Paper prepared for EUSA Conference, 31 March - 2 April 2005, Austin Texas Explaining delays in the transposition of utilities directives: A sociological institutionalist account Sara Berglund Utrecht University, Faculty of Social Sciences University College, P.O. Box 80145 3508 TC Utrecht, The Netherlands s.berglund@fss.uu.nl Abstract It is often claimed that there is a problem with delayed transposition of directives within the EU (European Union). However, the data on which many of these claims are made show some serious shortcomings. By using a new dataset on the transposition of utilities directives that overcomes many of these problems, this paper set out to examine if there are problems with transposition delays in the area of utilities, and if so, how these problems can be explained. It was found that in nearly half of the cases, transposition was delayed, and the average delay was about seven and a half months. There were also significant differences between the member states in this respect. In search for explanations for these differences, existing literature on the compliance with EC (European Community) law was reviewed, and this produced an extensive list of potential explanatory variables. Rather than to test all these variables, a theoretical approach, namely sociological institutionalism, was adopted in order to bring theoretical focus to the analysis. The variables that could be derived from this approach were subsequently selected and tested. Of the explanatory variables that were tested, only the variable "Government efficiency" remained significant when the grouped structure of the data was controlled for. The fact that only a low amount of the variance was accounted for in the analysis, and that the policy area (telecommunications or energy) to which a directive belongs proved to be an important factor in accounting for delays, suggests that the explanatory variables are too broad. This calls for further research in which the explanatory variables are operationalised in such a way that they are more specific to transposition and to the characteristics of a policy area within the member states. Introduction In the area of utilities, the member states of the EU have agreed on ambitious policies that aim to open up entire markets for competition and to regulate these markets. These policies are often made in the form of directives. In contrast to regulations, directives have to be transposed into the national legislation of the member states. The member states have some freedom regarding the means used, but the transposition measures that the member states adopt must serve to fulfil the objectives of the directives, and the member states are obliged to transpose directives before a particular date. Transposition is thus a crucial first step in realizing policies in the utilities area in practice. However, it has often been claimed that there are problems with transposition in the form of delays (see e.g. Lampinen and Uusikylä 1998; Bursens 2002; European Commission 2003b). In this study, it is examined if the claim that there are problems with transposition holds for directives in the area of utilities, and if there are differences in the transposition records of the member states. This is done by studying the transposition of a large number of utilities directives in the UK, Germany, the Netherlands, Spain and Greece. A new dataset, consisting of data provided by the Commission and the member states has been created for this purpose. Provided that there is indeed a problem with the transposition of utilities directives and differences between the member states, the question of why the member states differ arises. A review of existing studies on compliance with EU law produces an extensive list of heterogeneous variables that attempt to explain such differences. These variables include the level of corporatism (Lampinen and Uusikylä 1998; Mbaye 2001), the number of veto players (Mbaye 2001; Giuliani 2003; Börzel, Hofmann et al. 2004), bargaining power in the Council (Mbaye 2001), citizen's support for the EU (Lampinen and Uusikylä 1998; Mbaye 2001), the rule of law (Börzel, Hofmann et al. 2004; Sverdrup 2004) etc. However, the results of the testing have been mixed. Rather than to attempt to test all of these variables, a theoretical approach, namely sociological institutionalism, has been selected in order to bring focus to the analysis. The paper is organised as follows. First, the area of utilities regulation and the regulatory frameworks adopted within the EU in this area are presented. Second, a new dataset, which overcomes many of the shortcomings of data used in existing studies, is presented. In the next section, existing studies on compliance with EC law are reviewed, a theoretical framework is developed, variables are selected and hypotheses are formulated. Finally, a presentation of the results of the testing of these variables is followed by conclusions and implications for further research. 1. Regulating utilities in Europe The telecommunications and energy (utilities) sectors have traditionally been characterized by state monopolies. Over the past decades, however, a process of privatisation, liberalisation and regulatory reform has transformed these sectors fundamentally. One of the driving forces in this process has been the development of EC regulatory frameworks, mainly within the context of the Internal Market. Other important factors have been technological and economic developments, internationalisation and new ideas concerning the benefits of competition (Thatcher 2001). 1.1 Telecommunications Telecommunications in Europe were viewed as a natural monopoly at least until the late 1970s, and this justified the granting of special and exclusive rights to a public telecommunications operator (PTO), which held a monopoly in the provision of services and over the infrastructure. Regarding telecommunications equipment, the PTOs generally engaged in privileged supplier relationships (Sauter 1995; Hunt 1997; Thatcher 2001a). Still, the liberalisation of the telecommunications sector in the EU proceeded rather quickly. The first steps towards a regulatory framework for telecommunications were taken in the mid-1980s, and by 1998, telecommunications were completely liberalised in most EU member states (DG Information Society 1999). The regulatory framework consists mainly of directives, and they can be divided into two categories, namely those aiming at liberalisation (deregulation) and at harmonisation (re-regulation). The liberalising directives aim to create competitive telecommunications markets, and the harmonising directives aim to prevent abuse of market power and to ensure public interests, such as "universal service" (Hunt 1997). The liberalising directives were issued by the Commission on the basis of treaty provisions on competition rules balanced against the national public interest, and this initially lead to resistance among the member states. However, in the court cases that followed, the European Court of Justice (ECJ) ruled in favour of the Commission (Sauter 1995, Schmidt 1998). The harmonising directives, on the other hand, were adopted by the Council and the European Parliament (EP). The first major step towards liberalising and harmonising the telecommunications sector was the Commission Green Paper on telecommunications in 1987, and this was followed by a number of directives. By the mid-nineties, the whole telecommunications sector, except for voice telephony and infrastructure (network operation), was liberalised. However, the excluded parts were quite important, since they accounted for about 85% of the revenues in the sector. In 1996 an important step was taken with the adoption of the Full Competition directive (96/19), which meant that by 1 January 1998 competition should be allowed throughout the telecommunications sector (Hunt 1997; Thatcher 2001). In 1999, the Commission launched a major review of the EC telecommunications law, which resulted in the adoption of a new regulatory framework for the electronic communications sector in 2002. The basic goals remain the same, and in addition, it aims to be flexible, simple and technology neutral. The new regulatory framework consists of six directives and one decision, and largely replaces previous legislation in the field (DG Information Society 2004). 1.2 Energy Although energy was a topic in two of the founding treaties, energy has for a long time been low on the agenda, and attempts to create common energy policies have largely failed. In the Commission White Paper on the Internal Market in 1985, energy policy was not mentioned, since it was considered a very problematic area. However, the energy sector was regarded as essential for the working of the Internal Market, since it was an important part of providing a "level playing-field". Thus, in 1989, attempts were started at creating an internal energy market (Andersen, 2000:4; Matlary, 1996:257-8,263). Since the oil sector was already largely operating within a free market, and since the coal and nuclear sectors were covered by Treaty provisions, the focus was on the gas and electricity sectors (Andersen, 2000:5). There has been some success in opening up these markets, but the process of liberalisation has gone much slower than in telecommunications. One reason for this can be that the member states have strongly diverging interests in this sector, for example depending on whether they are net importers or exporters of energy. There has also been a strong resistance among the member states towards liberalisation of this sector, since they fear that they might eventually lose control this sector to the EU (Matlary, 1996:265). Maybe the most important obstacle in liberalising the energy market has been that it has been considered to hold some characteristics of a natural monopoly, i.e. that a single firm can supply market demand most efficiently. This is the case for transportation, i.e. transmission and distribution of gas and electricity. The level of monopolisation in the member states was also very high (Andersen, 2000:6; Coen, 2001:80-2). However, following the reform of the sector in the UK in the early 1980s, a discussion about whether or not free-market rules might be beneficial for the sector started in Scandinavia, France and the Netherlands (Matlary, 1996:258). There were some minor steps at adopting directives liberalising the energy sector in the early 1990s, but the breakthrough did not come until the late 1990s. In 1996, a directive concerning a single market for electricity was adopted, and for natural gas in 1998. These directives introduced a gradual deregulation of the production and transport of electricity and gas. Although the original Commission proposal had been changed substantially in the process, these directives still served to transform the European electricity and gas markets (Andersen, 2000:15-6). In 2003, two new directives (2003/54 and 2003/55) establishing common rules for the internal market in energy and gas were adopted. In addition to these directives, a number of directives concerning energy efficiency and energy labelling of household appliances have been adopted in the energy sector in the past decade. 1.3 Characteristics of directives in the utilities area The utilities directives are usually quite complex and many of them concern big issues, such as opening up a market for competition. Although this is not true for all utilities directives, since, especially in the energy sector, some directives do concern purely technical issues, on the whole, utilities directives are generally quite complicated and receive relatively seen a lot of attention. This means that the process of transposition is more political than in many other policy areas, and although lower level legislation can usually be used for transposition in many other policy areas, utilities directives have often been transposed through primary legislation. 2. A new dataset on the transposition of utilities directives Many existing studies on transposition (see e.g. Lampinen and Uusikylä 1998; Bursens 2002) use data published in scoreboards provided by the Commission. In these scoreboards, it is reported what percentage of the directives that each member state has not yet transposed. The goal set by successive European Councils is that the transposition deficit should not exceed 1.5%. However, this goal has never been reached. In May 2002, the percentage of the Internal Market directives not yet transposed was at the lowest level of 1.8%, and since then, the transposition deficit has increased again (European Commission 2003a; European Commission 2003b; European Commission 2005). Although these percentages do not seem very high, they are taken very seriously both by the Commission and the member state governments. However, what does this data actually tell us? Since it is only a measure of the percentage of directives that the member states have not transposed at a certain point of time it does not tell us anything about the performance of the member states between these dates. In addition, the data has an automatic upwards bias, since the number of existing directives increases over the years. In order to overcome this problem, it is possible to look at the percentage of directives that the member states have transposed each year of the directives that were adopted that year. This has been done by Giuliani (Giuliani 2003). However, this still does not tell us much about transposition delays. In order to answer the question of how late directives are actually transposed in the Netherlands, Mastenbroek (2003) developed a database which includes information about the deadline of the directive and the date of adoption, publication and entering into force of the Dutch national instruments used for transposition. In her study, she showed that about 40 % of all directives were transposed late in the Netherlands in the period 1995-1998. This shows that the problem is more serious than is depicted in the Commission scoreboards. For this study, a database with a similar structure has been developed. The database includes information about all utilities directives with a deadline before 1 February 2002, including directives no longer in force, and the national instruments that have been used to transpose these directives by the five selected member states. As mentioned above, the selected member states are the Netherlands, Germany, the UK, Spain and Greece. This provides a good sample since they include member states that are large and small, northern and southern, that have a federal and unitary structure and a majoritarian and consensus style. Since Spain became a member of the EC only in 1986, only directives adopted from 1986 on are included in this study, in order to be able to make comparisons between the member states. The total number of directives for this time period is 58, and the number of directives is roughly equal per policy area (44,8 % energy and 55,2 % telecommunications). Information about the national measures used for transposition of these directives in the five selected member states was gathered in a three-step process. First, the Commission database Celex was consulted. This database is extensively used for finding references to national transposition measures (Nunn-Price 1997). Unfortunately, Celex is far from complete. For many directives, there are no references to national transposition measures for one or more member states, or only references to old legislation. This is actually very surprising, since it is reasonable to assume that the Commission has information from the member states about how all directives have been transposed. For the utilities directives, information about new transposing measures was to be found in Celex in merely 56 % of the cases. For an additional 6 % of the cases Celex gave references to national legislation adopted before the directive was adopted. Since Celex does not contain links to national databases containing the texts of the legal instruments used for transposition, it was also necessary to look up these texts in national online databases or in libraries. Second, it was possible to fill in some of the gaps in Celex by consulting reports by the Commission on the implementation of certain directives. In a third step, some of the remaining gaps were filled by information provided by ministries and other sources in the individual member states. In total, information about new transposition measures was found for about 71 % of the cases. For the cases where no new instruments were found, there were a number where old national legislation was reported in Celex or the other sources. Although it is not necessarily the case, for many of these cases, it was probably not necessary for the member state to transpose the directive. For other cases, it is explicitly reported in the sources that transposition was not necessary. For the cases where there is no information, it is possible that transposition has not yet taken place, that existing national legislation was considered sufficient or that the directive has been transposed although no measure has been reported. Although it is difficult to make any guesses about whether or not there is a systematic bias in this data, there is no a priori reason to assume there is. Ideally, the dataset should of course include information about all the cases, but the current dataset is still preferable to the Commission scoreboards or only the data provided in Celex. In order to calculate delay, i.e. the number of days after the deadline that a directive is transposed, the date of adoption of the first instrument used by the member states has been used. The reason for the choice of the first instrument is that it is not possible to know when transposition in a member state is complete merely by looking at references to national legislation in Celex or other sources. There is always the possibility that new measures will be added later, and the first measure might in fact be sufficient in order to comply with the requirements of the directive. To use the first instrument thus avoids the problem of exaggerating delays. In cases where only old national legislation, i.e. legislation adopted before the adoption of the directive, was reported in Celex, this was noted in the database, although these instruments were not used in the analysis. The date of adoption of the national instrument was then compared to the deadline specified in the directive. 3. Problems with transposing utilities directives within the deadline If delay is calculated as the date of adoption of the national instrument minus the deadline of the directive, delayed transposition was found in nearly half of the cases (44,5 %). In 26,5 % of the cases, transposition occurred before the deadline or within the week of the deadline. For the rest of the cases, there is only information about old legislation, information that transposition was not necessary or no information. The average delay is 233 days (about 7,5 months), and the longest delay was 2671 days (over 7 years). These numbers show that there is in fact a problem with transposition in the utilities sector, and that delays are not only frequent, but also tend to be rather long. The five member states included in the study have quite different transposition records. This is demonstrated in the table below. Table 1. Differences in delay between the member states Member state Average delay in days The UK 86 The Netherlands 166 Spain 236 Germany 326 Greece 391 The UK has the best transposition record and transposes the directives on average 86 days (almost three months) after the deadline. Greece has the worst record and transposes on average 391 days (about a year and a month) late. The other member states have an average which falls somewhere in between these extremes. There are also differences between the two policy areas, energy and telecommunications. Table 2. Differences in delay between the policy areas Member state Policy area Average delay in days The Netherlands Telecommunications 361 Energy -63 Germany Telecommunications 386 Energy 272 The UK Telecommunications 116 Energy 54 Spain Telecommunications 217 Energy 259 Greece Telecommunications 640 Energy 110 Total Telecommunications 331 Energy 124 The average delay for energy directives is 124 days (about four months), and for telecommunications directives it is 331 days (about eleven months). There are also differences between the member states in this respect. While the Netherlands and Greece appear to have much more trouble with transposing telecommunications directives than energy directives, the other member states do not experience such a big difference. Spain even transposes telecommunications directives slightly faster. Another interesting aspect is how the transposition record has developed over time. The average delay has steadily decreased over the years, from an average of about one year and four months in the period 1985 until 1990 to an average of merely two months in the period 2000 until 2005. While this seems to indicate a very positive development, possibly caused by the increased attention for timely transposition both by the Commission and the member state governments, these numbers should be treated with caution for two reasons. Firstly, the number of cases in the early years is very small, while the bulk of the directives were adopted from 1990 on. Secondly, not all of the directives in the period 2000-2005 have been transposed yet by all member states, and the average delay will thus increase when transposition does occur. For example, Greece had, at the time of the study, only transposed three of the ten directives included in the study for this time period. In fact, it seems that the trend towards decreasing delay has been brought to a halt and again started worsening. The transposition of the new telecommunications package and the new energy directives has not proceeded smoothly. The Commission has initiated infringement proceedings against a number of member states for failure to notify transposition measures for these directives on time (European Commission 2003c; European Commission 2003d; European Commission 2004a; European Commission 2004b; European Commission 2004c). 4. Towards explaining delays in transposition 4.1 Explanations in existing studies After the publication of the path-breaking work of Siedentopf and Ziller (1988), in which the transposition of directives in a number of member states was examined and the process of transposition was outlined, there has been a number of empirical studies on transposition. Initially, these were mainly case studies of one or a few directives (Duina 1997; Knill and Lenschow 1998; Haverland 1999; Börzel 2000). The focus of many of these studies is environmental policy, and in general they tend to ascribe to an institutionalist approach. Recently there have also been a number of quantitative studies on transposition and compliance with EU law (infringement proceedings) (Lampinen and Uusikylä 1998; Mbaye 2001; Giuliani 2003; Mastenbroek 2003; Börzel, Hofmann et al. 2004; Giuliani 2004; Sverdrup 2004). In the table below, the variables tested and the findings of these studies are summarised. Table 3. Explanatory variables tested in quantitative studies on compliance with directives Explanatory variable Dependent variable Author(s) Expected effect (- leads to better compliance, + leads to worse compliance) Hypothesis confirmed? Level of corporatism Transposition, infringements Mbaye, Lampinen and Uusikylä - No Number of veto players Transposition, infringements Giuliani (2003), Mbaye, Börzel et al + Only in Giuliani's study, but only explains small amount of variance Bargaining power in the Council Infringements Mbaye - No (significant positive correlation) Comparative economic power Infringements Mbaye - Yes Power of a state (political and economic) Infringements Börzel et al + Yes, but explanatory power small The degree of consensual democracy Transposition Giuliani (2003) - No Regional autonomy Infringements Mbaye + Yes Administrative capacity/government effectiveness Infringements Mbaye, Börzel et al, Svedrup - Yes (only exception is one of Mbaye's indicators of this variable, namely fiscal resources) Stable and efficient political institutions Transposition Lampinen and Uusikylä - Yes Citizens' support for the EU Transposition, infringements Mbaye, Lampinen and Uusikylä - No (Mbaye even finds a strong positive correlation) Support of EU institutions Infringements Börzel et al - Yes, but weak Approval of democracy Infringements Mbaye - No Political culture in the member states. Transposition Lampinen and Uusikylä - Yes Rule of law Infringements Börzel et al, Svedrup - Yes Length of membership Infringements Mbaye +/- Positive relationship Type of legal measure used Transposition Mastenbroek Statutes take more time than Orders in Council and Ministerial Orders Yes Number of instruments used for transposition Transposition Mastenbroek + No Organizational differences between ministries ('Chinese wall' between stages of preparation and transposition) Transposition Mastenbroek + Yes, but does not explain all variance Quality and politicization of directive Transposition Mastenbroek Commission directives are transposed faster than Council directives Yes New directive or amendment Transposition Mastenbroek Amendments are transposed faster No Length of deadline Transposition Mastenbroek + Yes Goodness of fit between EU and national policies and institutions (new national instrument or amendment) Transposition Mastenbroek - Yes Inter-ministerial coordination problems (number of ministries involved in transposition) Transposition Mastenbroek + Partly (only slower when three or more ministries involved) Use of QMV Infringements Mbaye + No Organisation of the executive (in relation to the definition of the national interest in EU bargaining) Transposition (standardized residuals) Giuliani (2004) Underperformance is expected where organisationally the EU issues has not achieved sufficient autonomy from traditional foreign policy Yes Involvement of parliament in EU affairs Transposition (standardized residuals) Giuliani (2004) - Yes National coordination capacity Transposition (standardized residuals) Giuliani (2004) - Yes Dimension and organisation of the national PR in Brussels Transposition (standardized residuals) Giuliani (2004) - No Coordination at the EU level (Coordination capacity and level of networking) Transposition (standardized residuals) Giuliani (2004) - Yes The policy style exhibited by a member state Transposition (standardized residuals) Giuliani (2004) Anticipatory policy style is comparatively better than a reactive one Yes, but weak result Although the number of studies included in this table is not very large (seven studies), the number of explanatory variables that have been tested is remarkably high. This large number of heterogeneous variables, combined with the mixed results of the testing, suggests the need for a theoretical focus. In the next section, the theoretical framework which will be used in this study will be developed. Those variables from existing studies that can be derived from this framework will then be selected and tested on the data on transposition of utilities directives. 4.2 Developing a theoretical framework: how does sociological institutionalism view transposition? In the past two decades, there has been a renewed interest for institutions in political science, and institutional analysis is now a central part of the discipline. The term "new institutionalism" is commonly used as a collective term for the approaches that adhere to this new focus. The most basic assumption, which holds these approaches together, is that institutions have a structuring effect on political actions and outcomes (Aspinwall and Schneider 2000). However, there is not one "new institutionalism". The "new institutionalist" body of literature can better be described as a collection of rather diverse theoretical approaches. Peters (1999) has identified seven different "new institutionalist" approaches, but more commonly, three basic approaches, namely sociological institutionalism, historical institutionalism and rational choice institutionalism, are referred to (Hall and Taylor 1996; Thelen 1999; Aspinwall and Schneider 2000). It has also been argued that the basic distinction between these approaches are between normative or sociological institutionalism, which emphasizes the importance of norms and identities, and rational choice institutionalism, which is a refined version of rational choice theory (Lowndes 2002). Expressed in another way, sociological and rational choice institutionalism form the two ends of a spectrum, with historical institutionalism lying somewhere in between (Aspinwall and Schneider 2001). In this study, the focus is on sociological institutionalism. How would the process of transposition be pictured from a sociological institutionalist perspective? How well can delays in transposition be explained by employing this approach? Sociological institutionalism, as the name suggests, has its roots in the rich literature on institutions in sociology (see e.g. Powell, DiMaggio et al. 1991). The interest for institutions within the sociological discipline dates as far back as to the leading theorists Weber and Durkheim (Peters 1999). In 1984, the sociological conception of institutions was brought into the domain of political science by March and Olsen. They criticised the lack of attention to institutions in contemporary political science, and the conception of politics as the sum of individual behaviour. Further, they identified the rise of a "new institutionalism" in political science, and outlined the possibilities for a political theory attentive to institutions (March and Olsen 1984). In subsequent works, they have worked out this theoretical framework more in detail (1989; 1996). They argue that the focus in the dominant works in political science on a "logic of consequentialism", according to which political actors aim to maximize their personal utility, does not suffice to describe behaviour in politics, but that there is also another logic of action at work, namely the "logic of appropriateness". According to the "logic of appropriateness", actors act according to their interpretation of the situation and the role they have in this situation. Thus, the personal preferences of the actor recede in importance, and it becomes more important for the actor to do what is appropriate or expected of him in the situation at hand. They thus emphasised the structuring effect of norms, values and culture on political actors, and their importance for political outcomes. The sociological notion of institutions as important in shaping the identities, norms and values of individuals has also found its way into political science by another route, namely through the development of the so called "constructivism" or "social constructivism" in the field of international relations. This approach shares many of the characteristics of the sociological institutionalism, and is often considered as part of the approach (Hall and Taylor 1996; Aspinwall and Schneider 2000). The constructivist approach builds on the classic work of Berger and Luckmann from 1966 (Hall and Taylor 1996), and has been further developed in political science by, among others, Wendt (1987; 1992; 1994) . The focus of the research is on different processes of socialisation in which the preferences of the actors are formed. Language plays an important role in two of these processes, namely arguing (Risse 2002) and persuasion (Checkel 1999). A third process, namely learning, has also been developed by Checkel (Checkel 1999; Checkel and Moravcsik 2001). So, how would the process of transposition be described if a sociological institutionalist/constructivist perspective is adopted? According to sociological institutionalism, action is guided not primarily by strategic calculation, but by the role models of the actors involved, and their perceptions about what is expected of them and which rules apply to the situation at hand. From the point of view of sociological institutionalism, what is important here are thus values and norms within the government and administration concerning the importance of following laws and rules and commitment to agreements. Such general norms and values would be expected to have a positive effect on the general transposition record of a member state. However, it is also important that the actors have the necessary resources in order to perform their tasks according to these norms, and that the rules applying in the situation are sufficiently clear. If there is a contending set of rules that could be applied to the situation or if the rules are not clear, it might be difficult for the actors to know which the proper line of action is (March and Olsen 1996). As pointed out by scholars adhering to the constructivist strand, processes of socialisation are of importance for how these actors view both the situation and their role in it. For example, it can be assumed that the transposition record is enhanced if actors at the national level which are involved in transposition are involved in the whole process leading up to the adoption of the directive. The mechanism behind this is that the actors are involved in a process of learning, which makes them more positively inclined both towards the EU in general and towards the policy embodied in the directive. An alternative mechanism would be that the involvement of actors involved in transposition, especially the parliament, in the negotiation stage would grant the directive a more legitimate status. 4.3 How can delays in the transposition of utilities directives be explained? Among the explanatory variables that have been tested in existing studies, there are a number of variables that can be connected to sociological institutionalism, namely "Rule of law", "Administrative capacity/government efficiency", "Political culture", "Chinese walls/National coordination capacity" and "Involvement of parliament". These variables can be divided into two groups. The first group, consisting of the first three variables, contains variables that are concerned with general characteristics of the member state and its administration. The second group, consisting of the latter two variables, contains variables that are concerned with the way in which the member state is organised regarding EU affairs. In the quantitative studies described earlier in this paper, these variables have been successful in explaining delays in transposition, and, to an even greater extent, the number of infringement procedures opened against member states. This is surprising, since sociological institutionalism is often associated with qualitative case studies. It is also promising, both for the present study, and for the use of sociological institutionalism in explaining differences in compliance with EU law. Below, it will be described how these variables have been used in the literature. The causal mechanism by which the variables could be expected to affect the timeliness of transposition will be worked out and corresponding hypotheses that will be tested in this study will be formulated. It will also be described how these variables will be operationalised in the present study. 4.3.1 Variables concerning the general characteristics of member states The first variable in this group is "Rule of law". The rule of law can be defined as "the extent to which agents have confidence in and abide by the rules of society" (Kaufmann, Kraay et al. 2003). Although this variable is very broad, it can be assumed, along the lines of sociological institutionalism, that domestic traditions of rule of law form the role models of civil servants and other actors involved in transposition in such a way that they feel that compliance with law, including directives, is important. In the studies in which this variable is tested, namely the studies by Börzel et al. (2004) and Sverdrup (2004), it is assumed that a high support for the rule of law in a member state should have a positive effect on the compliance with directives. In these two studies, this variable was quite successful in explaining differences between member states in the number of infringement procedures that are initiated and at which stage they are resolved. However, it has not yet been tested if this variable has an effect on the timeliness of the transposition of directives. As compliance with directives start with meeting the deadline for transposition, this variable should have an effect also in this stage, and if such is the case will be tested in the present study. The hypothesis to be tested is: H1: The stronger the support for the principle of rule of law, the more timely transposition will be. In the study by Börzel et al., this variable is operationalised by using opinion pool data by James L. Gibson (1996). This operationalisation has the advantage that it is measured specifically for European countries, but the drawback is that it is only measured at two points of time (1992-3). Sverdrup uses an other source in order to operationalise this variable, namely an indicator found a study by Kaufman, Kraay and Mastruzzi published by the World Bank (2003). This indicator includes perceptions of the incidence of crime, the effectiveness and predictability of the judiciary, and the enforceability of contracts. The advantage of this measure is that it is measured over a longer time period (1996-2002), although not the whole time period covered by this study, and it also seems to measure more aspects of the rule of law. A possible disadvantage is that it is measured for the whole world, since this often means that there is no variance between the western states. However, in this case, the variation between the selected member states is sufficient. For these reasons, the study by Kaufman, Klaay and Mastruzzi will be used to operationalise the variable "Rule of law" in this study. The second variable in this group is "Administrative capacity/government efficiency". This is also a very broad variable, which is defined and used differently by different authors (Mbaye 2001; Börzel, Hofmann et al. 2004; Sverdrup 2004). Despite the differing definitions, the authors seem to attempt to measure the same thing, and operate on the same assumption, namely that the member states are in principle willing to abide by European rules, but that they also need the capacity to do this. In his study, Sverdrup assumes that states which have the capacity to handle political matters in general in an efficient way will also handle European matters in an efficient way. From a sociological institutionalist perspective, it can be argued that if the values and norms of the actors involved in transposition lead them to pursue timely transposition, they will do so if they have the ability. This is in contrast to rational choice institutionalist accounts, which assume that actors involved in transposition will attempt to delay the transposition of directives that are not in line with their interests. All three authors found that this variable had a negative effect on the number of infringement procedures that are initiated. However, just as for the variable "Rule of law", it has not yet been tested if this variable has an effect on the timeliness of transposition. The hypothesis to be tested is: H2: The more efficient a government is, the more timely transposition will be. Mbaye and Börzel et al. operationalise this variable by using measures of fiscal resources, characteristics of civil servants (level of education etc.), characteristics of bureaucracy (performance related pay, permanent tenure etc.) etc. Sverdrup, on the other hand, uses the World Bank study on Governance indicators to measure also this variable. In this study, government efficiency is defined as "quality of public service provision, quality of the bureaucracy, the competence of civil servants, the independence of the civil service from political pressures, and the credibility of the government's commitment to policies". Here, government efficiency is measured in a much broader sense, and is based on subjective judgements. For the purpose of this study, this latter definition and operationalisation has been chosen, since it captures more aspects of the variable and touches upon elements of administrative culture, such as the neutrality of the civil service. In the remainder of this study, this variable will be referred to as "Government efficiency". A third variable which could potentially be derived from sociological institutionalism is "Political culture", which has been tested by Lampinen and Uusikylä (1998). However, the challenge is how to define and operationalise this variable so that it is logically connected to transposition. Although Lampinen and Uusikylä have shown in their study that this variable, together with positive attitudes towards the EU, leads to a high overall willingness to comply with directives and thus leads to more timely transposition, they fail to clarify the causal mechanism behind this. In order to operationalise this variable, they use a number of indicators measured by various indexes, namely the legitimacy of political system (electoral participation), satisfaction of citizens with democracy, degree of social fragmentation, and respect of individual rights. If this operationalisation is used, it is difficult to see how political culture could have an effect on transposition. In addition, since transposition is mainly performed by civil servants, "administrative culture" might be a more suitable variable to use. If this variable is to be used, an operationalisation which connects more directly to transposition is thus needed. In the present study, the variable is not tested. However, the variable "Government efficiency" arguably overlaps in some respects with this variable. 4.3.2 Variables concerning the organisation of EU affairs in the member states The first variable which relates to the organisation of EU affairs is based on the two variables "Chinese Walls" and "National coordination capacity". The first of these two variables, suggested by Mastenbroek (2003), concerns differences between ministries regarding the existence of "Chinese walls" between stages of preparation and transposition. It is assumed that departments with a separate professional division responsible for negotiations transpose slower. The reasoning behind this is that if preparation is carried out by policy officials and transposition by legal experts, there is no way of anticipating legal problems, and this can cause problems in the transposition stage. The second variable is more general and concerns the coordination of common position in a member state among the whole sub-system of institutions (for example ministers, departments, local levels of government, agencies, institutionalised interest groups etc.) involved. This variable is tested in the study of Giuliani, and an effective system of coordination is expected to lead to better compliance. From the point of view of sociological institutionalism, the existence of "Chinese walls" is of great interest. If the civil servants that perform transposition are also involved at the stage of negotiation of a directive they are involved in a process of learning and socialisation, which makes them more positively inclined both towards the EU in general and towards the policy embodied in the directive. These processes are described by for example Checkel (1999). One could also argue more generally that the roles of the civil servants are shaped within this context and make them perceive transposition in general and transposition of a specific directive as important. Similarly, the most interesting aspect of "National coordination capacity" is if this system of coordination leads to the setting up of a transposition project group already at the negotiation stage. If this is the case, the transposition actors are likely to undergo the processes of learning and socialisation described above. The creation of a team, as opposed to assigning the responsibility to a single civil servant ensures continuity even if a civil servant changes position. Since the most interesting aspect of the variable "National coordination capacity" actually is the existence of "Chinese walls", these two variables will be treated as one in this study, and it will be referred to as "Chinese walls". Mastenbroek has tested the variable "Chinese walls" on transposition in the Netherlands, and the hypothesis was confirmed, although not all variance was explained. However, it has not been tested on transposition in the other member states. Giuliani uses the variable "National coordination capacity" in an analysis of residuals from a previous study, and while the analysis suggests that this variable has the expected effect, this variable has thus not been tested, in the strict sense of the word, on transposition. H3: The participation of the transposition actors in the negotiations on a directive leads to more timely transposition. In order to operationalise the variable "Chinese walls", Mastenbroek assumes that ministries with a large international division responsible for the negotiations and a separate legal division responsible for transposition are more likely to have this type of Chinese Walls. Information about this is found in Quarterly overviews of the Dutch Ministry of Foreign Affairs and the Staatsalmanak. Since this only measures differences between ministries in the Netherlands, this operationalisation cannot be used for the present study. In his study, Giuliani operationalises the variable "National coordination capacity" by roughly synthesising (only for 10 countries) qualitative observations, mainly made by Kassim et al. (2000). It is in principle possible to use a similar operationalisation based on secondary literature for the current research. However, the amount of work required in order to make a good comparison of the different member states in this respect that takes account of changes over time rendered this beyond the scope of this paper. The last variable, also proposed by Giuliani (2004) is "Involvement of parliament". It is expected that the involvement of the national parliament in the stage of the negotiation on a directive leads to swifter transposition since the democratic legitimacy of the process is ensured. This will make the actors responsible for transposition, sometimes including the parliament itself, more positively inclined towards transposition. In Giuliani's study, residual analysis suggests that this variable has the expected effect. H4: The more the national parliament is able to have an impact in the negotiations on a directive, the more timely transposition will be. In Giuliani's study, this variable actually consists of three variables, namely information available to parliament, presence and meeting frequency of parliamentary European Affairs Committees, and the impact of the parliamentary action on EU affairs. In able to operationalise this variable, Giuliani mainly uses a publication by European Centre for Parliamentary Research & Documentation (ECPRD), and other secondary literature. For this study, a similar operationalisation has been used. Each member state was given a score on the degree to which its parliament is able to have an impact in the negotiations on European legislation, the scope of the information that it receives regarding EC affairs, and the existence and frequency of meetings of a European Affairs Committee (EAC). Regarding information available to parliament, only information on matters concerning the first pillar was considered, since directives fall under this pillar. These scores are mainly based on the ECPRD report, although internal reports based on secondary literature on the characteristics of the five selected member states made within the project of which this research forms a part, have been used in order to verify these scores and fill in some blanks regarding the date of creation of the EACs. Although this variable is mainly measured at one point of time, it partly captures the historical development by including the date of creation of an EAC in the different member states. 5. Testing of explanatory variables on the dataset 5.1 Applied method Since the data is grouped by the categorical variable "Member state", it is necessary to control for this in the analysis, especially since the variables that will be tested are on the member state level. There is another grouping variable present, namely "Policy area", and it could also be useful to control for this variable although there are no explanatory variables specific to the policy areas in this study. If the grouped structure of the data is not controlled for, there is a risk of getting inflated Type I errors, and thus overestimating the significance of the explanatory variables. In this study, the effects of the grouping variables are controlled for by using the ANCOVA method, and including grouping variables as fixed factors, and the explanatory variables at the member state level as covariates. In principle, multiple regression could just as well have been used, since it is in fact the same method (Miles and Shelvin 2001). In this case, the advantage of ANCOVA is that it saves the trouble of creating dummy variables for the categorical variables. Since the dependent variable "Delay" does not show a normal distribution, but is positively skewed, the log of the dependent variable was taken. This resulted in an approximately normal distribution, apart from one extreme outlier with a standardised residual of –9.45. Since this outlier distorted the data, it was removed. The case was the energy directive 2001/77/EC, which was transposed 575 days before the deadline by the UK. 5.2 Results In a first step, it was simply tested if "Member state" had a significant effect on transposition delays, and this proved to be the case. Table 4 Analysis of variance on "Member state" Variable F Significance Member state 2.722 .031 R Squared = .052 (Adjusted R Squared = .033). Dependent variable: Log of delay in days Secondly, the variables "Member state" and "Policy area" were put together into a model, and they both showed a significant effect on the length of transposition delays. Table 5 Analysis of variance on the grouping variables Variable F Significance Member state 3.009 .019 Policy area 11.605 .001 R Squared = .104 (Adjusted R Squared = .081). Dependent variable: Log of delay in days Subsequently, in order to test the explanatory variables that were derived from sociological institutionalism while controlling for "Member state" these variables were included into a model as covariates, while "Member state" was included as a fixed factor. Table 6 Analysis of covariance on "Member state" and explanatory variables Variable F Significance Government efficiency 5.393 .021 Rule of law .664 .416 Parliament .079 .778 Member state .485 .747 R Squared = .089 (Adjusted R Squared = .056). Dependent Variable: Log of delay in days Table 7 Regression coefficients of the explanatory variables Variable B Government efficiency -.260 Rule of law .301 Parliament -.039 Dependent Variable: Log of delay in days The result of the analysis is that only the variable "Government efficiency" is significant. The fact that "Member state" is no longer significant shows that the effect of "Government efficiency" is a better predictor of transposition delay than simply in which member state a directive is transposed. However, the variables "Rule of law" and "Parliament" are not significant, and "Rule of law" even has an effect opposed to what was expected. The amount of variance explained by this model is about 6 %, and thus very low. As was apparent from the analysis of the two categorical variables, "Policy area" appeared to have a significant effect on transposition. In a next step, this variable was also included in the model. Table 8 Analysis of covariance on "Member state", "Policy area" and explanatory variables Variable F Significance Government efficiency 10.517 .001 Rule of law 2.490 .116 Parliament .006 .937 Member state .793 .531 Policy area 16.567 .000 R Squared = .160 (Adjusted R Squared = .125). Dependent Variable: Log of delay in days Table 9 Regression coefficients of the explanatory variables Variable B Government efficiency -.358 Rule of law .570 Parliament -.010 Dependent Variable: Log of delay in days In this model, "Policy area" has a highly significant effect. The variable "Member state" is not significant in this model either. Of the explanatory variables, "Government efficiency" is still the only one that is significant. This model explains about 12,5 % of the variance if the adjusted measure of R squared is used. The policy area to which the directive belongs is thus important for the length of transposition delays. However, although the amount of explained variance is much higher in this model, it is still not very high. 6. Conclusions and implications for further research In this paper, three variables that can be derived from sociological institutionalism have been tested using data on the transposition of 58 utilities directives in five member states in the period 1986 to 2002. These variables have previously been tested in studies of compliance with EC law with good results. However, if these variables are tested on the current data, controlling for the fact that the cases are grouped into member states, only the variable "Government efficiency" is significant. This shows that this variable has an effect on transposition delays that goes beyond the simple fact that the directives are transposed in a particular member state. However, the amount of variance that is explained by this model is very low. One reason for this low amount of explained variation could be that the explanatory variables used are very broad, and not directly related to transposition. This is not only true for the variables selected for this study, but also for the other variables that have been tested quantitatively using data on transposition or infringements. Another problem with the very broad variables typically used for quantitative studies in this area is that they measure so much that it is difficult to know what within this variable it is that really has an effect on compliance. Therefore, it does not seem useful to add more of these broad variables to a model that aims to explain transposition, but rather to try to operationalise these variables in a way that is more specific for the transposition process. Another result of the analysis is that the policy area (telecommunications or energy) is very important for the length of transposition delays. In the descriptive analysis, it was also shown that the member states differ in this respect. For some member states, the policy area to which a directive belongs appears to be very important for the length of transposition delays, and for other member states it seems less important. If the variable "Policy area" is included into the model, a larger amount of the variance is accounted for. Since the directives in the two policy areas are different in character, this could be one possible reason for this difference. However, that does not explain the differences between the member states in this respect. For further analysis, it would thus be useful to take the characteristics of the policy areas in the member states into account. It is thus clear that in order to answer the question of what the causes for delays in transposition are, it is necessary to take a step further and operationalise the explanatory variables in a better way. This could of course be done in case studies. However, the advantage of having selected only a few member states is that it might be feasible to operationalise the variables better even for a quantitative study. It might also be possible to include variables that were not operationalised in this paper, such as administrative culture and "Chinese walls". Ideally, these variables should be operationalised not on the general member state level, but on the level of the policy area. For example, the administrative culture in the ministry in the Netherlands that currently transposes utilities directives, which is the Ministry of Economic Affairs, might be different from that of other ministries. In addition, before the liberalisation of the utilities sector, the utilities directives were transposed in the Netherlands by the Ministry of Transport, and this ministry most likely has a different administrative culture than the Ministry of Economic Affairs. References Aspinwall, M. D. and G. Schneider (2000). "Same menu, separate tables: The institutionalist turn in political science and the study of European integration." European Journal of Political Research 38: 1-36. Aspinwall, M. D. and G. Schneider (2001). Institutional research on the European Union: mapping the field. The rules of integration: Institutionalist approaches to the study of Europe. M. D. Aspinwall and G. Schneider. Manchester, Manchester University Press: 1- 18. Börzel, T. A. (2000). "Why there is no "southern problem". On environmental leaders and laggards in the European Union." Journal of European Public Policy 7(1): 141-62. Börzel, T. A., T. Hofmann, et al. (2004). Why do states not obey the law? Non- compliance in the European Union. Paper presented at the workshop "Transposition and Compliance in the European Union", Oud-Poelgeest, Leiden, the Netherlands, 11-13 June 2004. Bursens, P. (2002). "Why Denmark and Belgium have different implementation records: On transposition laggards and leaders in the EU." Scandinavian Political Studies 25(2): 173-195. Checkel, J. T. (1999). "Social construction and integration." Journal of European Public Policy 6(4): 545-60. Checkel, J. T. and A. Moravcsik (2001). "Forum section - A constructivist research program in EU studies." European Union Politics 2(2): 219-249. DG Information Society (2004). 2004. DG Information Society, E. C. (1999). Status report on European Union electronic communications policy. 2004. Duina, F. (1997). "Explaining legal implementation in the European Union." International Journal of the Sociology of Law 25: 155-179. European Commission (2003a). Internal Market Scoreboard Nr 12, European Commission. 2003. European Commission (2003b). Internal Market: further deterioration in Member States' implementation of rules, European Commission. 2003: press release. European Commission (2003c). Electronic Communications: European Commission launches infringement proceedings against eight Member States. 2003: press release. European Commission (2003d). Electronic Communications: Commission takes further step in enforcement action against seven Member States. 2003: press release. European Commission (2004a). Electronic Communications: Commission takes second legal step against eight Member States for not adopting new privacy rules for digital networks and services. 2004: press release. European Commission (2004b). Six Member States face Court action for failing to put in place new rules on electronic communications. 2004: press release. European Commission (2004c). Opening up of energy markets: 18 Member States still have to transpose the new EU rules. 2004: press release. European Commission (2005). Second implementation report of the Internal Market Strategy 2003-2006. Gibson, J. L. and G. A. Caldeira (1996). "The legal cultures of Europe." Law & Society Review 30(1): 55-85. Giuliani, M. (2003). Europeanization in comparative perspective: Institutional fit and national adaptation. The politics of Europeanization. K. Featherstone and C. M. Radaelli. Oxford, Oxford University Press: 134-158. Giuliani, M. (2004). "EU compliance: Macro, meso, micro or no institutions at all?" URGE Working Papers(6/2004). Hall, P. A. and R. C. R. Taylor (1996). "Political science and the three new institutionalisms." Political Studies 44: 936-957. Haverland, M. (1999). National adaptation to European integration: The importance of institutional veto points. Florence, European University Institute. Hunt, A. (1997). "Regulation of telecommunication: the developing EU regulatory framework and its impact on the United Kingdom." European Public Law 3(1): 93-115. Kassim, H., G. B. Peters, et al. (2000). The national co-ordination of EU policy: The domestic level. Oxford, Oxford University Press. Kaufmann, D., A. Kraay, et al. (2003). "Governance matters III: Governance indicators for 1996-2002." World Bank Policy Research Working Paper 3106. Knill, C. and A. Lenschow (1998). "Coping with Europe: the impact of British and German administrations on the implementation of EU environmental policy." Journal of European Public Policy 5(4): 595-614. Lampinen, R. and P. Uusikylä (1998). "Implementation deficit - Why member states do not comply with EU directives." Scandinavian Political Studies 21(3): 231-251. Lowndes, V. (2002). Institutionalism. Theory and methods in political science. D. Marsh and G. Stoker. New York, Palgrave Macmillan: 90-108. March, J. G. and J. P. Olsen (1984). "The New Institutionalism: organizational factors in political life." The American Political Science Review 78(3): 734-49. March, J. G. and J. P. Olsen (1989). Rediscovering institutions: The organizational basis of politics. New York, The Free Press. March, J. G. and J. P. Olsen (1996). "Institutional perspectives on political institutions." Governance 9(3): 245-64. Mastenbroek, E. (2003). "Surviving the deadline: The transposition of EU directives in the Netherlands." European Union Politics 4(4): 371-395. Mbaye, h. a. d. (2001). "Why national states comply with supranational law: Explaining implementation infringements in the European Union, 1972-1993." European Union Politics 2(3): 259-281. Miles, J. and M. Shelvin (2001). Applying regression & correlation: A guide for students and researchers. London, SAGE Publications. Nunn-Price, N. (1997). "Improving access to Community legislation and its transposition into national law." The Journal of Information, Law and Technology (JILT)(3). Peters, G. B. (1999). Institutional theory in political science - The "New institutionalism". London and New York, Pinter. Powell, W. W., P. J. DiMaggio, et al. (1991). The new institutionalism in organizational analysis. Chicago, The University of Chicago Press. Risse, T. (2002). Constructivism and international institutions: Toward conversations across paradigms. Political science: The state of the discipline. I. Katznelson and H. V. Milner. New York, W. W. Norton & Company: 597-623. Sauter, W. (1995). "The telecommunications law of the European Union." European Law Journal 1(1): 92-111. Siedentopf, H., J. Ziller, et al. (1988). Making European policies work: The implementation of Community legislation in the member states. London, Sage Publications Inc. Sverdrup, U. (2004). Compliance: A matter of ability? Paper presented at the workshop "Transposition and Compliance in the European Union", Oud-Poelgeest, Leiden, the Netherlands, 11-13 June 2004. Thatcher, M. (2001). The national politics of European regulation: institutional reform in telecommunications. Utilities reform in Europe. D. Coen and M. Thatcher. New York, Nova Science Publishers: 11-29. Thatcher, M. (2001a). "The Commission and national governments as partners: EC regulatory expansion in telecommunications 1979-2000." Journal of European Public Policy 8(4): 558-584. Thelen, K. (1999). Historical institutionalism in comparative politics. Annual reviews of political science. Palo Alto, Annual Reviews. 2: 369-404. Wendt, A. (1987). "The agent-structure problem in international relations theory." International Organization 41(3): 335-70. Wendt, A. (1992). "Anarchy is what states make of it: the social construction of power politics." International Organization 46(2): 391-425. Wendt, A. (1994). "Collective identity formation and the international state." American Political Science Review 88(2): 384-96. Acknowledgements: I would like to thank the Netherlands Organisation for Scientific Research (NWO) for financial support for the project "Analysing the transposition of EU directives in the area of utilities (energy and telecom) regulation". This project is part of a larger project called "Analyzing European Union policies: the transposition of directives" in which transposition in four different policy areas, namely utilities, transport, food regulation and social policy, is examined. I would like to thank the members of the research group of this project, namely Frans van Waarden, Bernard Steunenberg, Kees van Kersbergen, Antoaneta Dimitrova, Markus Haverland, Ellen Mastenbroek, Mark Rhinard, Ieva Gange, Marleen Romeijn and Michael Kaeding, for their support and the useful discussions we have had on this research. Strictly speaking, the assumption of equal variances is not acceptable, since the Levene's test is significant (p = 0,045). However, since this test is only just significant, the results should still be reasonably reliable. The reason for this is probably its strong correlation with "Government efficiency", since it does have the expected direction if it is tested separately with "Member state" as a fixed factor. 1