WHEN THE EU HITS HOME: NATIONAL DETERMINANTS OF NON-COMPLIANCE WITH THE EUROPEAN UNION LAW

This study analyzes 25,516 cases of violation of the European Union law by 28 Member States from 1993 to 2018. I strive to outline the national level determinants of differentiation in the pool of member countries by the total number of the EU law violations. As a key method of analysis, logistic regression is used, where factors of GDP per capita (PPP), polarization of the parliament, fragmentation of the party system, regional strategies and quality of governance are used as country attributes. The analysis demonstrates that all country attributes are significantly related to all four quartiles of the outcome, which rank member states depending on the number of violations during the period under review: from the smallest share of violated directives (Q1) to the largest share of violated directives (Q4). The results of the study demonstrate the empirical relevance of the theoretical perspective of “worlds of compliance” formulated by G. Falkner et al. (2007) for the categorization of EU member states in their reactions to the compliance efforts of the EU.


Introduction
The main goal of the European Union (EU) in the process of expanding and unifying its legal space and setting its authority and power is to force the EU Member States (MS) to transpose, implement and apply the parts of the acquis communitaire. In the academic literature this enforcement and the reaction of the EU states to the enforcement actions of the EU are labeled under the term "compliance with the European Union law" (Thomson, 2007;Schimmelfennig, 2007;Sedelmeier, 2012;Börzel & Sedelmeier, 2017). The desire of the EU to force all member countries to comply with the EU law is determined by two reasons. The first reason is that the unified normative space is directly related to the idea of creating a functioning supranational EU governance (Banchoff & Smith, 2005). The second reason is the EU's commitment to the ideological position "one size fits all", which implies not only normative, but also political, economic and cultural unification of the EU states (Börzel & Risse, 2004). It is clear though that the different degree of compliance of the member countries with the EU law becomes an obstacle to the implementation of the "united Europe" (Swedberg, 1994;Smith, 2008) project.
Hypothetical expectations related to the likelihood of multi-level governance problems within the European Union were confirmed empirically. Since the mid-1990s, researchers track the growing differentiation in the pool of the EU Member States in terms of their compliance with the European Union law. Over the past 25 years, countries such as Estonia, Lithuania and Slovenia have been incorporating parts of the EU legislation with minimal violations (Tallberg and Jönsson, 2001;Goetz, 2004;Verheijen, 2007;, while France, Italy and Portugal demonstrate failures in this process almost every year (Pollack, 2006;Thomson, 2007;. This study targets state autonomy and state capacity as two potential explanations for such a differentiation in the level of compliance of the EU Member States. The paper is divided into four sections. The first section looks at two basic approaches that strive to reveal the reasons for non-compliance in the European Unionstructural and consociational. The same sec-tion focuses on the factors of state autonomy and state capacity of the EU Member States, which underlie the strategies of voluntary and involuntary (non-)compliance. The second section describes the study design, variables' encoding and the stages of modelling. The third section presents the results of logistic regression. The fourth section is devoted to the interpretation of the results obtained and gives the conclusions of the study.

Structural and Consociational approaches to the Problem of Non-Compliance
The empirical puzzle associated with different progress of EU member countries in the implementation, transposition and application of the acquis is the subject of controversy in two theoretical approaches: (1) structural and (2) consociational.
In the models build by the proponents of structural approach, the advantage of the "regulatory dictate" is always on the side of one of the parties of the two-level game, while the other side is forced to follow the competitor's rules. P. Haas (1998), J. Tallberg (2002), A. Slaughter et al. (2000) stress that the EU acts as a "normative dictator" creating the uniform rules of the game for all Member States and monitoring their observance. It is the choice of the states whether to follow these rules, still the EU has an ultimate right and monopoly to punish the violators of the created "normative ideal" (Checkel, 2001;Heidbreder, 2011). An alternative approach to exploring the (non-)compliance patterns in the EU reverses the observed hierarchical model and highlights the EU Member States and their behavioral strategies. The proponents of this approach stress the ability of national political actors to find an equilibrium point that would maximize the gains from adopting EU's rules of the game and use the available resources to approximate the result of "bargaining" to this point of equilibrium (Mendrinou 1996;Paraskevopoulos 1998). In both versions of the structural approach, the state's decision on the preferred compliance strategy and the final motivation of (non-)compliance is determined by its economic and administrative resources and the efficiency of their distribution and allocation at the national level (Börzel, 2002;Knill & Lenschow, 2005).
Structural approach is often criticized for reductionism and hierarcisation of relations between the EU Member States and supranational institutions. According to W. Carlsnaes (1992: 247), issues related to multi-level authority and accountability, are not limited to clarifying the relationship of the "primacy" of a particular level of decision-making in the structure of the European Union. M. Gabel (1998), B. Soetendorp and R. Andeweg (2001), M. Bogaards (2002), P. Kraus (2006 and many other authors demonstrate that the decision-making architecture in the EU is closest to the model of consociational democracy, which was theorized by A. Lijphart (1971). M. Burgess (2002) emphasizes that the decision-making process in the European Union is based on cooperation mechanisms involving both the intergovernmental and supranational level. The focus of research on compliance within the framework of the consociational approach, is the principle of subsidiarity. The principle of subsidiarity, first formalized in the Treaty on European Union (TEU), explicitly states that if a question can be resolved at a certain level of decision-making, it should not be transferred to a higher level of authority. The consociational approach takes into account the influence of national actors and institutions upon (non-)compliance of the EU member countries and describes the struggle of national actors for an acceptable legal status quo (Börzel, 2002).
Thus, the structural and consociational approaches offer two broad explanations for the different levels of compliance with the EU law. For the structural approach, the reason for the compliance failures in some countries is that they are not able to follow the Brussels' guidelines due to a lack of necessary economic, administrative and bureaucratic resources. The consociational approach, in turn, considers the reasons for non-compliance with the EU law through the inability of national actors to reach a compromise on the compliance agenda. The next section describes two key concepts that underlie these two explanationsstate capacity and state autonomy.

State Capacity and State Autonomy in the Context of (Non-)compliance
State capacity emphasizes the importance of the amount and diversity of resources available, as well as their allocation by the national institutions and actors for the issues related to compliance. For example, J. Tallberg (2002), T. Verheijen (2007), A. Dimitrova (2011) and C. Kaya (2019) allocate administrative capacity as the basis for successful compliance. Using a set of 27 EU countries, E. Thomann and A. Zhelyazkova (2017) show that in most cases countries with high administrative capacity and professional bureaucracy adopt the EU directives timely with no or almost no violations.
The alternative concept is state autonomy which dominates the works of consociationalists. Most of the researchers that seek to identify the reasons for differentiation in the pool of the EU countries in terms of compliance understand state autonomy in T. Skocpol's (Skocpol, Evans and Rueschemeyer, 1999) terms, that is as the ability of actors in power to make decisions and form political agenda, abstracting from the potential influence of other actors and groups. Students of (non-)compliance measure state autonomy of the EU states in the number of veto-players, which participate in the process of transposition, implementation and application of the required directives (Hille & Knill, 2006;Noutcheva & Bechev, 2008;Sedelmeier, 2012). Political parties and national legislatures as a whole play an important role in the works of the state autonomy scholars. According to P. Taggart and A. Szczerbiak (2004), the parliament is the "guardian of national interests", which ratifies international agreements and makes changes in national legislation. D. Finke and T. Dannwolf (2015) stress that over the past 15 years, issues that relate to the European Union have become extremely politicized at the national level, and therefore the discussion of these issues often leads to a clash of values and interests of large parliamentary groups.
T. Börzel (2001) raises the important topic of bargaining between national and subnational levels of governance during the application and enforcement of the EU directives. Using the examples of Spain and Germany, the author argues that the success of compliance directly depends on the ability of national elites to negotiate with the regions and provide them with compensation for the costs that the regions incur due to adaptation to the changed normative reality and the creation of the necessary infrastructure. Börzel's idea is developed by N. Dörrenbächer (2017) and J. Schmälter (2018), who note that the regional elite and the actors of the national administrative system ensure "street-level compliance", that is, the implementation and application of directives on the ground.
H. Klüver (2009) and C. Kaya (2019) research the special role of interest groups and NGOs in the (non-) compliance process. The first role of interest groups and non-government organizations is to act as watch dogs for the European Commission: they monitor the actions of national authorities in the area of compliance and inform the normative regulator of any violations by a member state that were made at an early stages of compliance. Thus, theorists of state capacity and state autonomy propose two scenarios of non-compliance. The first scenario is caused by the inability of the Member States to follow the rules of the game established by Brussels, due to the lack of effective bureaucracy, financial and administrative resources. The second scenario indicates the reluctance of countries to comply with the EU law, which is fueled by their large political and economic resources and the absence of consensus among the leading national actors and groups on the compliance agenda. In the academic literature these two scenarios are labelled under the names of involuntary and voluntary non-compliance, respectively (Börzel, Hofmann, & Sprungk 2004;G. Falkner et al. 2004; Jano 2016; Thomann & Sager 2017).

Voluntary and Involuntary Non-compliance in the EU
One can identify the cornerstone reason for voluntary non-compliance with the EU law: a powerful economy that is able to neutralize the consequences of potential financial sanctions for non-compliance (Haas 1998). Thus, countries choose a strategy of voluntary non-compliance if they can avoid potential sanctions for non-compliance or if the cost of the sanctions imposed is insignificant for them.
States that voluntarily comply with the European Union law are the exact opposite of the countries described above. According to G. Falkner et al, "small countries with effective bureaucracy" (2004: 412) are the best compilers. In other words, good compliance in this group of states is provided either by a welldeveloped administrative apparatus, which allows countries to reduce costs from adopting new rules of the game or economic non-readiness to take on the costs of non-compliance, and thereby choose compliance as the most profitable alternative.
Involuntary non-compliance, in turn, is determined by the absence of a baseline characteristic, which is noted as a key condition for successful transposition, implementation, and application of parts of the acquis. This characteristic is an effective bureaucratic system. According to C. Knill and J. Tosun (2009), D. Toshkov (2007) and K. Staroňová (2013), the weakness of the bureaucratic system lies at the heart of most compliance failures. B. Steunenberg and D. Toshkov (2009), conducting a statistical study of non-compliance of the EU-15 and EU-10 countries, highlight the ineffectiveness of bureaucracy as a significant variable for explaining involuntary non-compliance. Another reason for involuntary non-compliance may be low state autonomy, which does not allow national actors to come to a consensus on the issue of compliance (Tallberg, 2000;Chandler, 2010).
The strategies to respond to compliance requirements are most fully explained by G. Falkner et al. (Falkner, Hartlapp, & Treib 2007). The authors identify three "worlds of compliance": "world of law ob-servance", "world of domestic politics" and "world of transposition neglect." In the "world of law observance", one can find states, for which "compliance goal overrides domestic concerns" (Falkner, Hartlapp, & Treib 2007: 405). Such countries rarely show non-compliance and quickly correct violations of the EU law. "World of domestic politics" includes countries for which home policy has an extremely important value and "obeying EU rules is at best one goal among many" (Falkner, Hartlapp, & Treib 2007: 405). Such countries often face problems of non-compliance and risk to fail transposition, implementation, and application of the required part of the acquis if it directly contradicts national interests. Finally, countries from the "world of transposition neglect" show an absolute lack of interest in the issue of compliance. They begin to act only after the Commission and the ECJ begin the process of infringement and often only imitate the implementation and application of directives (Falkner, Hartlapp, & Treib 2007: 405).
The next section details the analytical steps to identify the national-level determinants of the strategies of voluntary and involuntary (non-)compliance of the EU Member States. The analysis tends to check the relevance of G. Falkner et al.'s assumptions on the compliance behavior strategies of the EU countries.

Study Design
In order to analyze the influence of factors of state autonomy and state capacity on (non-)compliance of the EU member countries with the EU law, I take the time period from 1993 to 2018. The choice of this period is due to historical reasons. In 1992 the Maastricht Treaty was signed and in 1993 it entered into force. This treaty endowed the EU institutions (European Commission, European Court of Justice and the Council of Ministers) with great powers in imposing sanctions and punishments against countries that do not comply with the EU law (Lampinen & Uusikyla, 1998). In turn, 2018 is chosen as the end point, since the countries of the European Union continue to violate the EU law, and for some countries the number of violations increases from year to year (Börzel, Hofmann, & Panke, 2012) therefore it is important to trace the dynamics of these violations. At the same time, I strive not to reduce the sample, and therefore do not consider the period, when the UK's compliance with the EU law has not been monitored (2019-2020). The total number of observations is 25,583 (total number of the EU law violations), but there is a differentiation at the level of individual states, which is due to the different period of their EU membership. As a key method of data analysis, I use multi-level logistic regression.

Stage 1: Outcomes
The outcome variable share of violated directives of the European Union is ordinal and contains the following categories: 1. Q1 (best compliers): 0.1%-2.0% of total number of the EU directives 2.
Q4 (worst compliers): >8.0% of total number of the EU directives To encode this variable, I use the data of the Annual Reports on Monitoring the Application of Community Law by each country for the period from 1993 to 2018, as well as the Berlin Infringement Database (Börzel & Knoll, 2012) 1 . The percentage of violated directives was calculated for each member state individually from the date of the state's accession to the European Union. The thresholds for quartiles (Q1-Q4) are set in accordance with the thresholds used in the annual reports of the European Commission on monitoring of the application of EU law, which track the progress of member countries in the area of compliance.

Stage 2: Country Covariates
The pool of state covariates chosen for analysis is divided into two groups, correlated with the two concepts to be explored in this paper: state capacity and state autonomy.
Consider the factors of state capacity. The level of economic power is one of the most important factors determining the state's ability to provide a sufficient basis for a correct and rapid transposition and implementation of the acquis . Harmonization of the national legislation with the EU law often requires large expenditures of the national budget. It can be assumed that compliance process for richer member countries is less burdensome than for poorer ones, so they comply faster and with fewer violations. The factor of quality of governance is treated as the quality of public institutions. I assume that countries where bureaucratic institutions are better developed have less problems with the implementation of the parts of acquis and spend less time and resources on the compliance program.
The variable GDP per capita (PPP) is encoded using World Bank Open Data (2018). It is an index that includes three categories: high GDP per capita (value "1"), medium GDP per capita (value "2"), and low GDP per capita (value "3"). The covariate quality of governance is encoded using the Worldwide Governance Indicators (2018) data. It is the logarithmized index that includes three categories: high quality of governance (value "1"), medium quality of governance (value "2") and low quality of governance (value "3").
The second group of factors is devoted to state autonomy: (1) ideological polarization of the parliament, (2) fragmentation of the party system, (3) strategy of regional authorities. P. Statham and H. Trenz (2015) argue that the boundaries of ideological polarization often coincide with the dividing lines in the national parliaments on (non-)compliance with the EU law. I assume that the conflicts associated with the implementation of the parts of European Union legislation are integral parts of the full-scale debate between the parties in the parliaments of EU Member States.
The ideological polarization factor is encoded using data from the Manifesto Project (2018). This index has three categories: high polarization (value "1"), medium polarization (value "2") and low polarization (value "3"). The thresholds are determined based on the categorization provided by the Manifesto Project: 1, 0.5 and 0, respectively.
To encode the fragmentation of the party system parameter, I calculate the effective number of parties (ENP) for each country on a yearly basis, starting from the moment the country joined the European Union. The fragmentation factor has three categories: high fragmentation (value "1"), medium fragmentation (value "2") and low fragmentation (value "3"). The thresholds are defined as follows: value "1" -ENP of 5.5, value "2" -ENP of 3.5, value "3" -ENP of 2.5 and correspond to the thresholds, widely recognized in the literature on the party systems fragmentation (Laakso & Taagepera, 1979;Golosov, 2010).
The parameter of the strategies of regional authorities is based on data from transcripts of the Committee of the Regions plenary sessions and Monthly Summary of Council Acts, released by the Secretariat of the EU Council of Ministers. This is a binary variable with the following categories: "1"presence of a conflict between the center and the regions on a specific compliance agenda (confrontation strategy), "0"absence of a conflict between the center and the regions on a specific compliance agenda (cooperation strategy). Each member states appoints a specified number of delegates to the Committee of the Regions, which purpose is to represent the interests of regional and local communities at the supranational level and to provide recommendations to other EU institutions for decisions touching the regional level. The Committee does not have much normative weight in the EU architecture, however, it can serve as an arena for expressing dissatisfaction of regional authorities with the actions of the central/federal authorities and declaration of the existing conflict between the center and the regions. It can be assumed that a mismatch between the interests of regional stakeholders and the central government can lead to specific compliance violations due to the resistance of "state-level bureaucracy" to the compliance agenda, enforced by the central authorities (Dörrenbächer, 2017;Schmälter, 2018). As part of the encoding, a country received a value of "1" if at least once during the year under review, the vote of the majority of the deputies of its national delegation in the Committee of the Regions on a given directive differed from the vote of the representative (national minister) of this state in the Council of Ministers on the same directive.

Statistical Analysis
The statistical analysis was performed using RStudio 1.3. For bivariate associations between state autonomy and state capacity variables and the frequency of violations of the EU law, I calculated the percentages of violations in the total pool of the EU directives for the period from 1993 to 2018 for each state under consideration and performed analyses using the chi-square test. Analyses were performed for all samples, and were also stratified by geographical dimension West-East (EU-15 vs post-2004 members). I estimated the adjusted odds ratios (ORs) and 95% confidence intervals (CIs) for the total share of the EU law violations by random intercept, logistic regression models. All outcomes were grouped by quartile and included simultaneously in the model.

Results
The pool of EU Member States is differentiated by the total number of violations of the EU law for the entire period under consideration . Figure 1 serves to better illustrate this cleavage. In this section, I use the theory by G. Falkner et al. (2007) in order to trace and interpret the reaction of the EU member countries to the European Union's enforcement in the field of compliance. I consider EU states from two perspectives: (1) state capacity and (2) state autonomy. The state capacity perspective allows to identify the institutional and administrative characteristics of the states, which contributed or, conversely, impeded the transposition, implementation and application of the acquis. The state autonomy perspective, in turn, touches on the issues of the ability of national actors to promote, block or adjust the compliance agenda.  Table 1 summarizes the characteristics of state capacity and state autonomy of the EU countries. In total, Western EU countries violated 6.9% of EU directives from May 1993 to May 2018. For the countries of Eastern Europe for the period from May 2004 to May 2018 this figure is 4.5%. Both indicators are weighted. Bivariate analysis shows that all country covariates, that is, GDP per capita (PPP), polarization of the parliament, fragmentation of the party system, regional strategies and the quality of governance, affect the share of violated EU directives. However, it should be noted that the factor of fragmentation of the party system for the countries of the EU-East group does not demonstrate significant relation to the outcome. The results of the logistic regression with the outcome share of violated directives of the European Union are presented in Table 2. All attributes are related to the outcome. If one considers the results in terms of the outcome's quartiles, the least number of violations was likely to be committed by countries with medium GDP per capita (PPP), low legislature's polarization and fragmentation of the party system, low share of conflicts between the center and regions, and high quality of governance. The individual attributes for the second quartile are largely similar to the first one: medium GDP per capita (PPP), medium parliament's polarization and low fragmentation of the party system, cooperation strategy of the regions, and high quality of governance. The third quartile is related to the following individual country attributes: high GDP per capita, high polarization of the parliament and fragmentation of the party system, priority of cooperation in the relationships between the regions and the center, and high quality of governance. Finally, the fourth quartile, which denotes the worst compliers, is related to low GDP per capita, medium polarization of the national legislature, high fragmentation of the party system, conflict strategy of the regions, and average quality of governance.

Discussion and Conclusion
In this research, I used the combination of structural and consociational approaches to study the phenomenon of (non-)compliance. The major reason for this symbiosis is that I explored different contexts of (non-)compliance: both purely hierarchical relations and consociational models of communication. The theoretical perspective of "worlds of compliance" construct four quartiles of compliance.
The first quartile unites the "best compilers": Cyprus, Estonia, Latvia, Lithuania, and Slovenia. The reasons for their high compliance rate lie in the high state capacity, that is, the well-functioning bureaucracy and developed administrative apparatus and high state autonomy, which, first of all, is expressed in the Eurooptimistic position of the major veto players and cooperation strategy of the regional elites.
The second quartile unites Bulgaria, Croatia, Czech Republic, Poland, Romania and Slovakia. The key determinants of violations of the EU law in these countries are weak bureaucracy and relatively low level of economic power (Bulgaria and Romania), high polarization and fragmentation of national legislatures along the ideological and compliance lines (Czech Republic, Croatia and Slovakia), as well as conflicts between the center and regions on the distribution of say and pay rights in the context of compliance (Poland).
The third quartile compliance is the largest one: it unites Austria, Denmark, Finland, Greece, Hungary, Ireland, Luxembourg, Malta, the Netherlands, Portugal and Sweden. All countries of the Third world are distinguished by a fairly well-developed bureaucratic and administrative systems and, in most cases, noncompliance at the national level is explained by low state autonomy: Euro-skeptic position of the major parties, intra-party conflicts on the compliance agenda and conflicts between the center and the regions.
The fourth quartile includes the main violators of the EU law: Belgium, Germany, France, Italy, Spain and the UK. The countries of the fourth quartile assess the gains and losses from compliance and noncompliance with the EU law in each particular case, and often consider the potential sanctions for violating the EU law as less costly than adopting the acquis. The national determinants of non-compliance include the conflict nature of relations between the regions and the center (Spain, Germany and Belgium) and the general lack of political will, reinforced by party cleavages on the compliance agenda (France, Italy and the UK).
From this it follows that four quartiles of the EU states can be divided into three worlds, designed by G. Falkner et al. (2007): a world of law observance (first quartile), a world of transposition neglect (part of the second and third quartiles, fourth quartile) and a world of domestic politics (part of the second and third quartiles).
The symbiotic research seems to be effective when it comes to the issues of multi-level governance in the European Union. The generalizability and comparability of the results can be enhanced by performing a study with a wide range of factors of national and supranational nature and expanding the pool of cases to the EU candidate countries and specific types of the EU law violations.