Compliance with EU Law in Times of Disintegration: Exploring Changes in Transposition and Enforcement in the EU Member States between 1997 and 2016

This article explores whether disintegration, in the form of increased non&#8208;compliance with EU law, has occurred in the EU Member States between 1997 and 2016. The contribution is twofold. First, we develop hypotheses to test the argument that non&#8208;compliance with EU law increases when Member States experience growing economic constraints and political turbulence. However, the hypotheses are conclusively rejected. Second, while we demonstrate distinct differences between Member States' compliance by the end of the 1990s, our time&#8208;series analyses demonstrate that these disparities disappeared over time. Taken together, our findings not only suggest that economic factors, government capacity, and, above all, domestic politics become gradually decoupled from the European integration process but also that cultural differences in law observance become increasingly irrelevant to explain the behaviour of governments. We conclude by arguing that the strong trend towards harmonization probably is explained by a changed culture of EU transposition and enforcement.


Introduction
'We continue to be very concerned by the developments in some of our Member States …. The European Union is a community of law. Respecting the rule of law and abiding by Court decisions are not optional' (Juncker, 2018, p. 11).
In his 2018 State of the Union speech, Jean-Claude Juncker used these forceful words to underscore that the rule of law is at risk in the European Union (EU). While Juncker specifically addressed the political developments in Poland and Hungary, European elites appear to have reasons for deeper concerns. The EU has been exposed to multiple crises that led to changes in the EU political system over the last decades (Bickerton et al., 2015;Falkner, 2017;Schmitter, 2012). EU Member States have not only suffered from a profound economic recession, but they have also experienced growing political turbulence (Bickerton et al., 2015, p. 710). The warning signs are visible across Europe, as radical right parties with Eurosceptic agendas gain momentum (Vasilopoulou, 2018, pp. 135-137). This process of political fragmentation and Euroscepticism appears to have reached its peak with Brexit.
In light of these developments, a growing volume of literature anticipates that economic and political turbulence will spark a process of European disintegration (Saurugger The work behind this article is equally divided between the authors. However, BP is the lead author because the article is built on her research ideas. 2014, p. 13) intensify when Member States experience growing economic constraints and political turbulence.

II. Partial Exits in the Wake of Economic and Political Turbulence?
After the economic crisis in 2008, administrative authorities in many EU Member States experienced work overload and financial cuts. This, in turn, led to decreased government capacity (Council of Europe, 2014). We argue that this development may have diminished the capability of some Member States to implement EU law in a correct and timely fashion. We base this expectation on previous research in the field. On the one hand, a recent study shows that Member States' compliance with EU law depends more on administrative capacity than on political willingness (Börzel and Buzogány, 2019, p. 335). On the other hand, several studies provide evidence that greater administrative capacity is essential to counteract transposition delays and non-compliance (Börzel et al., 2010;Hartlapp, 2009;Haverland and Rumeijn, 2007;Linos, 2007). Accordingly, if the economic crisis primarily undermined the administrative capacity in some Member States (Council of Europe, 2014), the increased variation among EU Member States suggests that the importance of government capacity for predicting compliance with EU law has increased over time. This leads to our first hypothesis.
Hypothesis 1: (H1): Lower degrees of government capacity has increased the probability of partial exits in the form of non-compliance with EU law, particularly in the wake of the 2008 economic crises.
With regard to political factors that drive disintegration, scholars argue that partisan realignment and growing Euroscepticism in the Member States may threaten the survival of the EU as a whole (for example Hurrelmann et al., 2015). Building on two streams of literature, we hypothesize that these developments increase the likelihood of partial exits in the form of increased non-compliance with EU law. First, a widespread view held in research is that the smooth implementation of the acquis communautaire depends on a high degree of consensus between national political elites (Bartolini, 2005;Hooghe and Marks, 2009). However, if successful compliance with EU directives indeed is rooted in elite consensus (Bartolini, 2005;Mair, 2013), we argue that it is plausible to expect increased non-compliance with EU law as European party systems crumble and dominant parties lose momentum. This forms the basis of our second hypothesis.
Hypothesis 2: (H2): The fragmentation of party systems implies a growing strength of Eurosceptical parties and a rise in veto points, which undermines elite consensus on the EU and therefore increases the probability of partial exits in the form of non-compliance with EU law.
In addition, several authors argue that the 'permissive consensus' of the past has gradually been replaced by a 'constraining dissensus' (Down and Wilson, 2008;Marks, 2005, 2009). A consequence of this transformation is that European integration has become increasingly politicized in elections and referendums, which implies that the preferences of the somewhat Eurosceptic public have become important for jurisdictional outcomes (Hooghe and Marks, 2009;Green-Pedersen, 2012;Hutter and Grande, 2014;Treib, 2014, p. 24). To exacerbate the situation, the growing electoral strength of the highly Eurosceptic radical right throughout Europe (Usherwood and Startin, 2013;Vasilopoulou, 2018) is often viewed as a threat to the EU. Against this background, our third hypothesis states the following hypothesis.
Hypothesis 3: (H3): The more prominent Eurosceptic opinions are in a national party system and the stronger Eurosceptic radical right parties become, the higher the probability of partial exits in the form of non-compliance with EU law.

III. Do Cultures of Compliance Enhance Differences in Non-compliance?
While Hypotheses 1 to 3 focus on factors that may explain an overall trend towards disintegration, our fourth hypothesis highlights the possibility that national cultures of compliance determine such trajectories. Our hypothesis builds on the literature that addresses the so-called four worlds of compliance (Falkner et al., 2005, pp. 317-341;Falkner et al., 2007. Although previous research has raised doubts regarding this typology (Thomson, 2009;Toshkov, 2007;Treib, 2014, p. 12), we argue that it remains the primary categorization of salient divisions between the compliance cultures of Member States.
First, in the 'world of law observance' (Denmark, Finland and Sweden), Falkner and her colleagues argued that a culture of respect for the rule of law ensured timely and correct implementation (for example Falkner et al., 2007, p. 405). If this culture indeed is decisive for legislative outcomes, it indicates that the patterns of compliance most likely are resistant to change. Second, the compliance culture in the 'world of transposition neglect' (France, Greece, Luxembourg and Portugal) differs immensely as administrative authorities in these countries are often stymied by inactivity or inertia towards EU implementation. This, in turn, traces back to a certain degree of national arrogance, administrative overload, or inefficiency (for example Falkner et al., 2007, pp. 405-406). Given that administrative capacity decreased significantly in the wake of the economic crises (Council of Europe, 2014), it is therefore plausible to expect that the already poor track record of implementation has further deteriorated over time. Third, in the 'world of domestic politics' (Austria, Belgium, Germany, the Netherlands, Spain and the UK), Falkner and her colleagues argued that policy makers primarily implement directives in a correct and timely manner if EU provisions do not clash with domestic politics and interests (for example Falkner et al., 2007, p. 405). We argue that it is particularly plausible to expect a decreased willingness to implement EU law in this world of compliance in the wake of party system fragmentation and the rise of Eurosceptic opinions and parties. Lastly, in the 'world of dead letters' (Ireland, Italy, Czech Republic Hungary, Slovakia and Slovenia), implementation appears to be evident in the written law, but often it is not effective in practice (Falkner et al., 2008, pp. 169-173). As the economic crisis hit these countries particularly hard, we maintain that it is likely that one can expect a worsened implementation performance.
In summary, we argue that growing economic and political asymmetries between countries over time (see Jones, 2018, p. 449) most likely enhanced the culturally induced compliance cleavages among Member States. This leads to our fourth hypothesis.
Hypothesis 4: (H4): Over time, the differences between culturally induced patterns of compliance among EU member are further enhanced.

IV. Data and Methods
If disintegration is a continuing trend, we argue that we should see its effects within the Single Market. The reason for this is twofold. First, the area is inherently highly integrated and it constitutes the core of the EU project. Second, the Single Market reflects two contradictory developments during economic and political turbulence. On the one hand, the European Commission strives to widen markets and boost competition. On the other hand, increased Eurosceptic agendas in some Member States have intensified protectionist measures (Financial Times, 2014). This clearly indicates that the Single Market is increasingly prone to partial exits.
An additional reason for focusing on this area is the EU Single Market Scoreboard, which is thus far the most comprehensive database for evaluating the implementation performance of Member States, as the availability, comparability, and the stable compilation of the data over time is especially pertinent. This is despite the fact that the Commission's data does not detect each case of non-compliance (Hartlapp and Falkner, 2009;Haverland and Rumeijn, 2007;Mastenbroek, 2003). The database includes all directives pertaining to the Single Market over a 20-year period (between 1997 and 2016) and in all related fields that ensure its functioning. This results in a total of 1019 directives (European Commission, 2017), which enables us to explore changes in member state compliance with EU directives over time, including two dependent variables: the transposition deficit (the gap between adopted and transposed directives) and the infringement procedures. Implementation in our study thus refers to the transposition of EU directives into national law and their subsequent enforcement (Falkner et al., 2005, pp. 5-7;Treib, 2004, p. 61). This data provides access to comprehensive time-series cross-sectional data that contain a total of 520 year/country observations (20 observations for each country).
A limitation of focusing on a highly integrated field such as the Single Market is that partial exits obviously are far less common than in politicized and less integrated fields, such as, for example, justice and home affairs where opposition from certain Member States are more prominent (see for example Roos, 2018). However, if Eurosceptic agendas indeed promote protectionism (Meulenbelt, 2016), we are likely to see increased non-compliance also with Single Market directives.

V. Operationalization of Hypotheses and Control Variables
To operationalize the factors that are hypothesized to drive partial exits (H1-H3), we include four variables (A-D) in our analyses. First, Government capacity (A) is operationalized as the highly regarded World Bank measurement of 'Government Effectiveness'. This index gauges a government's capacity to produce and implement sound policies and deliver public goods. It combines into one variable measures on the quality of public service provision, the 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. However, as it is unrealistic to expect a linear effect of government capacity, we add a square term to control for curvilinearity. Hypothesis 1 gains support if we find that the association between the dependent variable and government capacity is either negatively linear or exponentially (negatively) curvilinear. Moreover, although this variable is our primary indicator of government capacity, we also employ OECD data to control for real GDP growth (per cent change from previous year), gross government debt (per cent of GDP), government deficit (net lending as per cent of GDP), and unemployment.
Hypothesis 2 is operationalized as Rae's index 1 of Party system fragmentation. It is supported if this variable has a positive effect on our dependent variable. Moreover, we employ two variables to operationalize Hypothesis 3. First, building on the Comparative Manifesto Project, Party system Euroscepticism is operationalized as a balance measure between negative 2 and positive 3 references to the EU. This particular variable provides information on the average degree of negative references (minus positive references) to the EU for all parties in the party systems, for each election between 1997 and 2016. Second, radical right support (D) is operationalized as the percentage of votes for all parties that are categorized as belonging to this party family in the Comparative Political Data Set. 4 Lastly, as research has demonstrated that the ideology of parties in governments may affect Member States' degree of compliance, with left-wing parties more prone to comply than right-wing parties are (Mastenbroek and Kaeding, 2006;Falkner et al., 2005, pp. 309-313;Toshkov, 2008;Treib, 2004), we include Schmidt's index 5 of government composition as a control variable. However, in the context of this paper, this variable is merely included as a control. Descriptive statistics are presented in the Online Table A1.
While our first and most basic test of H1 to H3 is to explore the main (stationarity) effects of variables A to D, we also control for 'structural breaks' (non-stationarity) in the time series by interacting the variables with time dummies (Plümper et al., 2005, pp. 345-346). Hypotheses 1 to 3 are supported if the marginal effect of variable A declines over time while the marginal effects of variables B to D increase. Lastly, Hypothesis 4 is operationalized as the interaction between year dummies and dummy variables for the four worlds of compliance. Furthermore, we interact the dummies for the worlds with variables A to D. H4 is supported if the results show varying effects of variables A to D across the worlds as well as increased divergences between them over time. While we follow the original classification of the 'worlds of compliance' typology (Falkner et al., 2007), we add Poland (see for example Leiber, 2007), Bulgaria, and Romania (see for example Falkner et al., 2008, p. 2) to the 'world of dead letters'. Moreover, we added Cyprus to the 'world of transposition neglect' as it has a similar system to Greece, with comparable implementation outcomes. The data for Croatia is insufficient as our analysis only includes the years until 2016, while the three Baltic countries and Malta also could not be clearly affiliated to any of the worlds. In the models (3-8) that control for the worlds of compliance, these countries constitute the reference category.

VI. Analysing Time-Series Cross-Sectional (TSCS) Data
Given that our data consist of 28 EU members, over a 20-year period, both serial correlation of errors and panel heteroscedasticity are likely to bias our estimates. A standard technique that allows one to cope with these problems is the introduction of a lagged dependent variable (for example Beck and Katz, 1995). However, our statistical models combine panel corrected standard errors (that are robust to panel heteroscedasticity) and time-fixed effects (to control for non-spherical errors in the time dimension) with a Prais-Winsten AR1 transformation that eliminates serial correlation. In other words, our models manage serial correlation by including lagged residuals instead of a lagged dependent variable. The reason for this choice of method is the lag structure of our data. In fact, when studying the development over time in countries with different institutions, Plümper et al. (2005, p. 244) argue that misspecification of the lag structure is highly likely. Given that these concerns most certainly apply to the 28 countries (and 20 years) in our data, our method of choice is the Prais-Winsten regression.

VII. Findings
The results of our panel regression analyses (dependent variable: transposition deficit) are presented in Table 1, whereas the same models are also regressed against infringement procedures between 2003 and 2016 in Online Table A4. In Models 1 to 2, the coefficients for all covariates estimate the overall impact for the entire period between 1997 and 2016 for all 28 EU countries. In these models, Hypotheses 1 to 3 are tested. Subsequently, Model 3 adds dummy variables for the different worlds of compliance, whereas Model 4 interacts them with year dummies (the full interaction terms are found in Table A2 in the Appendix). Lastly, Models 5 to 8 include interaction terms between variables A to D and the four worlds of compliance. In these models, Hypothesis 4 is tested. Furthermore, Table 1 is continued in Models 9 to 14 in Online Table A3 where we interact variables A to D with year dummies.
The results in Models 1 to 2 indicate that the coefficients for variables A to D almost entirely reject Hypotheses 1 to 3 (the predicted effects are illustrated in Online Figure  1A). First, a rather substantial increase is evident in the transposition deficit as government capacity improves, although the effect tapers off and declines at the highest levels (these effects are illustrated in Online Figure 1A). The data demonstrates that this decline at the highest levels is explained by the fact that the best government capacity is found in the three Nordic countries, which also have the best compliance performance. Nevertheless, the overall effect clearly shows that government capacity predicts non-compliance.       Tables A2 and A3. Moreover, as shown in Online Table A4, the same effect applies to infringement procedures. Therefore, Hypothesis 1 is rejected. Additionally, our analysis demonstrates that party system fragmentation has no effect on the transposition deficit or the infringement procedures. This means that Hypothesis 2 is also rejected. Lastly, the results provide mixed support for Hypothesis 3. On the one hand, our findings show that a higher degree of party system Euroscepticism (C) predicts a lower transposition deficit while it does not affect infringement procedures (Online Table A4). This contradicts Hypothesis 2. On the other hand, the increased electoral strength of radical right parties yields a slight, but very uncertain, increase in the transposition deficit. However, we find no effect on the infringement procedures. Therefore, our results thus far provide no, or very limited, support for Hypotheses 1-3.
However, given that the estimated effects of variables A to D in Models 1 to 2 are the same for each year in the time series, it is possible that they are trending. To control for such structural breaks (non-stationarity), the analysis in Online Appendix Table A3 allows for parameter heterogeneity between the years, and the results of this analysis are presented in Figure 1.
The results displayed in Figure 1 cast an even greater shadow of doubt over Hypotheses 1-3. The marginal effects not only confirm the main direction of the effects of variables A to D (found in Models 1-2) but also that these effects have gradually diminished over the years. Although our analysis shows that the effects of all variables differed profoundly in 2004 due to the Eastern enlargement, the overall trend towards convergence is unmistakable.
While Figure 1 indicates that all four variables (A-D) were related to significant variations in the transposition deficit at the beginning of the time series, it is clear that these effects have declined over time. The same tendency of convergence, rather than disintegration, is apparent when reviewing the impact of the economic performance of Member States. While the main (stationarity) effects in Model 2 in Table 1 indicate that both real GDP growth per capita and (somewhat surprisingly) government debt significantly decrease the transposition deficit, our control for structural breaks in the time series again show that these effects are trending (Models 9-10 in Online Table A3). These results are presented in Figure 2.
As illustrated in Figure 2, both government debt and real GDP growth per capita only had a statistically significant effect until 2004 (government debt) and 2007 (GDP growth). However, subsequent to this, the effects levelled out and disappeared. This suggests that the economic performance of EU Member States no longer affects variations in the transposition deficit in spite of the major economic crisis in 2008. In Online Figure A4, the same trend is noted in the infringement procedures (and government capacity ceases to have an effect over time). Our results therefore reject the proposition that disintegration, in the form of partial exits, increases in the wake of economic and political turbulence (Online Table A4 shows similar results for infringement procedures). Only Hypothesis 4 remains and the proposition that disintegration is enhanced by differences between cultures of compliance.
Online Figure A2 reveals important differences in the determinants of the transposition deficit between the worlds of compliance. While government capacity predicts a lower deficit only in the world of law observance (with insignificant effects in the other worlds), party system fragmentation only explains increased non-compliance in the worlds of 'domestic politics' and 'dead letters'. Moreover, while party system Euroscepticism only predicts compliance with EU law in the law-abiding Scandinavian countries, the increased strength of the radical right predicts increased non-compliance in this world as well as in the world of domestic politics. Overall, the coefficients for the dummy variables for the worlds (Model 3) and for the interaction terms (Models 5-8) not only indicate significant differences in the transposition deficit between the worlds of compliance, but also that variations in the predictors of compliance/non-compliance are stable over time. However, a review of Figure 3 indicates that our results point in the opposite direction. Figure 3 shows that patterns of compliance/non-compliance differed significantly between the worlds of compliance by the late 1990s but have converged to almost the same  (Table 1), between dummies for years and the worlds of compliance . The full interaction terms, for each year between 1997 and 2016, are found in the Online Table A2. The vertical axis on the left indicates the magnitude of the marginal effect. The vertical axis on the right is for the scatter plot, which depicts the distribution of the transposition deficit by year in 23 EU member states. [Colour figure can be viewed at wileyonlinelibrary.com] Figure A3, which shows that the variation between the worlds of compliance was significantly larger than the variation within the worlds until the Eastern enlargement in 2004. Nonetheless, the trend lines in Figure 3 clearly show that the increased convergence between the worlds began several years before a major event like the 2004 Eastern enlargement and, particularly, the 2008 economic crisis. Moreover, the trend lines in Online Figure A6 demonstrate the sharpest downfall already between 1997 and 2001. While this trend was temporarily interrupted in 2004, the underlying scatterplots demonstrate that the largest convergence between Member States occurred in the years subsequent to the enlargement and the crisis. However, Online Figure A7 demonstrates that the overall decline in the transposition deficit is especially visible among the EU 15, whereas Online Figure A8 demonstrates that the decrease in infringement procedures is mainly driven by the founding members of the EU (see Toshkov, 2008).
This means that compliance appears to be largely unaffected by crises and external events. Moreover, while the determinants of the transposition deficit and infringement procedures differ (see Online Table A4), Online Figure A5 shows that a similar trend towards convergence also applies here. Consequently, Hypothesis 4 is also rejected.
In summary, our analyses of the time series between 1997 and 2016 (the transposition deficit) and 2003 and 2016 (infringement procedures) show a strong convergence tendency between the compliance behaviour of EU Member States. The apparent decrease in the transposition deficit and infringement procedures displayed in the paper, contradicts the idea that disintegration in the form of non-compliance is an ongoing process in the EU. It rather indicates that EU legislation has become more harmonized.

Conclusions
Although growing economic and political turbulence in EU Member States are widely believed to have sparked a process of European disintegration, knowledge is surprisingly scant about the long-term effects of these developments on Member States' compliance with EU law. We contribute to filling this gap in the research by exploring Member States' non-compliance with Single Market directives between 1997 and 2016. In addition, we contribute to the literature by tracing patterns of compliance across Member States with different cultures of compliance. Summing up the results, our main conclusion is straightforward: Instead of finding a 'Eurosceptic combination of declining compliance and reduced means to enforce compliance' (Vollaard, 2018, p. 250), we find the exact opposite with regard to both the transposition deficit and infringement procedures.
We argue that our findings contribute at least two principal insights to the literature on European (dis)integration and member state compliance. First, we formulated three hypotheses to test the argument that Member States increasingly choose to partially exit the EU by not complying with EU law in the wake of growing economic and political turbulence (Vollaard, 2014, pp. 12-13). Building on previous research (Börzel et al., 2010;Hartlapp, 2009;Haverland and Rumeijn, 2007;Linos, 2007), we hypothesize that improved government capacity becomes increasingly important for counteracting Member States' non-compliance in the wake of economic turbulence (Hypothesis 1). Moreover, based on a vast body of literature, we postulate that party system fragmentation (Hypothesis 2) and the growing importance of Eurosceptic opinions and parties (Hypothesis 3) over time will increase the likelihood of non-compliance. However, our findings conclusively reject all three hypotheses. With regard to the main effects in all 28 Member States between 1997 and 2016, our results in fact indicate that greater government capacity increases the transposition deficit and the infringement procedures. Additionally, while party system fragmentation is primarily unrelated to compliance, Euroscepticism also predicts a slightly lower transposition deficit but has no effect on infringement procedures. The only result that is somewhat in line with our hypotheses is the finding that the increased electoral strength of radical right parties yields a slightly higher transposition deficit. Most importantly, when accounting for the time dimension in our data, we find that the effects of all four variables have vanished almost completely over time. Similarly, our analysis shows that the effects of government debt and GDP growth, which were significant until the mid-2000s, over time have ceased to affect variations in the transposition deficit and the infringement procedures in spite of the major economic crisis in 2008. While these findings indicate a process of harmonization rather than disintegration, they could also imply that Hypotheses 1 to 3 fail to accurately explain patterns of compliance across the economically and politically different Member States of the EU. This leads to our second contribution.
Our last hypothesis (H4) is based on the proposition that disintegration is enhanced by different cultures of compliance (Falkner et al., 2005(Falkner et al., , 2007. Initially, it appears to be supported by the data. Our results not only confirm that the basic differences between the worlds of compliance suggested by Falkner et al. (2005) indeed existed by the late 1990s (see Thomson, 2009;Toshkov, 2007), but they also reveal important differences in the determinants of the transposition deficit between the worlds of compliance. However, while the main effects in our models indicate that these variations are stable over time, our analysis of the time series point in the exact opposite direction.
Although Hypothesis 4 anticipates increased divisions between countries with different compliance cultures, we again find a strong trend towards improved compliance and harmonization, particularly after the economic crises in 2008.
Taken together, our findings not only suggest that factors such as government debt, GDP growth/capita, government deficit, government capacity, and, above all, domestic politics become gradually decoupled from the European integration process (see Mair, 2013, p. 115), but also that deep-rooted cultural differences have become increasingly irrelevant to explain government behaviour. Our results, in other words, sharply contradict the argument that the economic and political turbulence initiated reduced capability and willingness of some Member States to integrate (for example Vollaard, 2018).
Why, then, do we find increased compliance in times of crisis? We argue that our results may be explained in the light of recent research. Studies reveal a profound change in the enforcement culture of both the EU and its Member States. Several reasons for this phenomenon are discussed. First, new governance instruments, such as the EU Pilot and SOLVIT, were introduced to improve the 'out-of-court' problem-solving dialogue (Falkner, 2018). Second, the EU's direct enforcement power and authorities were expanded (Scholten, 2017). Third, the resources of the European Commissions to sue non-compliance decreased (limited capacity, less budget, shortage of workforce) (Falkner, 2018). Fourth, enforcement was outsourced to private litigants and national courts (Falkner, 2018; see also Hofmann, 2018). Lastly, the European Commission strengthened the compliance capacity within the Member States (Börzel and Buzogány, 2019, p. 331) and the nature of EU law has changed. EU legislation has become less demanding by amending existing legislation rather than creating new legislation, and the amended directives require less effort to implement (Börzel and Buzogány, 2019).
We contribute to this literature by demonstrating a cultural change in transposing EU directives over time. Macro-economic factors no longer influence the transposition performance of Member States. Moreover, important variables widely discussed in literature, which influenced the transposition up until 2008, such as government capacity, party system fragmentation, Euroscepticism and the strength of the radical right, ceased to influence the transposition deficit. This demonstrates that the transposition of EU law has become decoupled from the political process and is largely unaffected by external events. This, in turn, supports research which indicates that the EU policy process over time gradually becomes less politicized and more technocratic (for example Mair, 2013).
Overall, our results reveal a strong integration logic among countries with vastly different compliance cultures in the midst of the worst economic and political crisis facing the EU thus far. However, our study is limited to the Single Market. Therefore, future research in the direction suggested by this article should explore compliance over time in more contested policy areas (for example justice and home affairs). Therefore, our findings clearly demonstrate that the process towards increased harmonization appears to be stronger than ever in the centre field of the EUthe Single Market.