement of clinical examination of 90/167/EEC New animals prior to prescribing drug instrument dispact Actors authorized to dispense 2001/82/EC New prescription drugs instrument dispactmedf Actors authorized to dispense medicated 90/167/EEC Regulatory feedingstuffs standard dispamount Amount of prescription drugs which may 2001/82/EC New be dispensed instrument dispamountmedf Amount of medicated feedingstuff which 90/167/EEC Regulatory may be dispensed standard dispcat Dispensing categories 2001/82/EC New instrument disprights docdisp Dispensing rights of veterinarians and pharmacies which exceed the mere dispensing of drugs Required duration of storage of documentation of dispensing, by whom 2001/82/EC 2001/82/EC New instrument Regulatory standard docprescr Required duration of storage of prescription and by whom 2001/82/EC Regulatory standard manufactnonauth Actors authorized to manufacture drugs 2001/82/EC New which do not require a market instrument authorization prescmedf Medicated feedingstuffs require a 90/167/EEC Regulatory prescription on a standardized form standard prescract Actors authorized to prescribe veterinary 2001/82/EC New drugs instrument prescrex Exemptions from prescription 2006/130/EC New requirement instrument adminactofflabel Actors who may administer the drug 2001/82/EC Regulatory which is used off-label standard cascade Possibilities for off-label use in cases of 2001/82/EC Regulatory supply shortage standard docadmin Required duration of storage of 2001/82/EC Regulatory documentation of administration, by standard whom onfarmmedf Possibility of on-farm manufacturing of 90/167/EEC New medicated feedingstuffs by livestock instrument owners topdressing Is top dressing (manual adding of drug 90/167/EEC New into feed or water) allowed? instrument withdraw Withdrawal periods 2001/82/EC Regulatory standard Principal target group(s) Veterinarians, Livestock owners Veterinarians, Pharmacies Veterinarians Veterinarians, Livestock owners Veterinarians, Livestock owners Veterinarians, Pharmacies Veterinarians, Pharmacies Veterinarians, Pharmacies, Livestock owners Veterinarians, Livestock owners Veterinarians, Pharmacies, Livestock owners Veterinarians, Livestock owners Veterinarians Veterinarians, Pharmacies Veterinarians, Livestock owners Veterinarians Veterinarians, Livestock owners Livestock owners Livestock owners Veterinarians, Livestock owners 4.2 Method: Fuzzy set analysis The set-theoretical method of Qualitative Comparative Analysis (QCA) (Ragin 1987) is designed for, but not restricted to, intermediate numbers of cases and has a substantive qualitative element. Shaped by the back-and-forth between ideas and evidence (Ragin 11

14 1987), it focuses on the causal complexity behind a phenomenon (Berg-Schlosser et al. 2008). This entails the assumption of causal asymmetry which refers to the fact that the occurrence of an outcome can have a different explanation than its non-occurrence. The assumption of equifinality allows for different, mutually non-exclusive explanations of the same phenomenon. Instead of assuming isolated effects of single variables, the assumption of conjunctural causation foresees the effect of a single condition unfolding only in combination with other ( ) conditions. (Schneider and Wagemann 2012: 78, emphasis in the original). Since my research question aims at disentangling the effect of EU law relative to, and in its interplay with, domestic factors on customization, I consider QCA an appropriate method for the analysis. QCA follows a deterministic understanding of causality in that its core interest is identifying necessary and sufficient conditions for an outcome. These explanations most often appear in combinations of conditions (also called configurations or paths). An explanation X is considered as necessary ( ) for the outcome Y if Y cannot occur without X. On the other hand, an explanation X is sufficient ( ) for the outcome Y if X always leads to Y. Cases are members of sets which represent concepts, for instance, in the set of extensive customization of EU directive. Fuzzy-set QCA (fsqca) (Ragin 2000), which I apply here, allows that cases display features to different degrees. Anchors for a case s set membership are defined, ranging from 0 (full nonmembership; e.g. no customization) to 1 (full membership; e.g. extensive customization) with a crossover point at 0.5 (neither in nor out; neither absent nor extensive customization). Recent descriptions of the method are found in Ragin (2008a), Rihoux and Ragin (2009), and Schneider and Wagemann (2012). Based on theoretical and substantive knowledge, the membership of each case is determined in each condition (a process called calibration ), and then in each logically possible combination of conditions. These latter combinations are represented in the rows of a truth table. What follows is a logical minimization process during which the shortest possible causal expression for the combinations of conditions causing the outcome (the solution term) is formulated. The basic idea is that if an outcome D is present in a case displaying A, B and C as well as in another case which displays A and C, but not B, it obviously does not make a difference for the occurrence of D whether B is present or not. FsQCA uses the logical operators or (+) and and (*) of Boolean, respectively fuzzy, algebra. Denoting the presence of a condition with capital letters and its 12

15 absence with lowercase letters, we observed that A*B*C + A*b*C D, which can be minimized accordingly to A*C D. There are two performance indicators for the final solutions. Consistency measures the degree to which the statement of sufficiency or necessity corresponds to the empirical evidence. Coverage indicates how much of the observations is explained by the solution (Schneider and Wagemann 2012: 128; Ragin 2006: 291). The basis upon which appropriate levels for these measures are chosen should be research-specific (Schneider and Wagemann 2010: 406; Schneider and Wagemann 2012: 127, 143ff.; Ragin 2006: 293). 4.3 Data and measurement This study is based on evidence attained during a commissioned research project about veterinary drugs regulations in Europe (Sager et al. 2011). Methods for data collection comprised an extensive document analysis of the respective legal bases, documents concerning the policy process and secondary literature. In each country and for the EU legislation, between three and four semi-structured telephone interviews were conducted with agents of stakeholder groups and of the administration. Some stakeholder groups answered written questionnaires. To ensure the correctness of the results, each case study was cross-read by a legal expert familiar with the respective national regulations. I now turn to the measurement and calibration of my variables as resumed in Table 3 below. 2 Table 7 presents the coding of country variables, Table 8, the raw data matrix, and Table 9 (all appendix) contains the fuzzy set membership values of each case. Customization Customization has two dimensions, namely the density and the restrictiveness of the domestic regulations. The measure of customization loosely follows Knill and Lehmkuhl (2002: 260) and Knill et al. (2012) and takes on a comparative perspective: the domestic regulations are compared with each other and then classified according to whether they are absent, relatively moderate or extensive in their degree of density (amount of rules) 3 and 2 The raw data upon which the coding is based can be provided by the author upon request. 3 There are some instances when an EU rule facilitates something without specifically prescribing it. In such 13

16 restrictiveness (amount of substantial freedom of action left to policy addressees) in comparison to the EU template. If the domestic regulation entails no changes as compared to the EU directive, I assign it a value of 0. Moderate domestic changes receive a value of 1; extensive changes, a value of 2. A detailed documentation of the coding decisions was undertaken and can be provided by the author upon request. The added customization index ranges from 0 to 4, which is transformed into a four-value fuzzy set (Ragin 2009: 81; see Table 3). Index values of 0 indicate absent customization (fuzzy value 0; customization). With index value of 1, customization is more absent than extensive, whereas values of 2 and 3 stand for more extensive than absent customization. An index value of 4 is conceived as extensive customization (fuzzy value 1; CUSTOMIZATION). EU regulatory mode Each EU policy is classified following Knill and Lenschow (2003) and Treib et al. (2007). A responsive regulatory mode is conceived as a dichotomous crisp set and takes on the value 0 (respreg) for EU regulatory standards, and 1 (RESPREG) for new instruments. Issue salience An issue is deemed salient if it receives a high amount of public attention, i.e. it is specifically mentioned as subject of discussions in the interviews, or in the analyzed documents. An issue which is salient at the EU level is coded as such for all countries, whereas otherwise the salience of an issue is restricted to the particular country. Issue salience is also conceived as crisp set and is present with salient issues (1; SALIENT), whereas the absence of visible discussion is interpreted as lacking issue salience (0; salient). The visibility of an issue may indicate that it is disputed, or simply that it is granted particular regulatory importance. situations, it was not considered as a higher degree of density if the domestic regulation did not adopt this possibility. For example, EU policy leaves open the possibility of on-farm manufacture of medicated feedingstuffs; if Germany does not avail itself of this opportunity, this restriction is not coded as an additional regulation, but as extensive additional restrictiveness. 14

17 Domestic resistance The prevalence of domestic resistance rests upon two criteria. First, at least one of the policy s final addressees, i.e. of the actor groups involved in the processes of dispensing and administration of veterinary drugs, opposes the policy (see Table 2). For each country, the most relevant organizations representing these stakeholder groups were identified. The document analyses and surveys then enable a qualitative assessment of whether one of these groups opposes the policy (3) or not (0). Second, the amount of domestic resistance is contingent upon the potential of the opposed stakeholder groups to successfully influence domestic policy making processes. Their power to exert influence, in terms of networks with the public administration, political relevance, activites such as, and/or resources for, lobbying at the national and European level, was evaluated by the interviewees for each country. It ranges from absent (1) over medium (2) to significant (3) (Sager et al. 2011: 372). These two variables are added into an index. The final condition domestic resistance is conceived as a four-value fuzzy set. It is fully given (RESIST) if one of the target group opposes the policy and the opposed target group has a significant power to exert influence (index value of 6). Resistance is more given than not if a target group opposes the policy, but is only moderately or not influential (index value of 4 or 5). If no one opposes the policy, but one of the policy s principal target groups is influential (index value of 3), resistance is mostly absent, but we can still expect the addressees to bring in their interests in order to customize the policy to them. 4 When none of the target groups is opposed to the policy and they are all not or only moderately influential, resistance is absent (resist) (index value between 1-2). Institutional veto points Following Haverland (2000), the notion of institutional veto points entails, firstly, the degree to which decentralised actors of the implementation chain are involved into legislation. For each country, the degree of decentralization was determined in terms of whether the regions have no (1), partial (2) or far-reaching (3) legislative competencies in the area of 4 For policies with several addressees, the most powerful target group served as point of reference. 15

18 veterinary drugs regulations (Sager et al. 2011: 354). A second dimension refers to the strength of bicameralism. I rely on Huber et al. s (2004) measure as contained in Armingeon et al. s (2012) Comparative Political Data Set III. The value 0 indicates no second chamber or second chamber with very weak powers, 1 equals weak bicameralism, and 2 stands for strong bicameralism. Decentralization and bicameralism are added into one index of institutional structure, ranging from 1 (centralized) to 5 (fragmented). Third, the notion of corporatism here refers to the degree of corporatist interest intermediation (as opposed to other uses of the term, see Siaroff [1999: 176-7]). The involvement of stakeholder groups in the formulation of policies of veterinary drugs is routinely practised in all four countries via diverse consultation mechanisms, whereas the latters effectiveness is not primarily determined by their degree of (in)formality (Sager et al. 2011: 138, 184, 234-5, 282-6, 373). In the absence of a meaningful policy-specific differentiation, I rely on the Siaroff s (1999) notion of integration as a proxy. I use the Siaroff index of the year 1995, as contained in Armingeon et al. (2012), which ranges from 1 (least integrated) to 5 (most integrated). Since transposition processes often take place in administrative bodies, I grant corporatism equal importance as institutional structure, both of which are added into one index ranging from Few veto points (vetopoint) are given when the index value is 2, many veto points (VETOPOINT) are indicated by an index value of 8, with a crossover point at 5. Veto players I rely on the measure of the number of veto players in the updated dataset by Tsebelis (1995) 5 (for the year 1990 or 2000, depending on the EU directive concerned). Based on the empirical distribution (Table 7), this set ranges from index values of 1 (few veto players, vetoplayer), to 4.5 (many veto players, VETOPLAYER), with a crossover point at [last visit 25/04/2013]. 16

19 National policy styles I classify each domestic policy instrument according to the degree of authority exercised on its target population, rather than on the first agent in the implementation chain (Sager 2009, 540). Sermons are given the value 0, carrots the value 1, and sticks the value 2. These values are summed up for each country and then divided by the number of policy instruments. I distinguish between the regulations of dispensing and those of administration, because the countries differ in the relative regulatory importance they grant to these two aspects (Sager et al. 2011: 369; see also Table 7 in the appendix). 6 The resulting index represents the average degree of coerciveness of the policy style a) for the regulations of dispensing and b) for the regulations of administration. Since this is a highly policy-specific measure, the calibration is based on a comparison of the cases: Empirically, the index ranges ranges from 1, indicating a non-coercive policy style (coercive), to 2, indicating a highly coercive policy style (COERCIVE), with an average degree of coerciveness of 1.6 which serves as crossover point. 6 The dispensing process tends to be regulated more restrictively, and more extensively (see case selection), than the administration of drugs. Apparently, the regulators deem the dispensing policy, which regulates the access to drugs, a more effective tool to control the risks associated with the usage of veterinary drugs. The everyday use of the drugs on livestock farms is arguably harder to control. 17

20 Table 3: Outcome and conditions: measurement and calibration Set Extensive customization (CUSTOMIZATIO N) Responsive regulatory mode (RESPREG) Salient issue (SALIENT) Domestic resistance (RESIST) Many veto points (VETOPOINT) Many veto players (VETOPLAYER) Measurement Added index (from 0 to 4) of 2 dimensions: density and restrictiveness of domestic regulation as compared to EU directive, each ranging from 0 (absent) over moderate (1) to extensive (2) 1 Calibration (Anchors) Regulatory standards (0) and new instruments (1) 2 Explicitly mentioned public attention yes (1) or no (0) 1 - What issues regarding the regulation of veterinary drugs (access to and use of) are the relevant stakeholder groups discussing about? - Are there any problems related to the transposition of European regulations at the national level? Added index of opposition of target group (yes = 3, no = 0) and its power to exert influence (absent = 1, medium = 2, significant = 3) 1 - Could you please give me the names of 2-4 relevant interest groups which are influential in the policy making processes regarding the regulation of veterinary drugs? - How would you rate their power to exert influence (Networks with the public administra-tion, resources for lobbying )? Added index of 1) institutional structure (from 1 to 5, with 2 dimensions: decentralization 1 (1-3) and bicameralism 3 (0-2)), and 2) corporatism 3 (from 1 to 5) Number of institutional and partisan veto players (capacity to block the adoption of a policy) 4 Coercive policy Average share of sermons (0), carrots (1) and style (COERCIVE) sticks (2) in national regulations (distinction of regulations of dispensing and of administration) 1 1 Source: Sager et al Sources: Directive 2001/82/EC and Directive 90/167/EEC. 3 Source: Armingeon et al Source: Tsebelis 1995, updated veto player dataset

21 5 Policy and country profiles Prior to the causal analysis, the cases are described with regard to their average set membership in the respective policy characteristics or country characteristics, on the one hand, and the degree of customization, on the other. The resulting policy and country profiles provide an unprecedented systematic overview of the policy area of veterinary drugs regulations, the domestic settings, and the countries transposition style. Figure 2: Policy profiles Key of case IDs see Table 2. N = 76 Average set membership of cases sorted by policy. Values above 0.5 indicate that a factor is more present than absent, and vice versa. Firstly, Figure 2 presents the policy profiles, sorted from low to high levels of customization. The policies differ largely in the extent to which they are customized at the domestic level, but there is a tendency for responsive EU regulations which are salient and face domestic resistance to be customized more extensively. The most salient and opposed policies are those directly relating either to the freedom of competition of dispensing actors, or to the liberties of livestock owners. Figure 3 shows the profiles of the four countries with regard to the mean amount of 19

22 domestic resistance to the EU policies, the number of veto points and veto players, their average policy style and their customization behavior. In order to shed light on different customization behaviors, the indices for regulation density and restrictiveness, upon which the final customization index is based, are also provided in these country profiles. 7 Figure 3: Country profiles Average set membership of cases sorted by country. N = 76 Values above 0.5 indicate that a factor is more present than absent, and vice versa. The countries display considerably divergent domestic settings, regulatory and customization styles. Austria and France are comparable in their customization behavior; in line with the countries rather coercive policy tradition, the EU policies are rendered slightly more restrictive (especially in Austria), but not differentiated extensively. The policies face a low amount of domestic resistance on average. However, the two countries differ with regard to the number of veto points and player. Federal Austria has many veto points but few veto players, whereas the opposite is the case in centralized France. Germany and the UK provide two interesting counterexamples which indeed approximate the theoretical ideal types. Germany is the country which customizes the EU directives most extensively, mainly in terms of additional restrictiveness. This corresponds to its highly coercive policy style. Furthermore, domestic opposition to EU policies tends to be very high 7 For the sake of comparability, these indices were recoded into a fuzzy set: index values of 0 result in a fuzzy set membership of 0 (absent density/restrictiveness), values of 1 constitute the crossover point 0.5 and values of 2 constitute full membership (1; high DENSITY / RESTRICTIVENESS). 20

23 in Germany, as well as the number of veto points and veto players. The restrictive regulations of the German system are seldom differentiated in order to allow for more flexible arrangements. In contrast to this, the UK takes a very liberal approach to individual responsibility, relying on recommendations of best practise, and interprets legal definitions in an innovative manner in order to retain its liberal regulations. As a result, EU policies tend not to be customized, and if so, more in terms of regulation density, but hardly with regard to additional restrictiveness. The domestic setting in the UK does not make it susceptible to extensive customization, since averaged domestic resistance is rather limited and the number of veto players and of veto points is very low. In summary, these descriptive overviews suggest that both at the policy and at the country level, the proposed conditions for customization could be causally relevant in the expected direction. 6 Results I now turn to a causal assessment of the way in which the various interplays of these factors affect the customization of EU directives during transposition. I use the fsqca 2.5 software and follow recommendations of best practise in QCA (Schneider and Wagemann 2010). None of the conditions is necessary (i.e., passes the 0.9 consistency threshold; Schneider and Wagemann 2012: 143) for neither high nor low levels of customization (see Table 10 appendix). As the assumption of causal asymmetry implies, sufficient conditions for extensive and limited customization are assessed separately. The results are first briefly presented, and afterwards discussed in view of the postulated theoretical arguments by relating them to illustrative case examples. 6.1 Sufficient conditions for extensive and limited customization The combinations of conditions (also called paths or configurations hereafter) which are sufficient for an extensive customization of EU directives are presented in Table QCA analyses have to deal with the phenomenon of limited diversity: if n denotes the number of conditions, some of the 2n (here: 64) logically possible combinations of conditions are not observed empirically (logical 21