ACTA UNIVERSITATIS UPSALIENSIS Skrifter utgivna av Statsvetenskapliga föreningen i Uppsala

201

National Courts as Gatekeepers in European Integration

Examining the Choices National Courts Make in the Preliminary Ruling Procedure

Karin Leijon

Dissertation presented at Uppsala University to be publicly examined in Brusewitzsalen,Statsvetenskapliga institutionen, Gamla Torget 6, Uppsala, Friday, 8 June 2018 at 10:15 forthe degree of Doctor of Philosophy. The examination will be conducted in Swedish. Facultyexaminer: Professor Daniel Naurin (University of Oslo).

AbstractLeijon, K. 2018. National Courts as Gatekeepers in European Integration. Examiningthe Choices National Courts Make in the Preliminary Ruling Procedure. Skrifter utgivna avStatsvetenskapliga föreningen i Uppsala 201. 213 pp. Uppsala: Acta Universitatis Upsaliensis.ISBN 978-91-513-0332-1.

The national courts’ placement in the intersection between the EU and member state legalsystems makes them important gatekeepers in the process of European legal integration. In thescholarly debate, national courts are characterized as either supporters of legal integration or asdefenders of national sovereignty. The aim of this thesis is to improve our understanding of thenational courts’ role as gatekeepers in EU legal integration by shedding new light on the nationalcourts’ behavioral patterns and the reasoning of individual judges in the preliminary rulingprocedure. Empirically the thesis provides a detailed examination of the national courts’ twokey choices in the preliminary ruling procedure, both of which have important implications forthe scope and pace of integration: (1) whether national courts are allowing the Court of Justice ofthe European Union (CJEU) to decide politically sensitive cases in which national policies areat stake; and (2) whether national courts frame cases by expressing support for an integration-friendly interpretation of the EU law or whether they instead voice an opinion in defense ofchallenged national law. Theoretically, the thesis contributes to redefining the main theoreticalcontroversy on what role national courts play in EU legal integration by identifying previouslyun-theorized behavior patterns. The analysis demonstrates that national court behavior is notlimited to either supporting legal integration or defending national sovereignty. On the contrary,national courts frequently make choices that may alleviate parts of the inherent tension betweennational concerns and the EU legal obligations that member states must accept in order for theEU to function efficiently. Moreover, a case study of the Swedish judiciary shows that Swedishjudges are not reasoning as expected by the dominant theoretical outlook in the judicial politicsliterature, meaning that they are not primarily guided by the logic of consequentialism and self-regarding considerations when making decisions in the preliminary ruling procedure. Instead,the judges’ mode of reasoning centers on what constitutes the appropriate course of actiongiven their professional obligations and how their choices may impact the functioning of thepreliminary ruling procedure.

Keywords: European integration, National courts, Court of Justice of the European Union

Karin Leijon, Department of Government, Box 514, Uppsala University, SE-75120 Uppsala,Sweden.

© Karin Leijon 2018

ISSN 0346-7538ISBN 978-91-513-0332-1urn:nbn:se:uu:diva-347101 (http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-347101)

Till Ingegerd & Mats

Contents

Acknowledgements ........................................................................................ xi

1. Introduction: National courts as gatekeepers in European integration ..... 15 Aim and research questions ...................................................................... 19 Identifying the research gaps .................................................................... 19 Positioning the research questions ........................................................... 25

The first debate: Are national courts supporting EU legal integration or defending national sovereignty? ...................................................... 25 The second debate: What explains the variations in the national courts’ support for EU legal integration? ............................................ 26 The third debate: What drives the behavior of individual judges? ...... 27

Overview of the thesis .............................................................................. 29

2. A theoretical model for understanding the role of national courts in EU legal integration ............................................................................................ 32

The behavioral pattern of national courts: supporting EU legal integration or protecting national sovereignty? ........................................ 33 Explaining variations in national court behavior ...................................... 38

Explanations derived from the judicial empowerment approach ........ 39 Explanations derived from the sustained resistance approach ............. 40 Explanations derived from the Europeanization literature .................. 44

The micro foundations: What guides the behavior of individual judges in EU legal integration? ............................................................................ 47

Consequentialism in judicial politics ................................................... 48 Self-regarding preferences ................................................................... 50 Other-regarding preferences ................................................................ 53 Questioning consequentialism ............................................................. 54 The challenge from the logic of appropriateness ................................. 57

A theoretical model of judicial behavior in EU legal integration ............ 60

3. Methodological approaches and research design ...................................... 65 Assessing the national courts’ behavioral pattern: A quantitative approach ................................................................................................... 65

Identifying national court opinions ...................................................... 68 Identifying the political sensitivity of a request for a preliminary ruling .................................................................................................... 72 Empirical examples of the four different types of cases ...................... 76

Understanding the behavior of individual judges: A qualitative approach ................................................................................................... 79

Selecting respondents and carrying out the interviews ........................ 83

4. The behavioral patterns of national courts: Supporting EU legal integration or protecting national law? ......................................................... 86

What types of cases are national courts referring to the CJEU? .............. 86 Results ................................................................................................. 95

What types of opinions are national courts expressing in the requests for preliminary rulings? ............................................................................ 97

Results ............................................................................................... 106 Combining cases with opinions: four behavioral patterns ..................... 106 Summary of chapter ............................................................................... 110

5. Explaining what choices national courts make in EU legal integration ... 112 Theoretical explanations ........................................................................ 112 Hypotheses derived from judicial empowerment approach ................... 113

Inter-court competition ...................................................................... 113 Hypotheses derived from the sustained resistance approach ................. 115

Public opinion .................................................................................... 115 Majoritarian democracy ..................................................................... 117 The political sensitivity of the case.................................................... 118

Hypotheses derived from the Europeanization literature ....................... 119 Experience of the preliminary ruling procedure ................................ 119 Concrete judicial review .................................................................... 120

Overview of the hypotheses ................................................................... 122 Data and measurement ........................................................................... 122

Dependent variables........................................................................... 122 Independent variables ........................................................................ 123 Models and analysis ........................................................................... 125

Analysis I: The referral of cases with a high degree of political sensitivity ............................................................................................... 126 Analysis II: Explaining the variation in the types of opinions expressed ................................................................................................ 129 Summary of chapter ............................................................................... 132

6. The behavior of individual national judges in the preliminary ruling procedure .................................................................................................... 133

Identifying the driving forces of judicial behavior ................................. 134 The dominant mode of reasoning: Self-regarding preferences .......... 135 The challenge from reasoning based on other-regarding preferences ......................................................................................... 136 The challenge from reasoning based on formal codified rules .......... 136 The challenge from reasoning based on informal professional practices ............................................................................................. 137

Exploring judicial reasoning in the preliminary ruling procedure ......... 138 The first decision: Whether or not to refer cases to the CJEU ........... 139 The second decision: Whether or not to include opinions in the references to the CJEU ...................................................................... 148

Judicial reasoning beyond self-interest? ................................................. 155 Summary of chapter ............................................................................... 166

7. Conclusion: Findings and implications ................................................... 168 Main findings ......................................................................................... 168 Theoretical contributions ........................................................................ 172

National courts and national judges: The implications of incorporating the micro perspective .................................................. 172 The role of national courts: Gatekeepers who balance conflicting demands ............................................................................................. 175

Avenues for future research and concluding remarks ............................ 180

8. References ............................................................................................... 186

Appendices .................................................................................................. 198 Appendix A – Units of analysis ............................................................. 198 Appendix B – Regression tables and statistics ....................................... 202 Appendix C – Information to the respondents and the interview questionnaire .......................................................................................... 212

xi

Acknowledgements

I would like to thank all my colleagues at the Department of Government at Uppsala University, and in particular my excellent supervisors Christer Karlsson, Li Bennich-Björkman and Katrin Uba. I would also like to thank Margit Tavits, Matthew Gabel, Michael Bechtel and Simone Bechtel who made my stay as a visiting PhD student at the Department of Political Science at Washington University (St. Louis) a pleasure. Thanks also to Christoph Hönnige and Sarah Sinram for inviting me to the Institut für Politikwissenschaft at the Georg-August-Universität Göttingen and to Lisa Conant for taking time to meet with me in Denver.

Tack Det finns många personer som betytt mycket för mig under åren på forskar-utbildningen och som jag nu skulle vilja tacka. Till att börja med; mina tre alldeles utmärkta handledare. Katrin Uba har tålmodigt visat mig fördelarna med grafer, klarsynt identifierat otydligheter i mina kapitelutkast och bidra-git med ett välbehövligt utifrån-perspektiv på europeisk integration. Li Ben-nich-Björkman har hjälpt mig att lyfta blicken och se de övergripande forsk-ningsfrågorna. Du övertygade mig om värdet av att utvidga studien och även inkludera de enskilda domarnas perspektiv, vilket jag är väldigt glad för. Christer Karlsson, som en gång frågade mig om jag skulle vara intresserad av att skriva ”en masteruppsats till, fast en som är aningens längre”, har stöt-tat mig i vått och torrt. Du har alltid tagit mina avhandlingsrelaterade be-kymmer på största allvar och hjälpt mig att vrida och vända på svårigheterna till dess att jag kunnat reda ut dem själv. Du har också lärt mig att alla pro-blem går att lösa genom en FS-session (eller två)! Stort tack till er alla tre för all tid som ni har lagt ner, jag hade inte kunnat önska mig några bättre hand-ledare. Våra handledningsmöten har varit konstruktiva, inspirerande och en gnutta oförutsägbara – precis vad jag har behövt.

Flera personer har ställt upp och kommentaret utkast på avhandlingen un-der årens lopp. Jag vill därför rikta ett stort tack till Sverker Gustavsson, Sten Widmalm, Paula Blomqvist, Olof Larsson, Anna Wetter Ryde och Marcus Österman som alla gett mig värdefulla kommentarer i slutfasen av avhand-lingsarbetet. Vidare vill jag tacka Helena Wockelberg, Karl-Göran Algots-

xii

son, Per Ekman, Thomas Persson, Anna Michalski, Moa Mårtensson, Lud-vig Norman samt deltagarna vid Europaseminarierna vid juridiska institut-ionen för all konstruktiv feedback.

Alla kollegor på statsvetenskapliga institutionen i Uppsala har bidragit till att jag upplevet doktorandtiden som väldigt rolig och några av er vill jag särskilt lyfta fram. Först ut är 2012 års kohort: Rafael Ahlskog, Nadja Grees och Marcus Wangel. Ni tar forskning och undervisning på stort allvar och det har varit spännande att se hur våra respektive avhandlingsprojekt utveck-lats. Men vad jag verkligen uppskattar är den avslappnade och glada stäm-ningen i vår lilla grupp! Stort tack! Jag vill även passa på att tacka alla de kompetenta personer som jag har haft nöjet att undervisa tillsammans med på Metod C. Under de sista intensiva månaderna av avhandlingsskrivande har jag fått fantastiskt stöd av Camille Pellerin och Lars ”Lannister” Wik-man. Tack för att ni bjudit på välbehövliga pauser med cocos-bollar och memes.

En viktig del av forskarutbildningen är att träffa andra forskare och utbyta erfarenheter, antingen på konferenser eller genom att vistas en period i en annan forskningsmiljö. Bidrag som har möjliggjort sådana utbyten har kommit från: the Swedish Network for European Studies in Political Science (SNES), Thuns resestipendium, Borbos Erik Hanssons stiftelse, Gålöstiftel-sen, the U4 Network och Anna Maria Lundins stipendiefond. Tack!

En av förmånerna med att vara doktorand vid Campus Gamla Torget är att man får chans att lära känna doktorander från institutionen för freds- och konfliktstudier, juridiska fakulteten och the Institute for Russian and Eurasian Studies. Thank you Ausra, Axel, Ewelina, Ilmari, Johanna and Love for being awesome! Jag måste erkänna att jag inte riktigt kommer ihåg hur vi först träffades men jag minns tydligt en fantastisk resa till Vilnius, intensiva bokcirkeldiskussioner, Eurovisionevents och oförglömliga cock-tailpartyn. Från allra första dagen på doktorandprogrammet har Linda Mo-berg varit en fantastisk kollega. Jag har nog inte sagt det till dig Linda, men jag såg medvetet till att hamna i samma lag som dig på fotbollsgolfkick-off:en 2012 med det långsiktiga målet att vi skulle bli vänner. Du är en ut-märkt forskare, men för mig är du i först hand en vän som jag kan lita på i alla väder.

Utanför akademins värld finns många olika grupper av vänner som stöttat och roat mig under tiden som jag har varit doktorand. Kanongänget i Digital-rundan: Emelie Martinell, Johanna Alaeus, Sofia Jägerlind-Puuri och Sofie Eriksson ska ha ett stort tack för att ni alltid är redo att bjuda in till samtal om vardagliga såväl som livsomvälvande händelser. Jag har också haft turen att träffa många roliga, smarta och omtänksamma personer under min tid som student i Uppsala. Fredrik Sävje, Klara Nordström, Kristin Ivarsson, Sebastian Axbard, Sofia Ståhl och Zohreh Khoban: jag har så många roliga minnen tillsammans med er och jag har alltid vetat att ni kommer att lyckas med allt det som ni tar er för. Jag uppskattar det stöd som jag har fått från er

xiii

genom åren, men framförallt uppskattar jag er förmåga att dansa hela natten, oavsett om det är till Håkan Hellström eller Snoop Dog och oavsett om ni har en slips knuten runt huvudet eller ej! Jag vill också passa på att tacka en grupp personer som jag nu inser tillhör några av mina äldsta vänner. Tack Richard Ekman, Christian Lovering, Christian Wallgren, Martin Aalto, Ma-tilda Schuman och alla andra medlemmar av Stockholmsgruppen för alla de gånger som ni fått mig att kikna av skratt och för alla all-in nights på Deba-ser Slussen (även om jag misstänker att några av dessa kvällar i början av min doktorandtid är orsaken till att jag varit tvungen att arbeta helg de sen-aste månaderna).

Under de år som jag har skrivit på den här avhandlingen har jag insett hur viktigt det är att ha en bästa vän som man kan ringa på vägen hem från job-bet och fortsätta samtalet där man avslutade det dagen innan. I mitt fall är det Susanne Gehlert som svarar när jag ringer och för det är jag innerligt tack-sam! You are the Filip Hammar to my (inner) Fredrik Wikingsson.

Jag vill ge ett extra stort tack till min kära Gustav för att du tar hand om mig när det behövs och låter mig ta hand om dig när det behövs. Jag hoppas att du vet att jag älskar dig.

Som barn var mitt viktigaste mantra ”kan själv” men jag vet att få av mina planer hade kunnat genomföras utan stöd av mina föräldrar, Ingegerd och Mats. Ni har lärt mig att läsa, tagit er tid att gå igenom travar med hem-läxor och hjälpt mig att flytta otaliga gånger. Även om ni ibland varit skep-tiska till mina idéer (som att flytta till Uppsala för att studera), så har ni alltid ställt upp för mig då det visat sig att jag, trots allt, inte klarat mig helt själv. Ni är bäst. Den här boken är till er. Karin Leijon Uppsala, 23 april 2018

15

1. Introduction: National courts as gatekeepers in European integration

The scope of the economic and political cooperation between countries in the European Union (EU) has fundamentally changed the European political landscape. What started out as a small community that supervised the pro-duction of coal and steel in the 1950s has, in less than 50 years, evolved into a full-fledged political system that upholds the free movement of goods, capital, services and citizens between 28 member states. There is no doubt that the expansion of the EU is by far the most important political develop-ment in modern-day Europe. The width and depth of the integration are un-precedented in any other international organization, and the decisions made by EU institutions span from the enforcement of competition law to the har-monization of social policy between the member states, affecting the daily lives of millions of people.

This development raises overarching questions about what drives this co-operation between the member states forward and why European integration is occurring. What first and foremost has intrigued scholars of European affairs is whether EU institutions and subnational actors have been able to transcend the member states’ sovereignty and promote European integration (Garrett and Tsebelis 1996; Haas 1958; Kohler-Koch and Eising 1999; Moravcsik 1993; Pollack 1997; Stone Sweet and Sandholtz 1997). In partic-ular, a considerable amount of academic attention has focused on the process of European legal integration and what influences the Court of Justice of the European Union (CJEU), together with the national courts, have had on the overall development of the EU (Alter 1998a; Alter 2001; Burley and Mattli 1993; Weiler 1994).

In the 1960s, the CJEU laid the foundation of this EU legal integration in a number of path-breaking rulings. By formulating these legal doctrines, the CJEU have granted private litigants access to use the EU legal system to challenge national policies (Alter 2001: 16). The CJEU have ensured that EU law effectively functions as domestic law, indicating that the EU as a whole bears closer resemblance to a federal political system than to an inter-national regime (Weiler 1991: 2415). Normally, international legal acts must be incorporated into domestic law before they can be invoked in national

16

courts. In contrast, the CJEU’s legal doctrines of direct effect1 and suprema-cy2 have given the citizens new rights under EU law and have established the EU law as supreme to national law. These doctrines have meant that member state courts must enforce EU rights, and if these rights conflict with any piece of national legislation, national courts are required to disregard it and apply the EU law instead (Sweet and Brunell 1998b: 67). Importantly, the CJEU’s decisions have invited national courts to help to enforce EU law in the member states, in turn creating political costs for member state govern-ments that did not obey EU law (Alter 2001: 17). By extension, the CJEU’s legal doctrines have limited the possibility that member states could escape their obligations under the EU Treaties through non-compliance (Weiler 1991:2412).

It is undisputed that the CJEU has been a key actor in promoting the ex-traordinary legal integration that has served as the basis for the entire politi-cal and economic EU project. However, the CJEU would not have been suc-cessful in this endeavor were it not for the national courts. Without a steady influx of cases, the CJEU’s ability to influence the course of European legal integration is heavily circumscribed (Alter 2001: 33). Since the CJEU cannot initiate legal proceedings by itself, it must be activated by a national court.3 In fact, the CJEU is dependent on the national courts’ willingness to provide it with legal cases through the preliminary ruling procedure (Alter 2001; Burley and Mattli 1993; Weiler 1994). If national courts do not refer cases to the CJEU, it will be starved of references, and only limited EU legal harmo-nization can occur (Golub 1996: 376-377). In contrast, if national courts are willing to provide the CJEU with references from different policy areas, it will be able to effectively incorporate EU law into the member states’ legal orders (Wind 2010: 1052).

The national courts are therefore the crucial link between the CJEU and the member states’ 28 different domestic legal traditions. It turns the national courts in each and every member state into important gatekeepers in the process of legal integration and decentralized enforcers of EU law (Mayoral 2017; Tridimas and Tridimas 2004). The CJEU’s ability to foster EU legal integration depends on whether national courts are willing to open the gate to the domestic legal system. This fact provides an important starting point for the present thesis: how national courts have undertaken the task of gate-keeping in the preliminary ruling procedure has highly significant implica-tions for EU legal integration. Not only does the national courts’ behavior have a direct impact on the outcome of the case at hand, but it also has the potential to influence national policies, as well as the development of EU

1 Established in a ruling by the CJEU in 1963: Case C-26/62, Van Gend en Loos v. Neder-landse Administratie der Belastingen 2 Established in a ruling by the CJEU in 1964: Case C-6/64, Costa v. ENEL 3 Or by the member states or the EU institutions

17

law (Granger 2006: 37; Rytter and Wind 2011: 490). For instance, it is well known that the CJEU uses the preliminary references from national courts to foster integration and to force reluctant member state governments to change their policies (Alter 1998a; Golub 1996: 376). This means that national court references to the CJEU have the potential to fundamentally alter domestic public policy and to shape EU legal integration.

There is close to a consensus in the literature that national courts are im-portant actors in the process of legal integration (Alter 1998b; Alter 2001; Conant 2013; Davies 2012; Mattli and Slaughter 1998; Weiler 1994; Wind et al. 2009; Witte et al. 2016). However, there is strong scholarly disagreement regarding how national courts are expected to behave and what role they play in the EU legal system. Previous research has framed the theoretical debate over national courts in terms of how their behavior influences Euro-pean integration. National courts are understood either as supporters of EU legal integration or as defenders of national policies and member state sover-eignty (Conant 2013; Pollack 2013). The proponents of the former view argue that national courts will use their position as gatekeepers to further the legal integration process because it increases their powers in relation to polit-ical actors. For instance, the EU legal system grants national courts the right to exercise de facto judicial review over legislation. This right is the judiciar-ies’ most powerful tool for monitoring political actors. National courts are thus expected to support EU legal integration since they strive for this type of judicial empowerment (Alter 1998b; Burley and Mattli 1993; Weiler 1994). Other scholars have rejected this claim by arguing that the national courts’ support for integration is not as widespread as one might believe. Instead, they have proposed that national courts mount sustained resistance (Pollack 2013: 1273) to further integration to protect domestic legal coher-ence, public policy or the member states’ sovereignty from EU-level intru-sions (Dehousse 1998: 173; Golub 1996; Wind et al. 2009).

The present thesis engages in this discussion about how national courts and judges respond to being placed at the center of the EU legal integration process (Weiler 1994). It does so by exploring what the national courts’ en-gagement as gatekeepers in the preliminary ruling procedure actually entails. The argument of the thesis is that, to understand the role of national courts in EU legal integration and to resolve the scholarly debate, we must systemati-cally examine what choices national courts make in the preliminary ruling procedure. Being gatekeepers does not only mean that national courts control the number of cases that reach the CJEU, which has been the focus of much previous research (for an overview, see, e.g., Conant 2007: 54). What is more important is that national courts control which types of cases that the CJEU is allowed to decide upon, as well as how these legal cases are framed by the national courts’ opinions regarding the outcomes of the cases. These aspects of national court behavior have important implications for the scope and pace of European integration, and they will be referred to as the national

18

courts’ two key choices in the preliminary ruling procedure: (1) whether or to what extent national courts are willing to allow the CJEU to decide political-ly sensitive cases in which national policies are at stake; and (2) whether the national courts frame the cases by expressing support for an integration-friendly interpretation of the EU law or whether they instead voice an opin-ion in defense of challenged national law.

The theoretical debate in previous research over how national courts act in the preliminary ruling procedure calls our attention to three different as-pects of court behavior that are fundamental for understanding the role of national courts in EU legal integration. The first and second aspects concern how the courts behave on an aggregated level and why we might encounter variations in their behavioral patterns. These aspects are essential for as-sessing whether or the extent of which national courts act as supporters of EU legal integration or as defenders of national sovereignty. The third aspect concerns what drives the actions of individual judges. Underlying the over-arching scholarly debate on judicial decision-making is the assumption that national courts are populated by judges who make decisions based on ex-pected consequences (Dyevre 2010; Rehder 2014). This perspective on judi-cial behavior stems from the influential US judicial politics literature (for an overview, see, e.g., Baum 1997; Epstein and Lindquist 2017; Whittington and Kelemen 2008). However, not everyone agrees with this proposition. Other scholars have claimed that human action is rule-bound and shaped by the logic of appropriateness (Gillman 2001; March and Olsen 1989). Illumi-nating these three different aspects of the national courts’ role as gatekeepers is fundamental for understanding the driving forces of EU legal integration and, by extension, the overall development of the EU.

What this thesis proposes is thus a new approach to the main theoretical question about what role national courts play in EU legal integration. By focusing on the national courts’ two key choices in the preliminary ruling procedure and on both the aggregated behavior of national courts and the decision-making of individual judges, this thesis sheds new light on the na-tional courts’ behavior as gatekeepers in EU legal system. As the upcoming chapters show in detail, subjecting the assumptions formulated by previous research to empirical scrutiny leads to a redefinition of how we understand the role of national courts and judges in the process of EU legal integration.

19

Aim and research questions The aim of the thesis is to shed light the national courts’ and judges’ behav-ior as gatekeepers in the preliminary ruling procedure in order to improve our understanding of their role in EU legal integration. To achieve this aim, the thesis sets out to address three interrelated research questions.

The first question stems from the theoretical controversy regarding how national courts behave in EU legal integration and whether their decisions lead to a behavioral pattern that supports EU legal integration or to one that defends national sovereignty:

(1) What types of choices are national courts making in the preliminary ruling procedure?

The second question centers on why some national courts support EU legal integration while others defend national sovereignty, that is, why there are variations in the national courts’ behavioral patterns:

(2) What factors can explain the variations in the types of choices that national courts make in the preliminary ruling procedure?

The third question springs from the scholarly debate on what constitutes the micro-foundation of the national courts’ aggregated behavioral patterns. In particular, it relates to the judicial politics literature on whether or to what extent the judges’ decision-making is driven by the logic of consequential-ism or the logic of appropriateness:

(3) What types of reasons are individual national judges guided by when making choices in the preliminary ruling procedure?

Before introducing the particular research debates to which these questions relate, the upcoming section outlines the gap in the previous research that this thesis seeks to address.

Identifying the research gaps At the heart of EU legal integration, one finds the preliminary ruling proce-dure, which is regulated in Article 267 of the Treaty on the Functioning of the European Union (TFEU). It is within this procedure that the interaction between national courts and the CJEU occurs. It is also the procedure that makes national courts gatekeepers in EU legal integration. The national

20

courts are allowed, and sometimes are required, to send cases to the CJEU, asking it to clarify the interpretation of EU law (Broberg and Fenger 2013). The CJEU’s answers to the national courts’ questions are called preliminary rulings. These rulings can often be understood as a form of supranational judicial review since the CJEU examines whether national legislation is compatible with the supreme EU law (Alter 2009; Rytter and Wind 2011). The CJEU’s interpretation of EU law presented in the form of a preliminary ruling is binding on national courts. At the final stage of the procedure, na-tional courts use the answer given by the CJEU to decide the case at hand.

The CJEU is known for pursuing an integration-friendly agenda, and it is a well-established fact that it has used the preliminary references coming from national courts to foster EU legal harmonization and to force member states to change their policies (Alter 1998a; Golub 1996: 376; Rasmussen 1986; Stein 1981). Although the CJEU only has the formal authority to spec-ify the meaning of EU law and not the national legislation’s compatibility with the EU law, it has often gone further in its answers to the national courts’ questions. Former CJEU judges such as Federico Mancini have ex-plicitly stated that the CJEU encourages national courts to use the prelimi-nary references as a tool for reviewing national legislation and to assess whether or not it is compatible with the supreme EU law (Alter 2001: 10; Mancini 1989: 606). In the words of Mancini:

having paid this lip service to the language of the Treaty and having clarified the meaning of the relevant community measures, the court usually went on to indicate to what extent a certain type of national legislation can be regard-ed as compatible with that measure. The national judge is thus led hand in hand as far as the door; crossing the threshold is his job, but now a job no harder than child’s play (Mancini 1989: 606).

Legal scholars have argued that the CJEU’s pro-integration stance in inter-preting EU law has been necessary for the CJEU to be able to efficiently enforce the rules laid down in the treaties and avoid a complete breakdown of the EU political system (for an overview, see, eg., Burley and Mattli 1993: 46; Cappelletti et al. 1986; Pescatore 1974). However, others have criticized the CJEU for going too far in this quest and have claimed that the CJEU favors market integration over all other policy goals (Grimm 2015; Höpner and Schäfer 2012). Scharpf notes that “Subjective rights derived from (the interpretation of) European law may, in principle, override all countervailing national objectives, regardless of their salience as manifesta-tions of democratic self-determination” (Scharpf 2009: 193). The problem, according to these scholars, is that the CJEU’s behavior threatens the legiti-macy of the EU project since it undermines democratically undertaken deci-sions of the member states (Scharpf 2009: 189-190). Indeed, for some indi-vidual citizens and companies, the preliminary ruling procedure has been an

21

important means to challenge member state policies and to transform nation-al legislation, including even constitutional provisions (Carrubba and Gabel 2015: 62; Granger 2006: 37; Rytter and Wind 2011: 490).

There is no doubt that the preliminary ruling procedure has been contro-versial (Alter 2001: 2). The manner in which it has been used by the CJEU is expected to have upset the member state governments that prefer to remain in control over national policies (Alter 2001: 37; Blauberger 2014: 460). Moreover, member states that are unhappy with the CJEU’s rulings find it very difficult to limit the court’s powers since it requires a treaty revision that can only be adopted by consensus (Hix and Høyland 2011: 115; Scharpf 2009: 182-183; Tsebelis 1995). In summary, the potentially far-reaching implication that comes from sending references to the CJEU is what makes national courts gatekeepers in EU legal integration. National courts have the power to either activate the CJEU and allow it to influence national policies or to keep domestic policy matters beyond the CJEU’s reach (Tridimas and Tridimas 2004: 132). These circumstances raise the question of how national courts balance the demands from, on the one hand, the CJEU and its wish to further EU integration and, on the other hand, the member states’ govern-ments and their desire to maintain control over national policy.

An illustrative example of the importance of the national courts’ positions as gatekeepers and of the type of domestic policy issues that can be placed under judicial review in the preliminary ruling procedure can be found in the legal controversy surrounding the alcohol monopoly in Sweden. On the very same day that Sweden became a member of the EU, a grocery store owner started selling wine, claiming that the Swedish statutory monopoly on the retail of alcoholic beverages was incompatible with the EU Treaties. This incident was the beginning of a lengthy legal procedure in which the faith of an important public policy, the alcohol monopoly in Sweden and in other EU member states, was to be decided by courts. The storekeeper was prosecuted for selling alcoholic beverages without a license, and the domestic court hearing the case decided to send it4 to the CJEU and ask whether the national policy was compatible with EU law (Slot 1998: 1186).

The national court’s decision to involve the CJEU and EU law came as an unpleasant surprise to the Swedish government. The CJEU has been known for using its legal rulings to increase the supranational features of the EU and to simultaneously challenge national public policies that it claims are incon-sistent with EU law (Alter 1998b; Scharpf 2009). Indeed, many believe that the CJEU has pushed EU legal integration further than member state gov-ernments could have foreseen (Alter 2001; Sweet and Brunell 1998b: 68). In Sweden, as well as in other member states, there was a strong political con-sensus that a monopoly was the best way to minimize the alcohol-related problems in society. However, after Sweden’s accession to the EU, it was no 4 Criminal proceedings against Harry Franzén. C-189/95. ECLI:EU:C:1997:504

22

longer solely a domestic political issue (Ugland 2000). Instead, it turned into a question about the policy’s compatibility with EU law.

The actor responsible for this turn of events was a national court. It had the power to decide that the faith of a sensitive domestic policy should be evaluated by a supranational court with a reputation for overturning national policies for the benefit of further integration (Alter 1998a; Rasmussen 1986; Wasserfallen 2010). Moreover, the national court was also responsible for framing the case (Alter 2001: 61; Nyikos 2006) as being about the monopo-ly’s compatibility with EU law, thereby opening up the possibility for the CJEU to challenge the validity of the national policy.

These types of choices that national courts are making every time they are confronted with an EU legal case have implications for the CJEU’s ability to foster EU legal integration (Alter 2001; Nyikos 2006; Wind 2010). For in-stance, the CJEU is dependent on the influx of cases concerning the compat-ibility of national legislation with EU law. Being able to decide these types of cases is essential for legal harmonization between the member states and for EU integration. Similarly, the CJEU’s ability to expand the scope of EU legal integration could be impeded if national courts express opinions stating that only a narrow application of EU law will be accepted in the member state. National courts have been known to challenge the CJEU’s interpreta-tions when they do not like them. The CJEU is therefore well aware of the need to cater to national courts’ objections since it is dependent on their will-ingness to enforce EU law (Alter 2001: 61-62).

Although previous research has found that the national courts’ choices in the preliminary ruling procedure are important, empirical studies have so far been limited. A large amount of the existing research is instead concerned with the question of how many cases national courts send to the CJEU and what can explain the variations in these numbers between member states (for an overview, see, e.g., Conant 2007: 54). These works rest on the assump-tion that a greater number of requests for preliminary rulings provide the CJEU with more opportunities to influence legal development in the member states (Alter 2001; Burley and Mattli 1993; Sweet and Brunell 1998b; Weiler 1991).

However, while referral rates have been important for assessing the na-tional courts’ initial, and varying, acceptance of EU law, these numbers do not suffice if we want to understand the role of national courts as gatekeep-ers in a deeper and broader EU. The focus of previous research on aggregat-ed referral rates has been criticized because these numbers tell us very little about the attitudes among national courts toward further EU integration (Alter 2000: 501). Moreover, the number of references might not be that important for the pace of EU legal integration (Alter 2001; Conant 2002; Conant 2007; Conant 2013; Wind 2010). For instance, a large number of referred cases do not signify that the CJEU is given extensive possibilities to shape EU legal development if the bulk of these cases is confined to only a

23

limited area of EU rules and regulations (Alter 2000: 501) or if the national courts frame the requests in a manner that limits the CJEU’s ability to ex-pand the reach of EU law (Nyikos 2006; Wind et al. 2009: 64).

If we want to resolve the theoretical controversy regarding the role that national courts play as gatekeepers in EU legal integration, it is necessary to move beyond the number of referred cases. What instead merit attention are the national courts’ two key choices in the preliminary ruling procedure. The courts’ first choice regards deciding the types of domestic legal cases in which the CJEU should be allowed to be involved. Are national courts main-ly referring cases of minor political importance, such as technical cases, to the CJEU (Alter 2000: 501; Alter 2009: 99; Bebr 1983: 456-457)? Or are they also regularly referring politically sensitive cases regarding conflicts between national law and EU law? The choice of the types of cases to refer to the CJEU is a highly important decision in relation to European integra-tion. Above, we observed an example of how a national court allowed the CJEU to have a say in a politically sensitive matter, namely the future of the alcohol monopoly in Sweden. By referring these types of politically sensitive issues to the CJEU, national courts provide the supranational court with am-ple opportunity to expand the scope of EU law and further legal integration. Conversely, if national courts instead withhold politically sensitive cases, the CJEU is prevented from enforcing EU legal norms.

The national courts’ second key decision concerns the framing of the le-gal case, meaning the types of opinions, if any, that courts are expressing in requests for a preliminary ruling. The most interesting aspect is whether national courts express support for an integration-friendly interpretation of the EU law or whether they instead voice opinions in defense of national policies (Leijon and Karlsson 2013). When drafting a request to the CJEU, national courts can choose to include a written statement, an opinion, ex-pressing how they believe that EU law should be interpreted and the legal case at hand resolved.5 Expressing opinions has been theorized as a way for national courts to influence the CJEU’s final rulings (Nyikos 2006). National courts can, for example, argue that a national policy, while restricting the free movement of goods, should be deemed compatible with EU law since it aims to protect public health (referred to as the norm of proportionality; Davies 2012). Alternatively, national courts might express the opposite opin-ion, i.e., that they consider a national policy to be incompatible with EU law.

By including opinions, national courts have an opportunity to express support for either their own member state’s law or the EU law. Although the effects of opinions on the CJEU’s final ruling have not yet been systemati-cally examined, anecdotal evidence indicates that the CJEU sometimes has

5 The legal basis of this practice can be found in: European Court of Justice (2005) Information Note on References from National Courts for a Preliminary Ruling Official Journal of the European Union, 48, 1-5.

24

considered the views of national courts in its judgments (Alter 2001: 61-62; Nyikos 2006; Rosas 2007: 126). If successful in influencing the CJEU, the national courts’ behaviors might contribute to setting precedents that limit or expand the reach of EU law, not only in the case at hand but also in adjacent cases and policy areas.

Apart from focusing on the number of references sent to the CJEU, the other main feature of the existing literature on EU legal integration is the emphasis on the national courts’ behavioral pattern on an aggregated level. This feature is about describing and explaining the overall trends in behavior across member states. However, this research is based on assumptions about the incentives that individual national judges are faced with. In particular, judges are expected to have the incentive to either support or resist EU legal integration (Alter 1998b; Golub 1996; Mattli and Slaughter 1998). Most analyses of EU judicial politics rely on rational choice theory and its as-sumption that actors make decisions based on expected consequences (Dunoff and Pollack 2017; Dyevre 2010; Rehder 2014). For example, as the name suggests, the judicial empowerment approach builds on the notion that national courts are driven by their self-interest to become judicially empow-ered. According to this theoretical approach, the judges’ calculations of how to increase their own powers vis-à-vis domestic political actors are essential for understanding their behavioral patterns in EU legal integration (Alter 2001; Burley and Mattli 1993; Mattli and Slaughter 1998: 190; Weiler 1994). However, few previous studies have empirically explored what guides the behavior of individual judges in Europe (Hönnige 2011: 350; Rehder 2014).

How national judges make decisions has important implications for our understanding of the forces driving EU legal integration. Those who sub-scribe to the notion of methodological individualism believe that all social phenomena can in principle be attributed to individual actions (Elster 1982). Even if one does not want to go that far, it is evident that knowledge about how national judges make decisions provides an important piece to the puz-zle of understanding national court behavior in EU legal integration. Failure to consider the reasons that judges have for making different choices means that we run the risk of misinterpreting how they actually have accepted the responsibility of being gatekeepers in EU legal integration.

In summary, two main and interconnected gaps can be identified in the previous research on national court behavior in EU legal integration. Alt-hough previous research has identified the types of case being referred (Alter 2001: 36; Wind 2010: 1052) and the types of opinions being expressed (Nyikos 2006) as important for EU legal integration, there is a lack of com-parative empirical research on the two key choices. Furthermore, the theoret-ical debate over the role of national courts in EU legal integration rests on a consequentialist approach to human behavior. However, empirical studies on what drives the behavior of national judges are lacking. In particular, we

25

want to know what motivates judges to make different types of decisions in the preliminary ruling procedure. Taken together, we have only a limited understanding of how national courts and judges behave in the preliminary ruling procedure. How national courts act has important implications for the continued process of EU legal integration. Exploring the choices that nation-al courts actually make at both the macro and micro levels thereby provides new and unique insight into how they have taken on the task of being gate-keepers in European legal integration.

Positioning the research questions The first debate: Are national courts supporting EU legal integration or defending national sovereignty? As discussed on the very first page of this chapter, the main interest of pre-vious research has been to understand how national courts’ behavior influ-ences EU legal integration. The initial attempts to chisel out what role na-tional courts play in the EU legal system resulted in a lively theoretical de-bate with clear connections to the broader discussion about what constitute the driving forces of European integration (Alter 2001; Burley and Mattli 1993; Garrett 1995; Golub 1996; Mattli and Slaughter 1995; Wind et al. 2009). At the core of this scholarly debate is the question of whether national courts are best characterized as either supporters of EU legal integration or as defenders of national sovereignty (Conant 2013; Pollack 2013).

The former view is frequently referred to as the judicial empowerment approach to understanding national court behavior (Conant 2013; Pollack 2013: 1271). This approach builds on the influential neofunctionalist theory of European integration and identifies the driving forces of integration to be the self-interest of both subnational actors, such as national courts, and of EU institutions, such as the CJEU. The judicial empowerment approach the-orizes that national courts will use their gatekeeping position to support EU legal integration since it expands their judicial powers and increases their influence (Burley and Mattli 1993; Mattli 1998; Weiler 1994).

Weiler has argued that the EU legal system, with its “principles of su-premacy and direct effect binding on national governments and parliaments, meant an overall strengthening of the judicial branch vis-à-vis the other branches of government” (Weiler 1991: 2426). The strengthening of national courts mainly stems from the preliminary ruling procedure granting them the power to de facto exercise judicial review over legislation (Alter 2001). This power would for most national judges “be heady stuff” (Weiler 1991: 2426). The upshot of the judicial empowerment approach is that an important factor driving EU legal integration is the national courts’ interest in becoming em-

26

powered. The answer to the theoretical question of how we are to understand the role of national courts is that they are promoting EU legal integration.

The opposing view has been referred to as the sustained resistance ap-proach. It rejects the claim made by the judicial empowerment approach that national courts support EU legal integration (Pollack 2013: 1273). Instead, national courts are expected to exhibit a behavior that effectively limits the CJEU’s ability to foster integration (Golub 1996; Wind et al. 2009). Golub theorizes that it is domestic political factors that create disincentives for na-tional courts to promote EU legal integration (Golub 1996: 378-379). For instance, one disincentive for national courts to support EU legal integration is that doing so is likely to undermine the domestic judiciaries’ control over policy outcomes (Golub 1996: 381). Another disincentive is that EU legal integration could undermine legal certainty and the coherence of the domes-tic legal system (Alter 2001: 48; Dehousse 1998: 173; Maher 1998). As Dehousse put it, “European law is often a source of disruption. It injects into the national legal system rules which are alien to its traditions and which may affect its deeper structure, thereby threatening its coherence” (Dehousse 1998: 173).

In yet another account, member state governments are believed to influ-ence the national courts’ behavior by advising them to not jeopardize core national policies (Wind 2010; Wind et al. 2009). This practice can either take the form of an outspoken instruction or can be part of a broader implicit understanding within the domestic political system about how EU law should be approached and contained (Nergelius. 2001: 88; Wind et al. 2009: 73). For the aforementioned reasons, the sustained resistance approach ex-pects that national courts keep the gate to the domestic legal sphere mostly shut, effectively shielding domestic policies from EU legal intrusions.

This thesis seeks to resolve the debate between judicial empowerment and sustained resistance and to provide an answer to the question of whether national court behavior support EU legal integration or defend national sov-ereignty. It leads to the first research question: What type of choices are na-tional courts making in the preliminary ruling procedure?

The second debate: What explains the variations in the national courts’ support for EU legal integration? In the wake of the attempts by the judicial empowerment and the sustained resistance approaches to understand the national courts’ influence on EU legal integration, a growing body of research has developed hypotheses about what might account for variations in the courts’ behavior in the pre-liminary ruling procedure. Here, the main question, which is closely linked to the first debate, is why some national courts are supportive of further EU

27

legal integration, while others instead defend national policies (Conant 2007: 55; Conant 2013).

There is plenty of disagreement in the scholarly literature about the fac-tors that can account for the national courts’ varying dispositions toward EU legal integration (Carrubba and Murrah 2005; Mattli and Slaughter 1998; Sweet and Brunell 1998b; Wind et al. 2009; Vink et al. 2009). A hypothesis tied to the judicial empowerment approach proposes that lower national courts can be expected to gain more from EU legal integration than higher courts. We should therefore expect lower courts to support further integra-tion, while the courts of final instance instead protect national policies (Alter 1996; Alter 2001). Another hypothesis, based on the sustained resistance approach, is that domestic political factors influence the courts’ behavior. The proposal is that a low level of public support for EU legal integration in the member states causes national courts to be less likely to support further integration (Carrubba and Murrah 2005).

Empirical studies of national court acceptance or contestation of EU legal integration have mostly focused on differences in referral rates between member states (Sweet and Brunell 1998b), policy areas (Golub 1996) and court levels (Alter 2001). However, previous studies aimed at explaining the variation in the choices national courts make in the preliminary ruling proce-dure are few, indicating that we have limited knowledge about the factors that explain the variations in the national courts’ key choices. So far, there have been only two case studies of the types of cases that the UK (Golub 1996) and Danish (Wind 2010) courts refer. There has also been one previ-ous attempt to explain whether or not national courts include opinions in their requests for preliminary rulings (Nyikos 2006). However, there are no previous studies investigating why national courts express support for either the EU law or the national law in requests for preliminary rulings.

We thus lack information about the factors that can explain why national courts make different choices in the preliminary ruling procedure. Knowledge about these choices is important for understanding the role of national courts in the process of EU legal integration. To fill this gap in pre-vious research, the present thesis sets out to answer the second research question: What factors can explain variations in the type of choices that na-tional courts make in the preliminary ruling procedure?

The third debate: What drives the behavior of individual judges? While the first and the second research debates concern how we should un-derstand the national courts’ aggregated behavior in EU legal integration, the third debate centers on the actions of individual judges. Previous research has mainly portrayed national judges as actors driven either by an interest to increase their own powers or by an interest to defend national policies and sovereignty (Burley and Mattli 1993; Carrubba and Murrah 2005; Conant

28

2013; Golub 1996; Nyikos 2006). These assumptions are very much influ-enced by the rational choice perspective on human behavior, which domi-nates the US judicial politics literature. Indeed, most political science re-search on courts and judges assumes that judicial decision-making is based on the judges’ calculations of how to maximize their preferences (Baum 1997; Dunoff and Pollack 2017; Gillman 2001; Spiller and Gely 2008). In other words, the judges’ way of reasoning is believed to be driven by the logic of consequentialism (March and Olsen 1989). In terms of substantive reasons, judges are expected to make decisions based on, for example, their policy preference and, how to achieve different career goals (Baum 1997: 17-18).

However, modeling judges as rational actors who make decisions based on desired outcomes has been criticized (Gillman 2001; Rehder 2014; Tamanaha 2009). Scholars who employ a perspective based on the logic of appropriateness instead have suggested that human action is rule-bound (March and Olsen 1989), shaped by what actors, such as judges, perceive as being the appropriate course of action in a given situation (Clayton and Gillman 1999; Gillman 2001; Shapiro 2002). This means that legal reasons, such as clear and consistent interpretation of the law, might feature among the substantive considerations that shape the judges’ behavior (Baum 1997: 17; Perry 1991).

The logic of consequentialism is the predominant theoretical perspective in most studies of the role of national courts in EU legal integration. Howev-er, whether the claim holds up to empirical testing has yet to be established since there is a lack of studies examining how individual European judges actually reason (Dyevre 2010; Rehder 2014). This means that the question of what constitute the driving forces of judicial decision-making in EU legal integration is still looking for an answer. Moreover, we lack knowledge about the types of substantive reasons or goals that guide judicial behavior in the preliminary ruling procedure. Both of these aspects are crucial for under-standing the roles of national courts and judges in EU legal integration.

The present thesis therefore sets out to analyze the micro-foundations of the national courts’ observed macro-behavior in EU legal integration. It does so by answering the following research question: What types of reasons are individual national judges guided by when making choices in the prelimi-nary ruling procedure? This question is partly explorative since there is vir-tually no previous research investigating the national courts’ decisions in the preliminary ruling procedure from the perspective of individual judges.

29

Overview of the thesis The introductory chapter is followed by the theoretical chapter, which has four main sections. The purpose of this chapter is to chisel out a theoretical model that helps to describe and explain the national courts’ key choices at both the macro and micro levels of analysis. The first section outlines the development of the theoretical debate on national courts in the field of Euro-pean integration. It presents the two main perspectives on national court behavior, the judicial empowerment and sustained resistance approaches, and elaborates on the type of behavioral pattern that they would expect na-tional courts to exhibit. In the second section, the focus is on what might explain the variations in the national courts’ behavior. A survey of the litera-ture on EU legal integration and judicial politics identifies six explanatory hypotheses. The third section shifts the attention from the national courts’ aggregated behavioral patterns to the decision-making of individual judges. By drawing on the literature on judicial politics and historical institutional-ism, this section engages in a discussion about what constitutes the driving forces of judicial behavior. In particular, it focuses on the logic of conse-quentialism and the logic of appropriateness. The final section ties together the three research questions with the expectations formulated in the theoreti-cal model.

Chapter 3 presents the research design of the thesis. Given the thesis’ fo-cus on both the national courts’ aggregated behavior and the behavior of individual judges, the design of the study builds on two different methodo-logical approaches. To answer the first and second research questions re-garding the national courts’ choices at the aggregated level, a quantitative approach was chosen. The empirical material consists of an original data set including 470 randomly selected cases referred from the national courts to the CJEU between 1992 and 2012. The chapter provides theoretical defini-tions of (1) the different types of cases, with regard to their degree of politi-cal sensitivity, that national courts refer to the CJEU and (2) the different types of opinions that national courts express in their requests for prelimi-nary rulings. The process of categorizing the cases and opinions is also de-scribed. Answering the third research question about how individual judges are reasoning in the preliminary ruling procedure requires interviews with judges. Given the question and the method, a single country case study was deemed appropriate and the Swedish judiciary has been singled out as a suit-able case. The choice of studying the Swedish judiciary is motivated in part by practical considerations such as being able to get access to the respond-ents and, in part, by the fact that there is nothing in previous research that suggests that the Swedish judges’ mode of reasoning deviates from how other European judges reason in the preliminary ruling procedure. However, the chapter also discusses whether the Swedish judiciary can be understood as a critical case in relation to the dominant theoretical expectations formu-

30

lated by the literature on US judicial politics and EU legal integration. In the final section the chapter elaborates on the selection of respondents and methodological problems.

The thesis’ analytical chapters 4, 5 and 6 each address one of the research questions. Chapter 4 provides an answer to the question of the type of choic-es that national courts are making in the preliminary ruling procedure. Since this thesis is the first comparative study of the national courts’ two key choices in the preliminary ruling procedure, a systematic description of the decisions that are being made is indispensable. The chapter uses the theoreti-cal expectations derived from the judicial empowerment approach and the sustained resistance approach to make sense of the national courts’ choices. In particular, we want to know whether or to what extent the courts’ deci-sions lead to a behavioral pattern that supports EU legal integration or to one that defends national sovereignty. The analysis shows that the most common behavioral pattern among national courts is to support EU integration. How-ever, it turns out that, when examining the relationship between the two choices, the national courts’ behavior regularly deviates from the theoretical predictions, and two previously untheorized behavioral patterns are identi-fied. This thesis interprets the two behavioral patterns as attempts by the national courts to balance conflicting demands that stem from, on the one hand, the EU level and, on the other hand, the member state level.

Chapter 5 addresses the second research question: What factors can ex-plain the variations in the type of choices that national courts make in the preliminary ruling procedure? Theoretically we want to know why some national courts supporting EU legal integration, while others are defending national sovereignty. The findings show that first instance courts are more likely than higher courts, in particular courts of final instance, to support EU legal integration. Conversely, courts of final instance are in general more likely than all other courts to defend national law. This finding is in line with the expectations of the inter-court competition hypothesis, which claims that the EU legal system empowers lower national courts at the expense of the higher courts. In addition, courts in Denmark, Finland, Sweden and the UK are in general less likely than other courts to provide the CJEU with oppor-tunities to expand the reach of EU law.

Chapter 6 focuses on the micro foundations of the national courts’ aggre-gated behavioral patterns to answer the third research question: What types of reasons are individual national judges guided by when making choices in the preliminary ruling procedure? The analysis draws upon the theoretical assumptions in previous research about what drives judicial decision-making. The dominant perspective, which builds on the logic of consequen-tialism, is contrasted with the logic of appropriateness. The results show that Swedish judges are just as likely to make decisions in the preliminary ruling procedure based on expected consequences as they are to make choices based on considerations of appropriate behavior. Furthermore, and in con-

31

trast to the dominant view in the literature, consideration about how to max-imize one’s self-regarding preferences is not the main driving force of the national judges’ behavior. Rather, other-regarding preferences, formal codi-fied rules and informal professional practices guide the judges in the prelim-inary ruling procedure. In regard to the substantive reasons, the findings show that judges think about matters such as how to safeguard the coherence and functioning of the domestic legal system, how to uphold their own repu-tations among colleagues and how to obey the formal rules.

Chapter 7 concludes the thesis with an analysis and assessment of the findings in relation to the overall aim of the thesis. It discusses the collective contributions of the thesis and its theoretical and empirical implications for future research into national courts in multileveled political systems such as the EU. In particular, it emphasizes how the thesis’ focus on the choices that national courts make in the preliminary ruling procedure and on the behavior of individual judges contributes to redefine the overall debate about the role that national courts play as gatekeepers in EU legal integration.

32

2. A theoretical model for understanding the role of national courts in EU legal integration

The overarching aim of this thesis is to shed light the national courts’ and judges’ behavior as gatekeepers in the preliminary ruling procedure in order to improve our understanding of their role in EU legal integration. A neces-sary first step towards accomplishing this aim is to formulate a theoretical model for understanding judicial behavior in the EU context based on in-sights from the literature on EU legal integration and US judicial politics. The thesis’ starting point is that to understand the national courts’ role as gatekeepers in EU legal integration, their behavior must be examined both from a macro and a micro perspective. This means that the model needs to include both theoretical accounts that deal with aggregated court behavior as well as accounts that dissect the micro foundations of judicial decision-making.

The chapter has four main sections. First, it explores what tentative an-swers previous research has provided to the first research question: What types of choices are national courts making in the preliminary ruling proce-dure? As was discussed in the introductory chapter, two main approaches to national court behavior can be identified in the literature on EU legal inte-gration (Pollack 2013): the judicial empowerment approach, which claims that national courts are to be understood as supporters of EU legal integra-tion (Burley and Mattli 1993; Weiler 1994), and the sustained resistance approach, which characterizes national courts as defenders of national sov-ereignty (Golub 1996; Wind et al. 2009). The section elaborates on how these two behavioral patterns relate to the national courts’ key choices in the preliminary ruling procedure. In particular, it discusses what types of cases the judicial empowerment and the sustained resistance approaches would expect national courts to refer to the CJEU, and what types of opinions each of the two approaches would expect national courts to express in the requests for preliminary rulings.

The theoretical expectations regarding the national courts’ behavioral pat-terns are also connected to the second research question: What factors can explain the variations in the types of choices national courts make in the preliminary ruling procedure? Answering this question sheds light on the role of national courts in EU legal integration by assessing what factors can explain why some national courts are supporting EU legal integration while

33

others are defending national sovereignty. By surveying the literature on EU legal integration and judicial politics (for an overview, see, e.g., Baum 1997; Conant 2007: 54), the chapter identifies six hypotheses about what may ex-plain the variations in court behavior. One of those hypotheses has been derived from the judicial empowerment approach while three come from the sustained resistance approach. The remaining two hypotheses have been derived from the Europeanization literature.

In the third section, the theoretical focus shifts from the courts’ aggregat-ed behavioral patterns to judicial decision-making at the individual level. This change in perspectives is prompted by the focus of the third research question: What types of reasons are individual national judges guided by when making choices in the preliminary ruling procedure? Turning to previ-ous research about the driving forces of judicial behavior, the dominating theoretical perspective assumes that judges are rational actors who are trying to pursue their preferences (Alter 2001; Baum 1997; Burley and Mattli 1993; Dunoff and Pollack 2017; Nyikos 2006). That is, this account primarily sees human reasoning as being guided by the logic of consequentialism (March and Olsen 1989). However, an alternative perspective formulated as the log-ic of appropriateness sees human behavior, including judicial behavior, as rule-bound (Gillman 2001; March and Olsen 1989). Based on these two overarching and diverging perspectives on human behavior, four different characterizations of judicial reasoning in the preliminary ruling procedure can be formulated.

The fourth and final section of this chapter returns to the research ques-tions and provides an outline of how the theoretical model is applied throughout the analytical chapters. The model consists of the following three components (1) a descriptive account of the national courts’ behavioral pat-terns at the macro level and (2) an explanatory account of the variations in these behavioral patterns. Finally, the model includes (3) an account of judi-cial behavior at the micro level, that is, how individual judges are expected to reason when making decisions in the preliminary ruling procedure.

The behavioral pattern of national courts: supporting EU legal integration or protecting national sovereignty? The scholarly interest in the behavior of national courts in European integra-tion begun with a puzzle: Why did national courts, rooted in domestic legal traditions, accept the CJEU’s authority and its bold rulings, such as the su-premacy of EU law over national legislation? (Alter 2009; Burley and Mattli 1993; Shapiro 1992: 127; Weiler 1994). One of the proposed answers to this question forms part of the highly influential theory of European integration that is neofunctionalism.

34

Grounded in a functional logic of regional integration (Haas 1958) the core claim of neofunctionalism is that the EU institutions and subnational actors, such as interest groups, private companies and domestic courts, are the main driving forces of European integration and supranational rule mak-ing (Burley and Mattli 1993: 43; Stone Sweet and Sandholtz 1997: 300). For instance, companies looking to expand their intra-EU trade realize that they would benefit from common trade rules in all member states. Once in place, this legal harmonization increases the exchange across the borders. But for these trade policies to be efficient, adjacent policy areas such as trademark and environmental laws must also become integrated. The supranational institutions, in particular the European Commission and the CJEU, try to facilitate the cross-border exchanges by pushing for more legal harmoniza-tion and offering mechanisms for supranational dispute resolutions. This process is commonly referred to as functional spillover; the initial economic integration and the transnational activities in one area ‘spill over’ and create a need for legal integration in other policy sectors such as the social, envi-ronmental and financial areas (Burley and Mattli 1993: 55; Stone Sweet and Sandholtz 1997: 309).

Expanding on the neofunctionalist theory, Burley & Mattli (1993) place legal integration at the center of their analysis of the driving forces of the overall European integration. Successful EU legal integration is defined as a functioning legal system in which the supranational EU law gradually makes its way into the member states’ domestic legal systems. It includes the ex-pansion of EU law not only in economic policy but also in other areas such as social welfare, health and education (Burley and Mattli 1993: 43). Burley and Mattli (1993) theorize that the successful economic and political Euro-pean integration can be ascribed to the nonpolitical feature of the law. Con-flicting political interests make it impossible for the member states to reach agreements on many policy issues. The legal sphere, however, is governed by nonpolitical objectives such as the rule of law. The transfer of the con-tested policy issues from the political domain to the legal arena makes it possible to achieve outcomes that would have been impossible for political decision-makers to agree on. Central to this argument is that EU law func-tions as a mask that conceals the substantive political implications of the CJEU’s rulings. The political decision-makers in the member states, such as the member state governments, did not realize the political ramifications of the decisions until later (Burley and Mattli 1993: 72).

Given the importance of the law in this neofunctionalist approach, it is hardly surprising that the theory ascribes an important role to both the CJEU and the national courts in the process of European integration. Neofunction-alism claims that the EU institutions and the subnational actors are the driv-ing forces of supranational integration. But what are driving these actors? The short answer is self-interest. The EU legal system creates opportunities for individual litigants, companies, national courts and other subnational

35

actors to pursue their interests (Burley and Mattli 1993: 60). The CJEU has often been understood as an actor that wants to increase its influence over legal development relative to other EU institutions and national political actors (Alter 2001: 53; Alter 2009; Pollack 2013).

The neofunctionlist’s answer to the initial question, why national courts have accepted the CJEU’s authority, is that they did so because it expands their judicial powers (Burley and Mattli 1993: 63; Weiler 1994: 523). Exact-ly how supporting European integration empowers national courts has been further developed by Alter (2001), Mattli & Slaughter (1998) and Carrubba & Murrah (2005), among others. The theoretical basis of these accounts is throughout this thesis referred to as parts of the judicial empowerment ap-proach to national court behavior.

How then does the EU legal system empower national courts? It has been theorized that the self-interest of courts and judges is to increase their inde-pendence, influence and authority (Alter 2001: 45). Courts want to be inde-pendent from the political sphere and free to make their own decisions. Fur-thermore, they want to have the authority to issue final rulings that cannot be appealed. Lastly, independence and authority help the courts to increase their influence and ensure that their rulings have an impact on legal and public policy (Alter 2001: 45-46; Baum 1997). According to the judicial empow-erment approach, the EU legal system helps national courts pursue these interests by giving them new powers. Most importantly, national courts get the opportunity to exercise judicial review.

The de facto practice of supranational judicial review in the EU legal sys-tem means that the CJEU, in its answers to the national courts’ requests for preliminary rulings, often points out that a national policy is incompatible with the supreme EU law. The CJEU’s interpretation of EU law is binding on the national courts but it is still up to the national court to apply this in-terpretation to the case at hand. This means that national courts are entrusted to take part in reviewing whether or not national legislation is compatible with EU law, and issuing a final ruling which may invalidate national law. This practice is assumed to strengthen the national courts’ influence over policy vis-à-vis the domestic legislative and executive branches (Mattli and Slaughter 1998: 190).

The upshot of the judicial empowerment approach is that domestic courts are expected to cooperate with the CJEU to promote further legal integra-tion. How can this perspective help with formulating expectations about what choices national courts make in the preliminary ruling procedure? Re-turning to the national courts’ decision of what types of cases to refer to the CJEU, this choice is highly important since the CJEU is dependent on re-ceiving cases from national courts through the preliminary ruling procedure. Without a full docket, the CJEU will not be able to define how the EU law is to be interpreted and its ability to expand the scope of EU legal integration is circumscribed (Golub 1996: 376-377). However, it also matters what types

36

of cases the CJEU receives. If the bulk of the cases concern narrow technical issues that do not involve questions about the relationship between the EU law and national law, the CJEU will not be able to pave the way for further EU legal harmonization in the member states (Alter 2000: 501; Wind 2010: 1052). This means that in order for the CJEU to be able to foster integration, it needs to get its hands on cases that concern the relationship between na-tional law and EU law, that is, cases that have a high degree of political sen-sitivity.

The judicial empowerment approach would therefore expect national courts to allow the CJEU to decide even those types of cases which have a high degree of political sensitivity and that may challenge national policies. Similarly, turning to the decision about including opinions in the requests for preliminary rulings, the judicial empowerment approach would expect na-tional courts to make choices that favor EU integration. This means that national courts will try to frame the legal issue in a way that opens up for the CJEU to expand the reach of EU law. By expressing opinions in support of EU law, national courts can signal to the CJEU that an expansion of the EU legal scope is accepted and even encouraged.

The neofunctionalist take on EU judicial politics, which permeates the ju-dicial empowerment approach, is first and foremost a theory about the initial legal integration in the EU. As such, it has been criticized on empirical grounds for not being able to explain the many setbacks in the integration process. On theoretical grounds, many have questioned the neofunctionalist assumptions about the driving forces of European integration. In other theo-retical perspectives on European integration, such as intergovernmentalism (Garrett 1995; Garrett et al. 1998; Garrett R. and Barry 1993) and liberal intergovernmentalism (Moravcsik 1993), it is the member states’ govern-ments that set the pace for European integration and not supranational insti-tutions or subnational actors. As a consequence, these theories do not elabo-rate on what behavior we should expect from the domestic courts.

There are, however, other accounts that are critical of the neofunctionalist assumptions that are inherent in the judicial empowerment approach but that still view national courts as important actors. These accounts have been re-ferred to as part of the sustained resistance approach to national court be-havior (Pollack 2013: 1273). This approach claims that national courts are unlikely to become empowered by participating in the EU legal system. It is therefore also unlikely that national courts have an interest in supporting the CJEU. On the contrary, national courts are believed to face strong disincen-tives to engage in collaboration with the CJEU in the preliminary ruling pro-cedure because of the high political stakes involved (Golub 1996). The dis-incentives include the risk of the CJEU overturning national policies and limiting member state sovereignty as well as undermining legal certainty (Alter 2001: 48; Dehousse 1998: 173; Golub 1996: 361; Wind 2010; Wind et al. 2009).

37

Avoiding collaboration with the CJEU means in this account that national courts do not request preliminary rulings. By withholding references from the CJEU, national courts accomplish two things: they are able to slow the pace of EU legal integration, and they remain in control over specific policy outcomes (Golub 1996: 381). For example, avoiding the preliminary ruling procedure and instead deciding the case themselves allows national courts to protect traditional member state practices and policies (Golub 1996: 379). Why do national courts choose to protect the member state policies? Accord-ing to Golub, the reason for this is that domestic political factors of different types put pressure on the national courts to not refer cases to the CJEU (Golub 1996: 378). One of the proposed mechanisms is that member state governments issue authoritative statements instructing national courts to be restrictive when it comes to using the preliminary ruling procedure (Wind et al. 2009), or that such statements are made by the highest judicial authority in the member state (Craig 1998: 205).

Another proposed mechanism is that national courts are less likely to em-brace the preliminary ruling procedure due to a “prevailing political climate of Euro-pessimism” (Golub 1996: 377). Similarly, Volcansek argues that the refusal of national courts to refer cases to the CJEU can be understood in terms of overall negative sentiment towards EU integration based upon spe-cific economic and political conditions (Volcansek 1986: 206).

Finally, national courts have been theorized to put up a sustained re-sistance because they are concerned about the stability of the domestic legal order (Alter 2001: 48; Maher 1998). According to Dehousse (1998), it is common that national courts perceive that EU law has a disruptive effect on the coherence of the national legal order. The EU norms may collide with existing domestic legal traditions and undermine the predictability of the member state’s legal system: “what appears as integration at the European level is often perceived as a source of disintegration from the perspective of national legal systems” (Dehousse 1998: 173).

How would the sustained resistance approach answer the questions of what types of choices national courts make in the preliminary ruling proce-dure? Overall, this approach expects the national courts to exhibit a behav-ioral pattern that puts up resistance to EU legal integration. In the original account, this would mean that national courts do not participate in the pre-liminary ruling procedure at all, or that they only participate to a limited extent; for example, by avoiding referring certain types of cases to the CJEU (Golub 1996: 381; Wind 2010). Drawing upon this account, it can be hy-pothesized that national courts do not refer cases that can be considered to have a high degree of political sensitivity to the CJEU. The reason for this is that those cases concern the relationship between national law and EU law, and referring this type of case to the CJEU is something that is likely to ena-ble the CJEU to accelerate the pace of EU legal integration. This is contrary to the national courts’ goal of resisting further legal harmonization. When it

38

comes to the second choice, what types of opinions are included in the re-quest, the expectation derived from the sustained resistance approach is that national courts express their support for the national law. In that way, the courts signal to the CJEU that an expansion of the EU legal scope is not accepted in the member state.

In sum, two types of behavioral patterns emerge. Based on the judicial empowerment approach, we would expect national courts to support EU legal integration by referring cases with a high degree of political sensitivity and expressing opinions in support of EU law. In contrast, the behavioral pattern derived from the sustained resistance approach leads to the opposite prediction. National courts defend national sovereignty by avoiding making referrals in cases with a high degree of political sensitivity, and by including opinions in support of national law.

These two characterizations are useful for theorizing about the implica-tions of the national courts’ choices in relation to their role as gatekeepers in EU legal integration. However, all national courts are not expected to behave in the same way. In fact, the initial explorations of the national courts’ refer-ral patterns revealed considerable variations across member states (Sweet and Brunell 1998b). Furthermore, the results from the few existing empirical case studies on the national courts’ key choices have shown that the national courts in Denmark (Wind 2010) and the UK (Golub 1996) appear to display a sustained resistance behavior. Another study looks at the content of the opinions expressed by the Swedish courts and finds that they come closer to the expectations derived from the judicial empowerment approach (Leijon and Karlsson 2013). This opens up for a theoretical discussion about what may explain variations in national court behavior, a question which is ex-plored in the next section.

Explaining variations in national court behavior The initial explorations of the national courts’ behavior in the EU legal sys-tem revealed that they had been cooperating with the CJEU by requesting preliminary rulings. This finding was taken as support for the claim that national courts had uniformly accepted the CJEU’s authority and the su-premacy of EU law (Burley and Mattli 1993; Weiler 1994). It is therefore not surprising that the early neofunctionalist theory did not formulate any hypotheses explaining the differences in the national courts’ behavioral pat-terns. But while it is true that the CJEU’s caseload has continued to increase over the years thanks to the national courts’ references, there have also been variations in the relative share of cases sent from courts in different member

39

states.6 This discovery has inspired researchers to formulate hypotheses about why we see this variation in judicial behavior in the preliminary ruling procedure. These hypotheses frequently draw upon the theoretical outlooks of the judicial empowerment and the sustained resistance approaches (Carrubba and Murrah 2005; Sweet and Brunell 1998b).

As discussed above, there is a lack of studies on what types of choices na-tional courts make in the preliminary ruling procedure. The main empirical focus of previous research has instead been to explain the variation in the number of referred cases. Can these studies in some way contribute to an-swering the question about what explains the variation in what choices na-tional courts make? Yes, since the research on referral rates has been driven by the same theoretical interest in understanding why national courts either support EU legal integration or defend domestic policies. It is therefore both possible and fruitful to elaborate on how these explanations can help us un-derstand variations in the national courts’ key choices: what types of cases they refer to the CJEU and what types of opinions, if any, they express in their requests for preliminary rulings. The following section outlines the theoretical basis behind each of the six hypotheses.

Explanations derived from the judicial empowerment approach Researchers who have expounded on the original neofunctionalist account have maintained that judicial empowerment is key for understanding nation-al court behavior. However, they differ in one important aspect. They do not claim that EU legal integration and the preliminary ruling procedure is em-powering all national courts. Instead, it is the lower domestic courts that gain influence at the expense of the courts of final instance. This theory is com-monly referred to as the inter-court competition hypothesis (Alter 1998b; Alter 2001).

Inter-court competition The inter-court competition hypothesis is based on the neofunctionalist claim that the preliminary ruling procedure empowers the national courts by giving them the opportunity to exercise judicial review. By referring cases to the final authority on EU law, the CJEU, the lower courts will be able to deliver rulings that the highest domestic courts will also have to accept. With the support of the CJEU’s answers to the legal questions, all national courts can create precedents and exercise judicial review, powers that previously were reserved for the courts of final instance (Alter 2001: 48-49). Hence, the pre-liminary ruling procedure flips the traditional judicial hierarchy.

6 The differences remain even after controlling for population size, which is believed to be correlated with the share of EU-related disputes in a country (Wind 2009).

40

A court’s place in the domestic legal system is believed to have an effect on how it will behave in the preliminary ruling procedure. However, it is not clear what behavior we should expect from the intermediate courts (Sweet and Brunell 1998b: 71). Sometimes high courts are referred to as ‘courts of last instance’, suggesting that this category only includes the courts against whose decision there is no judicial remedy. Lower courts, on the other hand, are sometimes referred to as ‘first instance courts’ (Alter 2009: 100) while other times as ‘all courts other than the Constitutional Court or the Supreme Court’ (Alter 2001: 49).

Building on interviews with different judicial actors and an in-depth study of doctrinal change in Germany and France between 1963 and 2000, Alter found support for the inter-court competition hypothesis among German and French judges (Alter 2001). Subsequent studies have set out to test this hy-pothesis by comparing referral rates across court levels (Sweet and Brunell 1998b: 89-90; Wind et al. 2009). The starting point of this research is that a relatively high number of references indicate that national courts are con-tributing to EU legal integration. Descriptive data on the overall number of referrals made between 1961 and 1995 show that first instance courts and courts of appeal are together responsible for the vast majority of the refer-ences sent to the CJEU (Sweet and Brunell 1998b: 89-90). However, the courts’ behavior varies between member states. In four member states7, first instance courts refer the greatest amounts of cases to the CJEU. Yet in seven member states, the majority of the cases come from intermediate courts8 (Sweet and Brunell 1998a: 89).

The theoretical proposition and the empirical findings suggest that the courts’ position in the judicial hierarchy is a factor that needs to be taken into account when trying to understand variations in national court behavior. This gives us reason to believe that inter-court competition also may play a role for other kinds of court behavior in the preliminary ruling procedure, such as what types of cases national courts refer to the CJEU and what types of opinions they express in the request.

Explanations derived from the sustained resistance approach Research that draws upon the sustained resistance approach has formulated three different hypotheses aimed at explaining the variation in national court behavior. These hypotheses center upon the institutional characteristics of the member state, the public support of European integration and the charac-ter of the legal case at hand.

7 Belgium, Germany, Greece and Portugal. See Sweet, Alec Stone & Brunell, Thomas L. (1998b) The European Court and the National Courts: A Statistical Analysis of Preliminary References, 1961–95. Journal of European Public Policy, 5, 66-97. 8 Denmark, France, Ireland, Italy, the Netherlands, Spain, and the UK. See ibid.

41

Majoritarian democracy Previous research has suggested that “the institutional features of law and politics and the traditional means of organizing their mutual relationship might influence the willingness of national courts to refer matters to the ECJ” (Wind et al. 2009: 71). The proposed hypothesis is that national courts from member states that have a tradition of what Wind et al. (2009) refer to as majoritarian democracy are reluctant to participate in the preliminary ruling procedure when compared to courts from so-called constitutional democracies. The key difference between the two types of democratic tradi-tions is how the relationship between the political and judicial branch is structured (Wind 2010; Wind et al. 2009).9

According to the authors’ definition, a majoritarian democracy is charac-terized by a strong tradition of parliamentary sovereignty while the role of courts as a counter-majoritarian force is restricted. The member states that have been identified as belonging to this tradition are Denmark, Finland, Sweden and the UK. The judicial branch in these four countries has a limited tradition of actively reviewing legislation in accordance with the constitu-tion. For instance, judicial review was prohibited in Finland up until 2000 and even though the practice is allowed in Sweden and Denmark it is seldom used (Wind et al. 2009: 72). The other EU member states are identified as having a tradition of constitutional democracy. What is characteristic of the-se member states is that the courts not only have the formal right to perform judicial review but also that this practice is seen as legitimate (Wind et al. 2009: 73).

How then does a majoritarian democratic tradition play a role in the na-tional courts’ behavior in the preliminary ruling procedure? One of the pro-posed mechanisms is that the member state government issues authoritative statements instructing national courts about how they should use the prelim-inary ruling procedure. Commonly, national courts have been instructed to limit the number of cases national courts send to the CJEU. For example, in Denmark the Judicial Committee, part of the Danish Ministry of Justice, recommends that national courts only request preliminary rulings in those cases in which the Commission has already initiated direct action against Denmark (Wind et al. 2009: 76). Another example can be found in the UK, where representatives of the House of Lords instructed other courts to inter-pret the EU law themselves rather than sending those cases to the CJEU (Craig 1998: 205; Wind et al. 2009: 77-78). What is important to note in

9 The way Wind et al. (2009) and Wind (2010) define a ’majoritarian democracy’ differs from other authors. For example, Lijphart makes a distinction between majoritarian democracy (or the Westminster model of democracy) and consensus democracy (Lijphart, Arend (1999) Patterns of Democracy : Government Forms and Performance in Thirty-Six Countries, New Haven, Conn.: Yale University Press.) In this thesis, however, the term “majoritarian democ-racy” will refer to the definition proposed by Wind et al. (2009) and Wind (2010), which focuses on a strong tradition of parliamentary supremacy and a limited role for the courts.

42

both cases is that there appears to exist a common understanding between the judiciary and the political branch that they should work closely together in order to provide for an efficient and legitimate relation between EU law and national law. As a result of this shared understanding, national courts in majoritarian democracies are expected to pay attention to the statements issued by the government and follow the instructions (Wind et al. 2009: 76).

The close cooperation between the judicial and the political branch, and the authoritative statements from the government, are believed to have left an imprint on the courts’ behavior in the preliminary ruling procedure. In-deed, the results from a survey among Danish judges show that close to 70 percent of the respondents list the advice from the judicial committee, pre-sented to them by the Danish state prosecutor,10 as the main reason for not requesting preliminary rulings (Wind 2010: 1051; Wind et al. 2009: 76). It should be emphasized that this is a unique and rather controversial institu-tional set-up. That a body tied to the Ministry of Justice and the Ministry of Foreign Affairs advises courts on matters of EU law is problematic if the judiciary is to be independent from the other branches of government. Wind goes so far as to note that the separation of powers between the courts and the legislative and executive branches is weak in Denmark (Wind 2010: 1052).

Although this particular institutional set-up might be unique to Denmark, it can be theorized that other member state governments with the same ma-joritarian tradition also try to signal to the judiciary how and when the pre-liminary ruling procedure should be used. It can either take the form of out-spoken advice from the government representatives or be part of a wider implicit or explicit understanding within the domestic political and judicial system about how to deal with the preliminary ruling procedure. National courts are expected to take these signals into account since the tradition of majoritarian democracy leads the judiciary to be loyal to the political branch and show deference to the legislator (Nergelius. 2001: 88; Wind et al. 2009: 73). In sum, it is theorized that courts in member states that have a tradition of majoritarian democracy (Denmark, Finland, Sweden and the UK) are more reluctant to engage in the preliminary ruling procedure and less sup-portive of EU legal integration than courts from other member states.

Support for the hypothesis and the proposed mechanism can be found in a number of case studies (Craig 1998; Wind 2010; Wind et al. 2009). In addi-tion, a statistical analysis of all requests for preliminary rulings in the period 1961-2004 found that courts in member states that have an institutional lega-cy of majoritarian democracy have fewer referrals to the CJEU when con-trolling for other relevant variables such as population size and intra-EU trade (Wind et al. 2009: 79). Since the member state’s tradition of majoritar-ian democracy has been found to influence how national courts engage in the 10 Also known as the Legal Advisor to the Danish State (Kammeradvokaten)

43

preliminary ruling procedure, it can be theorized that it also has an impact on what key choices national courts make in the preliminary ruling procedure.

Public support for EU integration The question of whether public opinion influences what decisions courts and judges make has been the topic of much scholarly debate in both the US and European judicial politics literature (Brace and Boyea 2008; Casillas et al. 2011; Giles et al. 2008; Mishler and Sheehan 1996). Previous research has suggested that one of the causal pathways for a direct influence of public opinion on judicial decision-making is the ‘political adjustment’ thesis (Mishler and Sheehan 1996: 174), or the ‘rational anticipation’ hypothesis (McGuire and Stimson 2004). The argument is that any court that cares about its perceived institutional legitimacy must try to anticipate whether its decisions will be respected and followed by the public (McGuire and Stimson 2004: 1019). The rational behavior of a court confronted with pub-lic opinion opposed to its decisions is to modify its behavior in order to pro-tect the legitimacy of the court and policy effectiveness. Otherwise, the risk is that the courts’ decisions are rejected (Giles et al. 2008: 295).

In the EU legal context, it has been theorized that low public support for European integration in a member state makes national courts less likely to support EU legal integration and also less likely to refer cases to the CJEU (Carrubba and Murrah 2005: 405-406; Mattli and Slaughter 1998). This is explained by the rational anticipation hypothesis. The courts need to be per-ceived as legitimate by the public, and their legitimacy is closely tied to the citizens’ support of the courts’ decisions (Caldeira and Gibson 1992; Staton 2006). If the citizens are skeptical to EU membership, they will not be fond of having their own courts sending cases to the CJEU, which has a track record of delivering rulings that foster European integration. The Euroscep-ticism among the citizens works as a constraint on the behavior of the na-tional courts in the preliminary rulings procedure. It makes them less likely to accept the CJEU’s doctrines and, as a consequence, less likely to engage in the preliminary ruling procedure (Mattli and Slaughter 1998: 197-198).

Previous empirical research on the impact of public opinion on national court behavior has yielded mixed results. Some statistical analyses of the national courts’ references to the CJEU have found support for the hypothe-sis. That is, in member states in which the support for the EU is relatively high, national courts are more likely to refer cases to the CJEU than courts from member states in which the support for EU is relatively lower (Carrubba and Murrah 2005: 412; Vink et al. 2009: 18). However, other studies found no such effect (Hornuf and Voigt 2015: 306; Mayoral 2012: 91). Although there is no conclusive evidence that the level of Euroscepti-cism in the member state matters for how national courts behave in the pre-liminary ruling procedure there is enough for at least considering its impact on other aspects of national court behavior. That is, does the level of public

44

support for the EU matter for what types of cases national courts refer to the CJEU and what types of opinions they express?

The political sensitivity of the case Research on what types of choices national courts make in the preliminary ruling procedure has pointed out that what types of cases courts are con-fronted with may influence whether or not they include opinions in the re-quest for a preliminary ruling. Nyikos has theorized that opinions are an “opportunity for national courts to stack the interpretive deck for the entire decision-making process” (Nyikos 2006: 527). The goal of including opin-ions in the request for a preliminary ruling is, according to Nyikos, to influ-ence CJEU’s interpretation of EU law (Nyikos 2006: 631).

However, writing opinions is time consuming. National courts are there-fore expected to be more likely to invest that time when they are more cer-tain that the CJEU will actually take their views into account (Nyikos 2006: 532). It has been suggested that the CJEU is more likely to listen to the opin-ions of the national courts when the legal issues in the case at hand concern the controversial question about the relationship between EU law and na-tional law (Nyikos 2006: 533). Empirical research shows that national courts are in fact more likely to include opinions in the request for a prelimi-nary ruling when the case concerns a policy area in which CJEU interven-tions are less accepted (Nyikos 2006). Examples of those opinions show that national courts express support for national law.11 This finding suggests that the character of the EU legal case being referred matters for what types of opinions national courts express.

Explanations derived from the Europeanization literature Apart from the two main approaches, sustained resistance and judicial em-powerment, explanations for national court behavior have also been derived from the Europeanization literature. In general, the main interest of this body of research is to understand how the member states are affected by EU membership and, alternatively, how the member states shape the EU (Börzel 2002; Börzel 2003; Green Cowles et al. 2001). This includes how national courts, an important member state institution, are influenced by EU legal integration and, in turn, influence the EU (Conant 2001; Ladrech 2010).

11 See, for example, Judgment of the Court of 10 April 1984. Sabine von Colson and Elisabeth Kamann v Land Nordrhein-Westfalen. Case C-14/83. ECLI:EU:C:1984:153 and Judgment of the Court of 4 October 1991. The Society for the Protection of Unborn Children Ireland Ltd v Stephen Grogan and others. Case C-159/90 ECLI:EU:C:1991:378. Nyikos (2006) discusses opinions found in these two cases.

45

Experience of the preliminary ruling procedure A core assumption in the theoretical framework of Europeanization is that the EU exerts an adaptational pressure on the member states’ governance structures, which is expected to result in institutional change (Green Cowles et al. 2001). However, national institutions, including courts, do not change their behavior overnight. It usually takes time for institutions to adapt to the rules and procedures in the EU political system (Conant 2001; Ladrech 2010). It has been suggested that changes in the domestic structures are the result of a learning process in which the member state institutions over time become familiar with the new EU governance system. This process consists of repeated interactions between the national institutions and the EU level by which the institutions start to adapt their behavior to the EU practices (Börzel 2003).

In the case of the preliminary ruling procedure, it means that by referring cases to the CJEU, national courts will learn how to formulate questions and what information to include in the requests. For example, there is a tendency that courts in countries that have been EU members for a long time refer more cases to the CJEU compared to courts in the newer member states (Golub 1996; Hix and Høyland 2011; Mayoral 2012). Furthermore, the judi-cial dialogue that takes place between the CJEU and the member state courts is expected to enhance the legitimacy of the preliminary ruling procedure in the eyes of the national judiciaries. The repeated interactions with the CJEU will convince the national courts that the CJEU and the procedure have a necessary role in the EU political system (Ladrech 2010: 118). Thus, nation-al courts’ past experience of the EU legal system is believed to have an ef-fect on their behavior in the procedure, making them more positive towards the procedure and EU legal integration.

That a learning curve exists when it comes to the number of requests for preliminary rulings has been demonstrated by Stone Sweet & Brunell (1998). That the number of requests for preliminary rulings has increased over time is also what Mayoral (2012) finds. However, in a study by Wind et al. (2009), the hypothesis is rejected. Although previous research has pro-duced mixed result regarding the importance of experience for the national courts’ behavior in EU legal integration, it is a factor worth considering in relation to the national courts’ key choices in the preliminary ruling proce-dure. It remains to be seen if experience of the preliminary ruling procedure matters for what types of cases national courts refer to the CJEU and what types of opinions are included in those cases.

Concrete judicial review Previous research has put forward the domestic tradition of judicial review as a factor that affects the national courts’ behavior in the preliminary ruling procedure (Carrubba and Murrah 2005; Mattli and Slaughter 1998; Sweet

46

and Brunell 1998b; Tridimas and Tridimas 2004). The claim is that courts that are allowed to review the constitutionality of national legislation are more likely to accept the preliminary ruling procedure than courts from legal systems that lack judicial review. The explanation lies in the similarity of the two legal procedures; they both require courts to review the constitutionality of legislation in concrete cases (Carrubba and Murrah 2005; Wind 2010). In this context, the term ‘judicial review’ refers to the power of all courts to review the constitutionality of legislation in the actual cases brought before them, sometimes described as ‘concrete’ judicial review or ‘decentralized’ judicial review (Mayoral 2012; Vink et al. 2009). In the Europeanization literature, this is understood in terms of the relative ‘goodness of fit’ between the individual member state’s political system and the EU. A relatively bet-ter fit occurs when the member state shares institutional features with the EU (Börzel 2003; Green Cowles et al. 2001).

For the courts that already perform judicial review at the national level, the preliminary ruling procedure is simply a natural extension of their na-tional judicial powers (Vink et al. 2009). Conversely, courts from domestic legal systems in which judicial review is prohibited are expected to have a difficult time accepting the preliminary ruling procedure since it requests them to review and set aside legislation, a practice that stands in sharp con-trast to what they normally are allowed to do (Sweet and Brunell 1998b: 68). The suggestion is that we can expect to find a difference in behavior between courts that are familiar with reviewing the constitutionality of legislation and courts that have never previously practiced judicial review.

The empirical examination of this hypothesis has so far yielded mixed re-sults. Most of the existing research uses quantitative methods and focuses on how different traditions of judicial review influence how many requests for preliminary rulings are made by national courts. A study of variations in referral rates between 1970 and 1998 shows that member states in which judicial review is allowed do not produce more references than member states without a tradition of the practice (Sweet and Brunell 1998b: 73). However, courts in member states in which only abstract judicial review is allowed make fewer references than courts in member states without any type of judicial review (Carrubba and Murrah 2005: 412). Similar results have also been found in research on the behavior of national courts from 1995 to 2006 (Vink et al. 2009: 18). A study covering the years 1973-2011 reveals that courts in member states in which only high courts can perform judicial review refer more cases to the CJEU than courts in other member states (Mayoral 2012: 91). In addition, a tradition of concrete judicial review has also been found to make courts more likely to include opinions in the requests for preliminary rulings (Nyikos 2006: 541).

In sum, previous research has produced mixed results when it comes to the impact of judicial review on national court behavior. However, since this is the first study to investigate the variation in the national courts’ key choic-

47

es, it is reasonable to investigate whether or to what extent a practice of con-crete judicial review has any relationship with what types of cases national courts refer to the CJEU and what types of opinions they express in the pre-liminary ruling procedure.

The micro foundations: What guides the behavior of individual judges in EU legal integration? This chapter has so far been occupied with formulating theoretical expecta-tions about the national courts’ aggregated behavioral patterns. However, it is not the court as an institution that decides what types of cases to refer to the CJEU or what opinions to express; it is the individual national judge. If we do not pay attention to how the judges experience the preliminary ruling procedure, we run the risk of misinterpreting what constitutes the driving forces of EU legal integration. To understand the national courts’ role as gatekeepers in EU legal integration it is therefore necessary to theorize about the judges’ actions on the micro level. In particular, we want to know what types of reasons judges have for making decisions in the preliminary ruling procedure.

Assumptions about what is driving the actions of national judges are both explicit and implicit in previous research. As was elaborated on in the first section, the two main theoretical approaches, judicial empowerment and sustained resistance, have formulated statements about why the individual judges can be expected to make certain choices. For instance, the judicial empowerment approach assumes that judges support EU integration because it expands their powers (Alter 2001; Burley and Mattli 1993; Weiler 1991). These approaches subscribe to the notion of methodological individualism, meaning that all social phenomena can, in principle, be attributed to individ-ual actions (Arrow 1994; Elster 1982). In this case, the national courts’ ob-served macro behavior to either promote or resist further integration is traced back to considerations made by the individual judge. But can we be sure that national judges actually think about how to expand their powers when mak-ing choices in the preliminary ruling procedure? Assessing the validity of this theoretical expectation requires a discussion about how previous re-search understands the driving forces behind human behavior and judicial decision-making.

What things do most people take into consideration before making a deci-sion? This question of what factors determine human behavior is perennial in social science research and its answer is crucial for understanding social phenomena. No one would argue that there is only one factor that explains human action. Yet researchers commonly assume that the actor’s behavior is guided by one primary way of reasoning. Although there are many different

48

ways to define and label those modes of reasoning, this thesis uses two basic theoretical concepts: the logic of consequentialism and the logic of appro-priateness (MacIntyre 1988; March and Olsen 1989: 23). These concepts provide two clear and contrasting definitions of how human action can be understood. The logic of consequentialism builds on the assumption that people are rational actors whose choices shape societal rules and institutions (Downs 1957: 3; Epstein 1990; Heath 1976). The term is often used inter-changeably with rational choice (Ostrom 1991: 238). The decision-making process for an actor guided by the logic of consequentialism begins with the question: How do I maximize my preferences in this situation? (March and Olsen 1989: 23) In contrast, the logic of appropriateness refutes this vision of human behavior and proposes instead that human action is rule-bound. In a given situation, actors are believed to be asking themselves questions such as: What is the appropriate response to this situation given my position and responsibilities? Human behavior is in this mode of reasoning shaped by what rules, norms, institutions and procedures prescribe as constituting the appropriate course of action (March and Olsen 1989: 23-24).

Together, the logic of consequentialism and the logic of appropriateness capture the principal divide in the scholarly discussion about what consti-tutes the main driving forces of human behavior (March and Olsen 1989: 23; Shapiro 2002: 99). This divide is also present in the literature on judicial decision-making, which will be explored in the upcoming sections.

Consequentialism in judicial politics The dominating theoretical outlooks in the political science literature on judicial behavior are strongly informed by the logic of consequentialism. The key assumption is that judges base their decisions on calculations of how to achieve the realization of their preferences, be it preferences over policy outcome (Epstein and Knight 1997: 3; Segal and Spaeth 2002), career prospects (Epstein et al. 2013; Posner 1993) or earning respect from others in the legal community (Baum 1997; Epstein and Knight 1997: 4; Maltzman et al. 1999: 46). The vast majority of the scholarship on judicial behavior has had the US legal context as its main field of study. As a result, the predomi-nant theories about judicial decision-making originated as attempts to ex-plain the behavior of US judges (Baum 1997; Dyevre 2010). This has impli-cations for how the theories are to be applied in research on European judg-es, a matter we will have reason to return to later.

A useful starting point for exploring the field of judicial politics is to fo-cus on the research on the most well-known US judges, namely the nine justices who make up the US Supreme Court. Since the US Supreme Court is ranked as one of the most powerful courts in the world, it is not surprising that much previous research has been dedicated to finding out what deter-mines the justices’ decisions. What distinguishes theories on judicial behav-

49

ior in the US Supreme Court from other theories on judicial decision-making is the focus on the justices’ policy preferences (Baum 1997: 26-27). The justices are described as rational actors who try to introduce their policy preferences into the law (Maltzman et al. 1999: 46). To be sure, few if any scholars would argue that policy preferences alone shape the justices’ behav-ior but many would maintain that it is one of the most important factors (Baum 1997).

However, while most researchers agree that justices primarily care about policy outcomes, they disagree about the extent to which justices are free to pursue those goals. According to proponents of the influential attitudinal model of judicial decision-making, justices do have the ability to vote their ideological preferences when deciding cases on their merits (Segal 1999: 238). This is made possible by the justices’ unique professional position (Baum 1997: 35). For example, lower court judges may think about how their behavior will either increase or decrease their chances of being promot-ed to a more prestigious position (Baum 2006; Black and Owens 2016; Epstein et al. 2013). In contrast, the justices of the Supreme Court have al-ready attained the highest judicial office and are granted life tenure. For this reason, the justices are not expected to think about their career (Segal and Spaeth 1993: 69-72). Instead, the justices’ institutional protection ensures that they are free to decide cases on the basis of their sincere policy prefer-ences (Segal 1997; Segal and Spaeth 1993; Segal and Spaeth 2002). For example, this means that the justices who hold liberal beliefs (as compared to conservative ones) are expected to vote in favor of the most liberal out-come in the cases brought before them (Baum 1997; Dyevre 2010: 300). Previous research claims to have found support for this hypothesis. For ex-ample, high correlations have been discovered between the justices’ ideolog-ical positions and their votes (Segal 1999: 239).

The attitudinal model is challenged by another prominent theory on judi-cial behavior: the strategic model of judicial decision-making. The strategic model questions the assumption that justices are as insulated as the attitudi-nal model suggests that they are. This critique builds on the notion that jus-tices of the US Supreme Court face various constraints that influence their behavior. According to the strategic model, ideological preferences are still key but the behavior of the justices is not always a reflection of their genuine policy preferences. Instead, they are expected to make decisions based on strategic consideration of how to achieve their policy goals (Bailey and Maltzman 2011; Epstein and Knight 1997; Helmke and Sanders 2006: 867; Murphy 1964). Being strategic means that justices think about the conse-quences of their actions and what may be the likely response from other actors. The justices’ choices are therefore shaped by how they anticipate other actors, such as the legislative and the executive branches, to behave (Maltzman et al. 1999: 47). For instance, if justices fear that their decisions will be overturned by Congress, they will strategically adjust their vote in

50

order to avoid legislative override (Knight and Epstein 1998; Maltzman et al. 1999). There are numerous empirical studies that claim to have found sup-port for such strategic behavior (for an overview, see, eg., Epstein and Knight 2000: 640).

Although the US Supreme Court in many ways differs from other courts, the theoretical assumption, in particular about how judges employ a conse-quentialist mode of reasoning and try to maximize their preferences, has been highly influential in the broader theoretical debate on judicial behavior both in the US and beyond (Baum 1997; Hall 1995; Hönnige 2009; Pellegrina and Garoupa 2013). When turning to research on judicial behavior in European integration, one finds that the logic of consequentialism also occupies the dominant position in this field (Carrubba and Gabel 2015; Carrubba and Murrah 2005; Conant 2013; Larsson and Naurin 2016; Obermaier 2008; Sweet and Brunell 1998b; Tridimas and Tridimas 2004).

A review of the literature shows that the assumptions about the behavior of national judges in the EU legal system closely resemble those articulated in the discourse on US judicial politics (Conant 2007; Dyevre 2010; Rehder 2014). Weiler was one of the first to suggest that the national courts’ ac-ceptance of EU law likely stemmed from “plain and simple judicial empow-erment” (Weiler 1994: 523). Other have claimed that the national judges are expected to be “ideologically motivated actors seeking to maximize their policy objectives ” (Nyikos 2006: 530). Similarly, Tridimas and Tridimas argue that: “National courts will refer to the ECJ and consequently apply its ruling when the expected net utility gains (benefits minus costs) from doing so exceed the utility gains from not referring” (Tridimas and Tridimas 2004: 135).

What can be concluded is that the literature on judicial decision-making mainly perceives of judges as actors who make decisions based on expected consequences. This is the case in both the US and the EU contexts. When faced with making a decision, the judges are assumed to think about what alternatives they have and what consequences each different alternative has for the materialization of their preferences. They will then choose the course of action that is likely to produce the best outcome given their preferences. But what preferences are individual judges trying to maximize, apart from the special case of policy goals? The next sections survey how the literature has defined the judges’ preferences, beginning with those related to the judges’ self-interest.

Self-regarding preferences In line with the consequentialist understanding of human action, much of the previous theoretical works in the field of judicial politics emphasize the judges’ wish to promote their self-interest (Dunoff and Pollack 2017; Spiller and Gely 2008). This research commonly draws upon economic concepts of

51

utility, costs, benefits and self-regarding preferences. The point of departure is that judges are not that different from other people and we should there-fore expect them to try to maximize the same types of preferences that eve-rybody else does (Posner 1993). A common understanding of a reasoning based on self-regarding preferences is that the purpose of the action is to achieve a personal benefit (Holley 1999: 3). For judges, personal benefits have been theorized to include goals such as securing a promotion to a high-er court, being popular and respected in the legal community and having power within the court (Baum 1997: 17; Dunoff and Pollack 2017: 241; Hall 1995; Miceli and Coşgel 1994; Schauer 1999).

To take one example, Miceli and Coşgel (1994) have argued that judges consider how their behavior, especially their rulings, may affect their reputation. In their work the judges’ self-interest is defined as protecting their reputation. Why then do judges care about their reputation? The argument is that reputation is important both for its own sake and for how it might influence the judges’ careers. The judges are expected to think about how they are viwed by collegues, legal scholars and, the general public (Miceli and Coşgel 1994: 32). The authors identify the judges’ way of justifiying their decision through opinions as key for understanding how their reputation is formed. In the opinions the judges provide argument as to why they have followed precedents or not. In the case that a judge presents a new precedent in the opinion, he or she wants other courts to accept this precedent and follow it. If this happens, the judge’s reputation is enchanced. But if the judge fails in this attempt to set a precedent and the decision is not followed but instead reversed by a higher court, his or her prestige might decrese (Miceli and Coşgel 1994: 35).

The judges’ potential concern about their reputation has also been inves-tigated in a study on how judicial actors experience the use of plea barging. Heumann finds that judges accept plea barging because it gives them several rewards while simultaneously protecting them from sanctions (Heumann 1978: 144-146). A plea barging means that the defendant relinquishes his or her right to go to trial and pleads guilty in exchange for reduction in the charge or the sentence (Heumann 1978: 1). Going along with the plea bar-gain system means that the judges can avoid the possibility of appeals and having their decisions overturned by a higher court. One of the interviewed judges states that:

Oh yes, we do worry about reversals on appeal, certainly. Some judges are apprehensive about their reputations, not only among fellow judges and the bar, but also on the Supreme Court. If you make a lot of wrong decisions, the Supreme Court keeps sending down ‘error, error, error’ (Heumann 1978: 144).

52

Based on the findings, Heumann theorizes that some judges support the use of plea barging since it is time-efficient and thereby helps them satisfy ad-ministrative demands for disposition of cases. The interviewed judges say that they felt pressure from their peers on the court to dispose of cases quick-ly. They express that they are evaluated by how fast they are able to dispose of cases and “move the business” (Heumann 1978: 146). One of the inter-viewees describes how the chief administrative judge makes them fill out forms and count what they spend their time on and what cases they get rid of. What receives positive praise is when judges have worked out strategies for increasing the guilty plea rate (Heumann 1978: 145).

Coming from an economic and legal background, Landes and Posner ar-gue that the judges’ policy preferences can be understood as self-regarding.12 This builds on the assumption that the objective of the judges is to derive utility by imparting their own policy preferences on the community (Landes and Posner 1976: 272). This is because a judge “derives personal satisfaction by preferring one party to the lawsuit over another or one policy over anoth-er” (Landes and Posner 1975: 887). Other accounts inspired by economic theory suggest that judges have a preference for leisure and therefore strive to reduce their workload (Bainbridge and Gulati 2002; Posner 1993; Posner 1995: 123-126).

In theories about EU legal integration, it is also commonplace to view self-regarding preferences as the main determinant of the judges’ behavior. The national courts’ applications of EU law have been described as instru-mental strategies of empowerment (Alter 2001; Conant 2013: 408; Nyikos 2000). Take, for example, the judicial empowerment approach. According to this theory it is the judges’ calculations of how to increase their own powers vis-à-vis the domestic political actors that is key for understanding why na-tional courts are exhibiting a behavioral pattern that supports EU integration (Alter 2001; Burley and Mattli 1993; Mattli and Slaughter 1998: 190; Weiler 1994). Similarly, scholars who draw upon the sustained resistance approach have hypothesized that national judges increase their powers by being able to decide when to withhold cases from the CJEU and when to allow the CJEU access to domestic legal cases. By withholding cases from the CJEU, nation-al judges are able to maintain control over important policy outcomes (Golub 1996: 379).

The upshot of reviewing previous theory and research on judicial deci-sion-making is that judges are mainly portrayed as rational actors who strive

12 Similar models of judicial self-interest can be found in: Spiller, Pablo T & Gely, Rafael (1992) Congressional Control or Judicial Independence: The Determinants of US Supreme Court Labor-Relations Decisions, 1949-1988. The RAND journal of Economics, 463-492. Schwartz, Edward P, Spiller, Pablo T & Urbiztondo, Santiago (1994) A Positive Theory of Legislative Intent. Law and Contemporary Problems, 57, 51-74. For example, Schwartz, E. P., Spiller, P. T. & Urbiztondo, S. argue that the US Supreme Court is a self-interested, ideologically motivated institution (1994:57).

53

to maximize their self-regarding preferences. However, in both the academic disciplines of political science and economics, critique has been raised against the idea that self-interest is the only driving force of rational actors. It has been argued that apart from self-regarding preferences, actors may also have preferences relating to the well-being of others, known as other-regarding preferences (Fowler and Kam 2007; Mansbridge 1990). Yet an-other type of critique that has been launched against the overarching logic of consequentialism in the judicial politics literature goes even further. It ques-tions the very foundation of consequentialism: that human action is prefer-ence-oriented and driven by calculations of expected outcomes (Gillman 2001; Tamanaha 2009). The next sections discuss the content and the impli-cations of these different types of critique in greater detail.

Other-regarding preferences What motivates human behavior if not self-interest? One of the answers provided by previous research is other-regarding preferences (Fowler and Kam 2007; Mansbridge 1990). A common way to understand other-regarding preferences is through the concept of altruism (Kurz 1978). An actor who is motivated by other-regard cares not only about their own wel-fare but also about how their behavior benefits others (Fowler and Kam 2007: 813). It is therefore possible to distinguish between different types of consequentialist considerations, those that are self-regarding and those that are other-regarding. A theoretical approach that sees human behavior as driven by other-regarding motivations still expects judges to base their deci-sions on rational calculations of how desirable different outcomes are. But instead of self-regarding preferences, the judges consider the benefit for others, beyond themselves, when choosing how to act.

But who are ‘the others’? The answer to this question varies among re-searchers. Some expect actors to care about the welfare of those who one has a personal relationship with, such as colleagues and relatives (Fehr and Fischbacher 2002: 2). It has also been theorized that actors may care about different social or political groups in society (Fowler and Kam 2007: 813) as well as public service values (Stout 2001: 1610). Stout has further elaborated on what other-regarding preferences means in the judicial context and claims that the judiciary in contemporary democracies is founded on the expectation that judges will behave in an altruistic fashion (Stout 2001: 1609). It means that they serve the social goal of providing an impartial application of the law despite the lack of financial incentives (Stout 2001: 1605-1606).

The judicial politics literature offers a few examples of what types of oth-er-regarding preferences judges may have. In principle, policy preferences can be other-regarding. Baum takes a firm stand on the matter when claim-ing that the Supreme Court justices’ policy goals do not involve their self-interest in any direct way since the outcome of the judges’ decisions seldom

54

affects the justices themselves (Baum 1997: 61). Instead, the justices’ rulings primarily affect the parties to the case and possibly society at large (Baum 1997: 27, note 8). This builds on the definition that self-interested behavior means pursuing an action with the purpose of achieving only personal tangi-ble benefits (Baum 2006: 10-11; Holley 1999). However, as noted under the subheading of self-regarding preferences, others have theorized that the jus-tices’ self-interest is connected to policy goals (Landes and Posner 1975; Landes and Posner 1976; Schwartz et al. 1994: 57). For example, Landes and Posner argue that judges derive personal satisfaction by imposing one policy over another on the community (Landes and Posner 1975: 887).

How do these different theoretical understandings of policy preferences go together? The important distinguishing factor between policy preferences based on self-interest and policy preferences based on other-regard is what objective the judges have. In the case that the policy preferences are about benefitting oneself in any way and not any other member of society, those preferences can be understood as self-regarding. But if the judges decide cases based on their preferred policy and that is about benefitting others, then the behavior is other-regarding. In the present thesis, other-regarding preferences are understood in a broad sense. Judges may care about how the outcome of a case affects individual litigants, social groups and society at large. For example, judges may care about the rights of criminal defendants or minority groups (Baum 1997: 58).

Within the field of EU legal integration, the sustained resistance approach finds that other-regarding preferences are a source of judicial motivation. Golub puts forward the occurrence of political incentives as one type of mechanism which can explain the judges’ reluctance to request preliminary rulings (Golub 1996: 379). The claim is that national judges shield domestic policy from unfavorable CJEU rulings by withholding references from the CJEU. The judges’ choice to not make a reference, and thereby avoid rulings that may overturn domestic legislation, is believed to be the result of the judges’ considerations for how to protect domestic policies and national sovereignty (Golub 1996: 378). Those preferences are other-regarding in the sense that the objective is to uphold the member state government’s control over domestic legislation. In addition, it can be seen as an attempt to protect the coherence and stability of the domestic legal order. The CJEU’s deci-sions have been perceived as undermining legal certainty and causing dis-ruption within national legal systems (Alter 2001: 48; Dehousse 1998: 173).

Questioning consequentialism What can be gathered from this outline of the judicial politics literature is that the assumption about how national judges are guided by the logic of consequentialism in EU legal integration stems from theories about judicial decision-making in the US. This body of research, in turn, has mainly fo-

55

cused on the behavior of the Supreme Court justices. Overall, this outline raises two main issues that are relevant to discuss in relation to this thesis’ ambition to understand what considerations individual judges make in the preliminary ruling procedure.

First, how reasonable is it to expect that the assumption about the primacy of self-regarding preferences holds outside the US context? The US legal and political system has a number of distinct features that suggest that the US judges have an inherently different role than their European counterparts. These differences may lead to variations in the judges’ mode of reasoning. In particular, judges in the US have in general much stronger ties to the politi-cal arena in comparison with the European judges.

To begin with, there is no doubt that the judicial branch is an important political institution in the US (Dahl 1957; Maveety 2009; Rehder 2014). As Tocqueville wrote: “[t]here is hardly any political question in the United States that sooner or later does not turn into a judicial question” (Tocqueville 2010: 441). This has several important implications for judicial behavior in the US. First and foremost, the persons who are appointed by the President of the United States to serve as justices in the US Supreme Court are more or less expected to share the president’s overall ideological position (Epstein and Segal 2005; Segal and Cover 1989). In contrast, the political influence of most high courts in Europe is limited in comparison to the US Supreme Court (Dyevre 2010).

Moreover, US judges in general work in what can be described as a polit-ical environment. What characterizes political actors in a political institu-tional sphere is that they are expected to make considerations of how to max-imize self-regarding preferences, such as being re-elected (Crisp et al. 2004; Spiller and Gely 2008: 34-35). Many states in the US choose their judges through popular elections in which the judges’ political party affiliation is explicit (Brace and Boyea 2008; Gibson 2012). While most federal judges hold office until retirement, many other judges are appointed for only a shorter term. Those judges may face many opponents in the re-election cam-paign, making the battle for a new term expensive and heated (Baum 1997: 34; Hall 1995). In contrast, the judges in most European countries are re-cruited on the basis of merits and technical professional competences (Bell 2006: 17).

The US judges’ exposure to these political circumstances is expected to be mirrored in their mode of reasoning. From the literature, one finds that the US judges’ self-regarding preferences may include policy objectives, the wish to be (re-)elected and professional advancement. Taken together, the overall differences between the judiciaries in the US and in Europe raise the important question of whether judges working in the European context really can be expected to be guided by the same reasons as their US counterparts. The European judges’ lack of exposure to politics in the EU member states is expected to make self-regarding considerations less common among the

56

judges. This means that one of the key assumptions in the literature on EU legal integration is called into question: We have reason to doubt that na-tional judges are mainly motivated by self-regarding preferences in the pre-liminary ruling procedure.

The second issue is whether one agrees with the broader assumption that much of the previous research on judicial decision-making is based on; namely, that human behavior is driven by the logic of consequentialism. The consequentialist logic is appealing in its simplicity. We tend to motivate our decisions with references to our intentions and expected outcomes. But the most widespread critique against the theory is that its assumptions about human behavior are in fact too simplified (Cook and Levi 1990; March and Olsen 1989; Steinmo et al. 1992). Actors are expected to have information about how different choices will play out in relation to their preferences. But to what extent is it possible to make the types of calculated choices that the consequentialist logic presumes? In reality, we seldom know exactly what a specific choice will lead to. What is more, even if we did have perfect in-formation, human action may also be driven by other considerations. By only employing a consequentialist approach to studying judicial decision-making there is a risk that we misinterpret the judges’ behavior. For in-stance, Tamanaha (2009) argues that the field of judicial politics have failed to seriously try to understand the judges’ behavior by assuming that they are “enrobed politicians engaged in the single-minded pursuit of policy prefer-ences” (Tamanaha 2009: 117).

Researchers who are critical of consequentialism have claimed that it is instead considerations about appropriateness that primarily guide human action. What is appropriate is in turn defined by the surrounding institutional contexts, meaning the collection of rules and practices in a particular situa-tion (March and Olsen 1989; Steinmo et al. 1992: 8). In contrast to the con-sequential approach in rational choice institutionalism, rules and norms not only constrain but also shape human preferences and behavior (Steinmo et al. 1992). The concept of appropriateness stems from March and Olsen’s influential institutional perspective on politics (March and Olsen 1989; March and Olsen 1995; March and Olsen 2009). Similar theoretical outlooks feature in much previous research including studies of judicial decision-making (Clayton and Gillman 1999; Gillman 2001; Maveety 2009; Skowronek 1995; Smith 1988). In this literature, scholars who are critical of the dominant consequentialist perspective often examine judicial decisions-making through a historical institutionalist lens.

Given the thesis’ ambition to understand how individual judges are rea-soning in the preliminary ruling procedure, these points of critique against the dominance of a consequentialist and self-regarding approach to judicial behavior suggest that we need to explore alternative theoretical perspectives on human behavior. The theoretical outlooks in previous research, which

57

primarily draw upon the logic of appropriateness, are outlined in the next section.

The challenge from the logic of appropriateness Although the logic of consequentialism dominates the political science lit-erature on judicial decision-making, it is not universally prevailing. A num-ber of scholars instead subscribe to the understanding of judicial behavior as being guided by internalized institutional rules and norms (Brigham 1999; Burton 1992; Clayton and Gillman 1999; Davis 1999; Gillman 2001: 486; March and Olsen 1989: 161; Tamanaha 2009). The fundament is that human behavior is rule-bound (March and Olsen 1984; March and Olsen 1989: 23). It means that actors are not expected to make calculations of expected utility when making decisions. Instead, they are believed to follow internalized rules of what constitutes appropriate behavior (March and Olsen 2009: 3; Steinmo et al. 1992: 8).

How does the logic of appropriateness come to expression in previous re-search on the driving forces of judicial behavior? Although the theoretical outlooks differ somewhat, the key point is that sets of rules shape the judges’ behavior. March and Olsen’s understanding of the term ‘rule’ is very broad. Rules may include formal codified law but also routines, beliefs, conven-tions, established practices and so on (March & Olsen 1989:22). In the judi-cial decision-making literature, ‘rules’ about appropriate behavior have mostly been defined as the formal codified law and established legal princi-ples. The concept of ‘the law’ has been understood as an institution or a set of ideas that inform the choices judges make (Kahn 1999: 45). For judges, these institutions impose a distinct set of responsibilities (Brigham 1999: 21). The law is thus one of the institutions that define the values judges should adopt and what routines they should follow (Smith 1988: 95; Whittington 2000: 623). This strand of research finds that judges are acting non-legally, meaning that they do not follow legal principles, if they decide cases on the basis of “personal or political preferences, or interests, or felt identification of the parties” (Burton 1992: 45).

According to previous research, the most obvious types of rules influenc-ing the judges’ behavior are thus legal statues and binding legal doctrine such as precedents (Clayton and Gillman 1999: 15; Friedman 2006: 262). Those rules can analytically be described as formal codified rules. Turning instead to legal principles, these include the institutional norms that govern the courts’ operations. The institutional norms can be said to encompass a set of internal practices and traditions within the judiciary (Whittington 2000: 615, 621). Those types of rules can therefore be defined as informal profes-sional practices. It includes, for example, the judges’ shared understanding of important legal principles such as the concept of liberty (Clayton and Gillman 1999: 15). In the case of the US Supreme Court, informal profes-

58

sional practices include the norm of secrecy about deliberations prior to the announcement of decisions, the majority rule in deciding cases, and respect for seniority in assigning opinions (O´Brien 1999: 91).

The political science research on judicial behavior that was reviewed un-der the subheadings of self-regarding and other-regarding preferences is skeptical of this portrayal of judges. Judges are believed to do more than simply apply the law and their understanding of ‘good policy’ has an impact on their decisions (Baum 1997:57). However, scholars such as Gillman (2001), Clayton (1999) and Tamanaha (2009), among others, would not deny that many factors are likely to influence the judges’ behavior. They acknowledge that judges’ decisions may very well be influenced by policy preferences but they argue that the law also matters (Gillman 2001: 492). It means that rules of different kinds not only constrain human behavior, as have been suggested by the strategic model of judicial behavior, but also contribute to shaping it (Steinmo et al. 1992: 8; Whittington 2000). To un-derstand this argument, it is necessary to delve into the details about how rules, such as legal principles, guide the judges’ behavior.

Describing a behavior as driven by rules of appropriateness is to see deci-sion-making as a process of matching a particular situation to the demands of a person’s position. This mode of reasoning evolves around three ques-tions: “What kind of situation is this? What kind of person am I? What does a person such as I do in these types of situations?” (March and Olsen 1989: 23-24). To answer these questions, the actor make use of his or her experi-ence, knowledge and intuition (March and Olsen 2009: 4). It is important to note that a behavior guided by rules of appropriateness, such as legal institu-tional principles, does not mean that decision-making is mechanical (March and Olsen 1989: 22; March and Olsen 2009: 8). March and Olsen elaborate on why the matching between one’s identity, situation and action can be a complex process. Both one’s identity and the specific situation can be inter-preted in different ways depending on the actors’ prior knowledge and expe-rience. It may result in different understandings among actors of what consti-tutes the appropriate response to a given situation (March and Olsen 2009: 10; Orren and Skowronek 1994). Similarly, in the legal context, different legal principles can stand in conflict with each other, meaning that the judge has to use his or her judgment to decide which rule to apply and how (Dworkin 1978: 26; Gillman 2001: 487). Legal rules are thus expected to motivate and shape the judges’ decisions without fully determining the re-sults. This is why different judges may decide cases differently (Gillman 2001: 488).

How do actors find out what constitutes the appropriate behavior? The proposition is that they draw upon their experience and the criteria of simi-larity. For instance, actors that are unsure about what the appropriate action is in an unfamiliar situation may think about how they have acted historical-ly in comparable situations and what rules applied then (March and Olsen

59

2009: 9). Moreover, in the case of colliding rules, actors may try to distin-guish the higher rules from the lower ones. The resulting behavior is that judges may have to disobey one rule because they find it more appropriate to adhere to another one (March and Olsen 2009: 10).

Within the judicial sphere, one example would be the principle of hierar-chy among legal rules, which means that the constitutional law trumps any other piece of legislation in case of a conflict of norms. An example of con-flicting rules can be found in a study of how Polish judges have adapted to being EU law judges. Jaremba (2013) notes that most of the judges find the principle of EU law supremacy to constitute a binding obligation. However, the judges sometimes run into problems when trying to decide how to handle conflicts between EU law and national law. They consider the national con-stitution to be a paramount source of law that clashes with the formal su-premacy of EU law. The result of this is that they under some circumstances apply the constitution over EU law despite the supremacy of the latter (Jaremba 2013: 275-276)).

Furthermore, research on the US Supreme Court has found evidence sug-gesting that legal considerations do influence the justices’ decisions. Epstein and Kolbylka argue that doctrinal change in the death penalty jurisprudence was driven largely by the way litigants presented their legal arguments. The justices responded to arguments from litigants that were framed according to principles that the justices themselves had articulated in prior decisions. Therefore, the authors believed that the justices were influenced by the logic of their prior holdings rather than by their policy preferences or public opin-ion when deciding the cases (Epstein and Kobylka 1992).

In sum, what answer does previous research give to the question of how national judges behave in the preliminary ruling procedure? The dominant assumption in previous research is that judges reason according to the logic of expected consequences. In particular, they strive to maximize various self-regarding preferences, such as increasing their power and influence. Alternatively, the judges may be driven by other-regarding preferences; for example, to uphold the coherence of the domestic legal system. Both these understandings of the judges’ modes of reasoning, and what reasons they have for making choices, is challenged by accounts that draw upon the logic of appropriateness. This perspective claims that human behavior is guided by the actors’ conceptions of what constitutes appropriate behavior. These con-ceptions, in turn, are shaped by sets of formal codified rules and informal professional practices, organized into institutions.

This outline of judicial action has tried to chisel out contrasting and op-posing theoretical views on human behavior. However, the driving forces of human behavior cannot be reduced to only one mode of reasoning. In some situations and under some circumstances, actors may try to maximize their self-interest while in others they may follow informal practices about what constitutes appropriate behavior. A fruitful approach to judicial behavior is

60

therefore to view the behavioral logics and the modes of reasoning as com-plementary. The task is then to specify theoretically and investigate empiri-cally under what circumstances the different behavioral logics and modes of reasoning are more or less likely to guide the actors’ behavior (March and Olsen 2009: 19).

A theoretical model of judicial behavior in EU legal integration This chapter has elaborated on how previous research would answer the three research questions posed in the thesis. The result is a theoretical model consisting of three components. Two of these components provide theoreti-cal starting points for analyzing the national courts’ behavior at the aggre-gated level while the third component provides a framework for analyzing the behavior of individual judges.

Chapter 4 addresses the first research question: What types of choices are national courts making in the preliminary ruling procedure? This question is driven by a theoretical interest to understand what implications the courts’ aggregated choices may have for EU legal integration. Two distinct expecta-tions about the national courts’ behavioral pattern have been identified by reviewing the literature. According to the judicial empowerment approach, national courts are exhibiting a behavioral pattern that supports further EU legal integration. In contrast, the sustained resistance approach claims that national courts protect national sovereignty. These two approaches shape the overarching debate about what role national courts play in EU legal integra-tion.

Given these two theoretical approaches, what choices should we expect the courts to be making? Table 2.1 below illustrates the four possible combi-nations of how national courts can be expected to behave. The rows show a division between the EU legal cases that can be understood as having a high degree of political sensitivity and the ones that have a low degree of political sensitivity. The columns illustrate the two main types of opinions that na-tional courts express: opinions in support of the national law and opinions in support of the EU law.

61

Table 2.1. The national courts’ four behavioral patterns

Opinions supporting EU law

Opinions supporting national law

Cases with a high degree of political sensitivity

A

Judicial empowerment

B

Cases with a low degree of political sensitivity

C

D

Sustained resistance

The behavioral pattern which draws upon the judicial empowerment ap-proach can be found in position A. National courts are expected to refer cas-es with a high degree of political sensitivity and express opinions supporting EU law. This behavioral pattern suggests that national courts are supportive of EU legal integration. In contrast, the expectations derived from the sus-tained resistance approach can be found in position D. National courts are referring cases with a low degree of political sensitivity and expressing sup-port for the national law. This combination of choices is interpreted as a resistance to further EU legal integration.

This is the first time the national courts’ key choices are analyzed in con-junction. As can be seen in Table 2.1, bringing them together provides us with two additional behavioral patterns (B and C) that do not follow from either of the two main theoretical approaches in the literature. This thesis argues that these previously untheorized behavioral patterns can be under-stood by adopting the perspective that national courts are gatekeepers, placed in the intersection between the domestic political system and the EU. In par-ticular, the two behavioral patterns can be interpreted as an attempt by na-tional courts to balance two conflicting demands. What those demands are can be exemplified by briefly discussing behavioral pattern B. According to the EU Treaties, national courts are required to refer all cases in which the interpretation of EU law is unclear to the CJEU, including politically sensi-tive cases. Despite this formal rule, referring cases to the CJEU can be met with opposition in the member state. The member state governments do not want the CJEU to challenge the validity of national law in sensitive national policy areas (Alter 2001: 130; Blauberger 2014: 460). One way for national courts to get around this conflict is by providing the CJEU with politically sensitive cases just as required by the codified EU rules. However, at the same time, they defend the validity of the member state’s policy by express-ing opinion supporting national law. Making these two choices means, first of all, that national courts cannot be accused by the Commission and the CJEU of systematically withholding politically sensitive cases from the pre-

62

liminary ruling procedure. Secondly, by including opinions supporting na-tional law, national courts may also try to convince the CJEU to not overrule domestic policies. This means that national courts also show consideration for the wishes of the member state government, which wants to remain in control over national policies. National courts thereby balance the interests of the EU institutions and the member state government. The implications of the two previously untheorized behavioral patterns are further elaborated on in chapter 4.

In sum, the four combinations in Table 2.1 should be understood as styl-ized patterns of behavior. The question of interest is how common each of the four behavioral patterns is. Answering this question makes it possible to assess what implications the national courts’ aggregated behavior has for EU legal integration.

Chapter 5 addresses the second research question: What factors can ex-plain the variations in what type of choices national courts make in the pre-liminary ruling procedure? Here, the theoretical interest is to understand what makes some national courts are supportive of EU legal integration while others are instead defending national sovereignty.

Previous research on EU legal integration and judicial politics has pro-posed six hypotheses that all aim to explain variations in national court be-havior. One of those hypotheses is derived from the judicial empowerment approach while three of them have their theoretical basis in the sustained resistance approach. The remaining two hypotheses are derived from the Europeanization literature. Table 2.2 provides an overview of how the dif-ferent explanatory factors are expected to influence the national courts’ be-havioral patterns. How each of these hypothesized relationships matter for what types of cases national courts refer to the CJEU, and what types of opinions they express in their requests, is developed in chapter 5.

63

Table 2.2. Overview of the hypotheses Hypotheses

Judicial empowerment Inter-court competition

Lower national courts are more likely than higher courts to support EU legal integration

Sustained resistance Euroscepticism

A Eurosceptic public makes the national courts less likely to support EU legal integration

Majoritarian democracy Courts in a member state that has a tradition of ma-joritarian democracy are less likely than other courts to support EU legal integration

The political sensitivity of the case

Courts are less likely to support EU legal integration in cases with a high degree of political sensitivity

Europeanization Experience of the preliminary ruling procedure

Courts with more experience of the preliminary ruling procedure are more likely than other courts to support EU legal integration

Concrete judicial review

Courts in a member state that allows concrete judi-cial review are more likely than other courts to sup-port EU legal integration

In chapter 6, the third and final research question is addressed: What types of reasons are individual national judges guided by when making choices in the preliminary ruling procedure?

In the judicial politics literature, the dominant assumption is that the judges’ mode of reasoning is based on the logic of consequentialism. The judges’ decision-making is believed to be influenced by consideration for how to achieve their preferences. Most previous research within the field of judicial politics and EU legal integration assumes that those preferences are self-regarding. For instance, the judges may strive to get promoted or in-crease their powers. However, as has been pointed out in other works, the judges may also have other-regarding preferences such as caring for societal groups. This means that the overarching consequentialist perspective on human behavior can be divided into two modes of reasoning: one that is self-regarding and one that is other-regarding. Scholars who are critical of the logic of consequentialism draw upon accounts that build on the logic of ap-propriateness. They argue that institutions, such as the law, not only con-strain judicial behavior but also shape it. Institutions or rules of appropriate-ness may in turn be broken down into two analytical categories: formal codi-fied rules or informal professional practices.

64

In sum, the third component of the theoretical model consists of two overarching characterizations of what drives human behavior. Each of these two logics can be divided into two modes of reasoning depending on what types of substantial considerations inform the actors’ decisions. The result-ing four modes of reasoning are summarized in Table 2.3. and further devel-oped in chapter 6.

Table 2.3. The judges’ expected modes of reasoning in the preliminary ruling procedure The logic of consequentialism The logic of appropriateness

Self-regarding pref-erences

Other-regarding preferences

Formal codified rules

Norms and practices

Reputation, career, leisure, increase one’s powers (judi-cial empowerment) Promote one’s pre-ferred policies (sus-tained resistance)

To care for the well-being of other individuals, groups or institutions

Article 267 TFEU, Infor-mation note on the preliminary ruling procedure, various codified rules at the do-mestic level

Norms and practices about appro-priate con-duct and what it means to be a judge

The four characterizations of judicial decision-making should be considered as ideal types. In reality, the diversity and differences between individuals makes it impossible to credibly claim that all judges always behave accord-ing to one, and only one, logic of action. What is interesting, however, is the dominance of the logic of consequentialism and the mode of reasoning based on self-regarding considerations in theories of EU legal integration. Yet their propositions, stemming from the US judicial politics literature, have rarely been put to the test in the EU context. The question is whether these assump-tions can be substantiated by empirical findings or if any of the other modes of reasoning better captures how the national judges reason. In addition, analyzing the judges’ considerations not only provides us with an under-standing about what mode of reasoning they are guided by, it also reveals what substantive reasons judges have for their decisions in the preliminary ruling procedure.

65

3. Methodological approaches and research design

This chapter elaborates on the type of research design that is appropriate in order to provide answers to the three research questions. Since the overarch-ing aim of the thesis is to understand the national courts’ role as gatekeepers in EU legal integration, the empirical focus is both on the courts’ aggregated behavioral pattern as well as on the reasoning of individual judges. This means that answering the research questions requires the use of two different types of material and methods. A quantitative approach is most suitable for capturing the national courts’ general behavioral pattern and its implications for EU legal integration. The empirical material used for this type of analysis consists of an original data set of a total of 470 requests for preliminary rul-ings. In contrast, understanding how individual judges reason in the prelimi-nary ruling procedure requires a qualitative approach. The empirical analysis of the judges’ mode of reasoning builds upon transcripts from the author’s interviews with 20 Swedish judges.

Assessing the national courts’ behavioral pattern: A quantitative approach A quantitative approach is appropriate when the aim of the research is to capture general patterns of behavior (Coppedge 1999: 468). In this thesis, the behavior of interest is what choices national courts make in the preliminary ruling procedure. Specifically, it is about what type of cases they refer to the CJEU and what type of opinions (if any) that they include in those requests for preliminary rulings. Thus, the units of analysis are the legal cases in which national courts have requested a preliminary ruling from the CJEU.

66

In order to capture the national courts’ overall behavioral pattern the se-lection of cases consists of a simple random sample of 47013 cases drawn from all requests for preliminary rulings (5590 cases in total) from the years 1992 to 2012.14 Most previous research has focused on national court behav-ior in the early days of the European integration process (Alter 2001; Nyikos 2006; Sweet and Brunell 1998b). However, less is known about how nation-al courts have acted in the wake of the Maastricht Treaty when the European Communities became the European Union in 1992. The reason for studying what choices national courts made between 1992 and 2012 is to capture the courts’ behavior as gatekeepers in a deeper and wider EU. Figure 3.1 below show the share of cases included in the data set by member state and the share of cases in the population by member state.

13 A total of 470 cases was deemed sufficient since the intention was to carry out statistical analyses. Previous research has shown that national court opinions are likely to occur in around 40 percent of the cases (Nyikos 2006). Thus, it was expected that around 188 out of the 470 cases would include opinions, which is sufficient to be able to perform logistic regres-sion analysis. A rule of thumb is that at least 100 observations are needed (Long 1997: 54) Long, J. Scott (1997) Regression Models for Categorical and Limited Dependant Variables, Thousand Oaks, Calif: SAGE. 14 Through the official database on European Union Law, EUR-LEX, it is possible to identify the number of requests for preliminary rulings that has been decided by the CJEU each year. Each request (or case) has a unique number, for example the case: “C-12/99” is case number 12 from the year 1999. Based on this information about the population of cases, a simple random sample was generated using statistical software.

67

Figure 3.1.The percentage of cases from each member state in the data set and in the population between 1992 and 2012. Source: Court of Justice of the European Union, (2017) and author’s own data set. Number of observation in the population: 5509, number of observations in data set: 470.15

As can be seen in Figure 3.1, courts from 23 out of 27 member states are represented in the data set. The five member states not included in the sam-ple are Croatia16, Cyprus, Malta, Slovakia and Slovenia. This is not surpris-ing given the low number of cases referred from courts in those countries to the CJEU.17 The proportion of cases from different member state courts in the sample is fairly similar to the proportion of cases from different member

15A table showing the frequency of cases from courts in each member state can be found in Appendix A. 16 Croatia became a member in 2013 and is therefore not included in the study 17 During the period in question Cyprus and Malta have referred two cases each, Slovakia has referred 20 cases and Slovenia has referred four cases (Court of Justice of the European Union 2017).

0 5 10 15 20 25

AustriaBelgiumBulgaria

Czech RepublicDenmark

EstoniaFinlandFrance

GermanyGreece

HungaryIreland

ItalyLatvia

LithuaniaLuxembourgNetherlands

PolandPortugalRomania

SpainSweden

UK

Percent

Percent in population

Percent in data set

68

state courts in the population.18 For example, German courts are responsible for the single highest share of referrals in the population of cases (22.1%) and in the sample (21.5%). In contrast, the smallest shares of cases, around one percent in the sample and in the population, come from courts in the newest member states. The data set is overall considered to be representative with regard to the cases referred from national courts to the CJEU between 1992 and 2012.

Information about what type of choices national courts make has been ob-tained from two documents: Judgment of the Court and Opinion of the Ad-vocate General. The document Judgment of the Court is the official ruling from the CJEU and it begins with a description of the legal issue at hand followed by the questions posed by the national court and a presentation of the observations made by the Commission and the member states.19 The document ends with the CJEU’s legal analysis of the issues and its answers to the national court’s questions. The documents range from 3 to 50 pages depending on the character of the legal issue. The document Opinion of the Advocate General follows a similar outline but discusses in greater detail the legal questions present in the case. This document contains a detailed de-scription of the character of the legal cases that national courts have referred to the CJEU.20 Both documents can be accessed from the official website of European Union law (EUR-LEX21) and are available in all official EU lan-guages. The documents have been used frequently in previous research on national court behavior (Carrubba et al. 2008; Carrubba and Gabel 2015; Leijon and Karlsson 2013; Nyikos 2006; Sweet and Brunell 1998b).22

Identifying national court opinions How has information about what opinions national courts express been iden-tified in these documents? Previous research (Leijon and Karlsson 2013; 18 Cases from some countries are somewhat over- or underrepresented in the sample. Refer-rals from courts in Austria, Belgium, the Netherlands and the UK are slightly overrepresented in the sample. Conversely, cases sent from courts in France and Italy to the CJEU are some-what underrepresented in the data set. For example, referrals from French courts make up 7.9 percent of the cases in the population but only 6.4 percent in the sample. 19 For research on the positions of the member states and the Commission before the CJEU see e.g., Carrubba et. al 2008; Larsson & Naurin 2016. 20The purpose of not only using the information in the document Judgment of the Court but also information found in the Opinion of the Advocate General is that the latter document gives a more detail picture of the case and the questions asked which has been useful in some of the more intricate legal cases. 21 EUR-LEX is the official website of European Union Law and other public documents of the EU. http://eur-lex.europa.eu 22 Ideally, one would like to have had access to the dossier, that is, the actual documents sent from the national court to the CJEU. However, since this file is not available to the public most research relies on the official documents provided by the CJEU. Looking at both the Judgment of the Court and the Opinion of the Advocate General provides the most encom-passing view of what choices national courts have made.

69

Nyikos 2006) has found that the opinions are usually placed in the fact sec-tion of the documents together with the questions the national court wants the CJEU to answer. In, for example, the document Judgment of the Court and under the heading “The proceedings before the national court” one finds the national court’s description of the legal issues, the questions and, some-times, an opinion about how the national court thinks that the case should be resolved. It is a fairly straightforward task to identify an opinion coming from the national court in this document because such opinions are referred to as being expressed by the national court in question. For example:

However the national court, Finalarte Sociedade de Construção Civil Ld.a and Portugaia Construções Ld.a consider that Article 48 of the Treaty also applies to the cases in the main proceedings [emphasis added] in that the chances of workers being taken on and posted abroad are reduced to the ex-tent that an employer may be deterred, as a result of the extension of the paid leave scheme, from exercising its freedom to provide services by pursuing activities in the Federal Republic of Germany. (Judgment of the Court of Jus-tice 25 October 2001, Urlaubs- und Lohnausgleichskasse der Bauwirtschaft v.Turiprata Construções Civil Ldª, C-54/98, EU:C:2001:564, paragraph 21).

Apart from reading this section in the search for opinions, the search tool in the PDF-reader has been used in order to make sure that no opinions that might have been placed in another section of the document have been left out of the analysis. The search terms include, among others: ‘according to the national court,’ ‘in the order of the reference,’ and ‘in the view of the national court’.23

What about the types of opinions? Previous research suggests that the substantive content of the national courts’ opinions can be divided into two main categories: opinions supporting national law or opinions supporting EU law (Leijon and Karlsson 2013). Supporting national law is here defined as expressing support for national legislative acts, regulations, practices, and decisions made by national agencies. In contrast, an opinion supporting EU law is defined as any instance in which national courts side with the EU le-gal acts, practices, regulations, or decisions.

An example of an opinion supporting national law can be found in the case DaimlerChrysler AG v Land Baden-Württemberg, referred from the German Federal Administrative Court (Bundesverwaltungsgericht). The legal issue in the case concerned the legality of a decree of the government and the Minister for the Environment and Transport of Land Baden-Württemberg. This decree made it compulsory for DaimlerChrysler to de-posit its production waste in Hamburg. The company claimed that this would be more costly than its current arrangement, to ship the waste to Belgium. In 23 The other search terms include the full name of the national court requesting the prelimi-nary ruling, for example “Bundesverwaltungsgericht” and “Rechtbank van koophandel te Gent”

70

DaimlerChrysler’s view the national law in question was a quantitative re-striction on exports and thus prohibited under EU law. In the request for a preliminary ruling the national court included the following opinion:

The Bundesverwaltungsgericht considers that the prohibition on exporting hazardous waste for disposal imposed by the contested decree must be con-sidered to be an imperative requirement of environmental protection, within the meaning of the Court's case-law. It concludes that the prohibition is not contrary to Article 28 EC (now Article 34 TFEU) [emphasis added] (Opinion of Mr. Advocate General Léger 20 September 2001, DaimlerChrysler AG v Land Baden-Württemberg, C-324/99, EU:C:2001:459, paragraph 36).

As becomes clear in the passage above, the national court not only makes an interpretation of the CJEU’s case law but it also claims that the national law (the prohibition on exporting hazardous waste) is in fact not contrary to EU law (Article 28 EC). This opinion is interpreted as an expression of support for the existing national legislation.

In case C-387/93 we instead find a national court, the Pretura Circondar-iale di Genova (District Magistrate’s Court, Genoa), which expresses support for EU law. The case concerned the legality of Italian tobacco laws. In the request for a preliminary ruling sent to the CJEU, the national court ex-pressed the following view:

The national court considers that, in view of the geographical spread of to-bacco outlets in Italy, their opening hours and functional deficiencies, such as inadequate supplies in some outlets of cigarette brands less frequently re-quested by consumers or occasional disruptions to supply because of strikes, the system of authorized tobacco outlets gives rise to restrictions on trade contrary to Article 30 of the Treaty [emphasis added] (Judgment of the Court of Justice 14 December 1995, Criminal proceedings against Giorgio Domin-go Banchero, C-387/93, EU:C:1995:439, paragraph 38).

Here, the Pretura Circondariale di Genova points out that it considers the Italian system with authorized tobacco outlets (a national monopoly) to be contrary to EU law (Article 30) since it imposes restrictions on trade.

Previous research has also identified a third category of opinions, which are not directly related to EU legal integration. An example of this type of opinion can be found in the case Medion AG v Thomson multimedia Sales Germany & Austria GmbH (C-120/04):

71

Applying the above principles to the present case, the referring court consid-ers that [emphasis added] there is no likelihood of confusion because the manufacturer designation ‘THOMSON’ characterises the overall impression conveyed by the contested designation ‘THOMSON LIFE’; the ‘LIFE’ com-ponent therefore does not characterise it on its own. (Judgment of the Court of Justice 6 October 2005, Medion AG v Thomson multimedia Sales Germa-ny & Austria GmbH., C-120/04, EU:C:2005:594, paragraph 15).

In this case the referring national court, Oberlandesgericht, considers that there is no risk of confusion between the two trademarks. It is clear that the referring court takes a stand on the legal matter in this case but this opinion cannot be understood as supporting either national law or EU law. Instead, the national court’s opinion falls into the category of “other position.” While this type of national court opinion is important for understanding the wider judicial dialogue between national courts and the CJEU (Jacobs 2003: 548; Rosas 2007: 126) it is not the focus of the present thesis. Instead, the empiri-cal analyses mainly center on the opinions that are directly related to EU legal integration: opinions in support of national law and opinions in support of EU law.

One minor caveat is worth making in relation to the procedure for identi-fying the national courts’ opinions. According to Nyikos (2006), it might be problematic to analyze cases sent from the national courts to the CJEU after 1995. The potential problem is that a section in the document Judgment of the Court, which used to contain national court opinions, has been rear-ranged. The claim is that even though the opinions of the national court can be found in other parts of the judgment, they might be incomplete (Nyikos 2006: 538). The accuracy of this statement has been evaluated in a study of 30 cases referred from Swedish courts to the CJEU between 1995 and 2009. It found that, when comparing the original document from the Swedish courts (which include the original statements of the national court) with the documents Judgment of the Court and Opinion of the Advocate General no opinions were missing in the EU documents (Leijon and Karlsson 2013).24 Hence, there is no empirical support for the claim that going beyond 1995 would in any serious way affect the reliability of the results.

24 In addition, the investigation of both the documents Judgment of the Court and Opinion of the Advocate General is believed to have further minimized the risk that opinions have been left out. All information which pertains to the question regarding the legal case should be included in those documents. Since the national courts’ opinions fall under this category, it is unlikely that any opinions have been left out. If a national court made a statement that has nothing to do with the legal questions posed, those opinions might have been excluded. How-ever, that is not a problem since those opinions are not the primary interest of this thesis. For an exploration of the Swedish high courts’ dialogue with the CJEU, see: Wallerman, Anna (2018) 'Referring Court Influence in the Preliminary Ruling Procedure: The Swedish Example'. In: Derlén, Mattias & Lindholm, Johan (eds.) The Court of Justice of the European Union : Multidisciplinary Perspectives. Oxford, UK Hart Publishing.

72

Identifying the political sensitivity of a request for a preliminary ruling As was discussed in chapters 1 and 2, previous research has argued that what types of cases national courts refer to the CJEU has important implications for the pace of EU legal integration. If the bulk of the preliminary references are concerned with interpreting technical EU legal matters then the CJEU will be prevented from expanding the reach of EU law (Alter 2000: 501; Alter 2009: 99; Bebr 1983: 456-457). On the other hand, if the requests regu-larly bring up questions regarding the validity of national law the CJEU is given ample opportunities to push for further EU legal integration (Alter 2001: 36). What is key is therefore the examination of what types of cases national courts actually refer to the CJEU. But how are we to distinguish between cases that are of minor political im-portance, such as technical cases, and politically sensitive cases in which national policies are at stake? In order to develop a theoretically grounded method for identifying cases with a high degree of political sensitivity it must first be established what political sensitivity is.

According to previous research, the question of a case’s degree of politi-cal sensitivity comes down to what values are at stake for the member state in the preliminary ruling procedure. This includes, for example, the member states’ control of policy outcomes (Golub 1996: 381). Blauberger (2014) describes the member state governments as typical supporters of the regula-tory status quo, that is, they prefer the existing domestic rules25 (Blauberger 2014: 460). For this reason they want to avoid CJEU rulings that demand domestic legislative reform or involve different kinds of financial risks, such as contractual penalties in public procurement or state liability for infringe-ment of EU law (Blauberger 2014: 461-462). In other words, the member state governments want to avoid rulings that declare domestic legislation to be incompatible with EU law. To be forced to initiate legislative reform can be considered to be a type of ‘political cost’ for the member state govern-ment. In addition, the member state governments want to avoid paying dif-ferent types of fines, which may be described as a type of ‘financial cost.’ Alter argues along the same line when theorizing that the member states typically want to avoid the CJEU’s decisions that “could upset public poli-cies or create a significant material impact (be it political or financial)” (Alter 1998a: 130).

How are political and financial costs defined? According to previous re-search, political costs primarily occur when national governments run the

25This is of course an assumption that can be questioned. Researchers have claimed that na-tional governments sometimes use the EU (including the CJEU) as a scapegoat for unpopular decisions (e.g. Novak 2013). Even though this might happen, it is reasonable to assume that the member state governments for the most part want to keep control over national policies and not be forced to change them by the CJEU.

73

risk of having to change existing national law. However, political costs may also arise when a CJEU ruling goes against public opinion and forces the government to impose and justify an unpopular decision. For example, the export of live calves from the UK to other member states with less rigid an-imal protection controls became, according to the request for a preliminary ruling, “a topic of considerable public concern in the United Kingdom”.26

Thus, public concern or widespread dissatisfaction among the electorate may also constitute a political cost. The financial costs, on the other hand, are to be understood as purely negative economic consequences brought upon the member state by a CJEU ruling. For example, economic costs may arise in a case that concerns the question of whether national social security arrangements should apply to a wider circle of people than what is currently the case. Other examples are when a CJEU ruling threatens the competive-ness of an important national industry. In such cases a preliminary ruling may lead to cut backs and fewer jobs, which in turn lead to higher unem-ployment.

It is worth emphasizing that the inquiry centers on what types of costs a preliminary ruling may impose on the member state. That is, it makes no claim to capture the actual economic or political effects of the ruling deliv-ered by the CJEU. Instead, it is about characterizing what types of cases national courts decide to refer to the CJEU. For instance, are the majority of the cases likely to bring about high political and economic costs for the member state?

Building on these definitions of politically sensitive issues makes it pos-sible to construct an analytical model for assessing the cases’ degree of polit-ical sensitivity. In sum, previous research claims that costs mainly arise when national law has to be changed and those costs can be either political or financial. The first key point for determining the degree of political sensi-tivity of a case is therefore to assess the nature of the EU legal dispute in the request at hand. Does it concern a conflict between national law and EU law or not? If it does, the CJEU is given the opportunity to invalidate national law, which in turn is likely to create different types of costs for the member state. For example, the member state government may have to give up its current policy and initiate a comprehensive legislative revision, thereby suf-fering both political and economic costs.

A conflict between national law and EU law is therefore necessary for a case to end up in the category of cases that have a high degree of political sensitivity. But not all of these conflicts are the same. Previous research points to the importance of taking into account the salience of the case. The potential political and economic costs for the member state are believed to

26 Judgment of the Court of 19 March 1998, The Queen v Minister of Agriculture, Fisheries and Food, ex parte Compassion in World Farming Ltd. C-1/96. EU:C:1998:113, paragraph 25) in the UK.

74

be higher in the salient cases because those cases challenge core institutions of the member state. For example, Wind argues that particularly controver-sial EU legal cases “may involve the extension of social rights or forcing a change in a sensitive national policy area” (Wind 2010: 1052). Nyikos writes that the CJEU’s interference in policy areas in which the EU only has limited competencies is seen as controversial by the member states (Nyikos 2006). Wind (2010) and Martinsen (2011) refer to social policies, health care, im-migration, and general welfare policies as traditionally contested policy are-as over which member states want to maintain control over (Martinsen 2011; Wind 2010). In other words, salient issues touch upon core national features and competencies (Bergvall 2007; Börzel 2003) as well as issues of wide-spread public concern such as the national organization of welfare, labor market regulations, or the tax system. This means that it is possible to distin-guish between politically sensitive cases, which are not salient, and highly politically sensitive cases, which are salient.

Thus far, the characteristics of cases that have a high degree of political sensitivity have been discussed. But what about those cases that do not con-cern a conflict between national law and EU law? If the nature of the legal dispute instead relates to different interpretations of EU law, without any involvement of national law, the case is less likely to give rise to substantial political or economic costs. The reason is that there is no national policy at stake meaning that regardless of the outcome, the member state does not run the risk of having to initiate legislative reform. This group of cases can therefore be considered to have a low degree of political sensitivity. When the type of legal dispute concerns different interpretations of EU law, the parties are often private actors.

However, it can also be the case that a government agency has made a de-cision based on EU law and an individual or a private company contests the agency’s interpretation of the EU provision. In the event that the CJEU would declare the decision made by the national agency invalid, the instruc-tions to the national agencies for how to apply the EU law may need to be changed. For instance, if the CJEU finds that the member state has failed to implement EU rules correctly, the state may face both political and econom-ic adjustment costs. This means that when the member state is a party to the case it is likely to increases the potential political and economic costs com-pared to when only private parties are involved. In sum, when the nature of the legal dispute concerns different interpretations of EU law, it is important to distinguish between cases in which the member state, in any capacity,27 is a party and those cases which only have private parties. The former are re-

27 It is true that the member states always formally become a party to the case when it is re-ferred to the CJEU and that they are allowed to submit observations to the Court. However, this thesis delimits the definition of “party to the case” to the original parties in the case brought to the national court.

75

ferred to as politically non-sensitive cases and the latter as somewhat politi-cally sensitive cases.

To summarize the procedure for determining the relative political sensi-tivity of a case, the first step is to decide whether the nature of the legal dis-pute concerns an interpretation of EU law or a conflict between national law and EU law. For those cases that concern a conflict between national law and EU law, it is possible to further differentiate between cases that concern a salient issue (for example, core national features like the organization of welfare) and less salient cases. The former are considered to be highly politi-cally sensitive cases while the latter are categorized as politically sensitive cases. In the other main group of cases in which the nature of the legal dis-pute instead concerns interpretations of EU law, it is the parties to the case that are of importance. If both parties are private the case is considered to be a politically non-sensitive case. Alternatively, if the member state (in any capacity) is a party to the case, it is defined as a somewhat politically sensi-tive case. A summary of the typical features of each type of case is described in Table 3.1 and the procedure for classifying the case can be found in Fig-ure 3.2.

Table 3.1. The definition of a case’s degree of political sensitivity Type of case Definition

A politically non-sensitive case The nature of the legal dispute con-cern the interpretation of EU law and the parties to the case are private actors.

A somewhat politically sensitive case The nature of the legal dispute con-cern the interpretation of EU law and the member state (in any capacity) is a party.

A politically sensitive case The nature of the legal dispute con-cern a conflict between EU law and national law and the case is non-salient.

A highly politically sensitive case The nature of the legal dispute con-cern a conflict between EU law and national law, and the legal case is salient.

76

Figure 3.2. The figure displays the process of classifying each request for a prelimi-nary ruling according to its degree of political sensitivity.

Empirical examples of the four different types of cases In a typical politically non-sensitive case the nature of the legal dispute is a conflict over the correct interpretation of EU law and the case involves only private parties. An illustrative example is the dispute between two compa-nies, Robeco Groep NV and Robelco NV, concerning the similarity of their company names.28 Robeco brought an action against Robleco before the national court claiming an order restraining Robelco from making use of the name Robelco or of any sign similar to Robeco as a trade name or company name. The legal question that arose in this dispute was how the word ’sign’ as used in Article 5(5) of the Trademark Directive should be interpreted: Does it refer both to identical and similar signs? In a case like this and in others belonging to the category of politically non-sensitive cases, the politi-cal cost for the member state, meaning the risk of a ruling invalidating exist-ing legislation, is negligible. The same goes for the potential economic cost. It is hard to imagine any outcome of this and similar cases in which the member state would suffer any substantial economic costs.

In a typical case in the category of somewhat politically sensitive cases, the nature of the legal dispute usually revolves around two conflicting inter-pretations of EU law, for example, a private party contesting a decision from a public agency. Hence, the case involves some branch of the member state as a party to the legal dispute. It might be as a public agency, a municipality, 28 Judgment of the Court (Sixth Chamber) of 21 November 2002. Robelco NV v Robeco Groep NV. C-23/01. EU:C:2002:706

Request for a preliminary ruling

Interpretation of EU law

Private partiesA politically non-

sensitive case

Member state partyA somewhat

politically sensitive case

Conflict between EU law and national law

Non-salientA politically

sensitive case

SalientA highly politically

sensitive case

77

or the state itself. An example of such a case29 is the dispute between the Danish Tax Authorities and the shoe company Ecco concerning the tariff classification for a sandal. Ecco argued that the sandal should have been classified under CN subheading 6403 as “footwear with (…9 uppers of leather’ instead of under CN subheading 6404 as “footwear with (…) uppers of textile materials’.” The legal question thus concerned the interpretation of headings 6403 and 6404 of the Combined Nomenclature set out in Annex I to Council Regulation. In a case like this and in the other cases in this cate-gory, the political cost for the member state, i.e. the risk of a ruling contrary to the political preferences of the national government, is minor but still exists. However, it is not any existing national legislation that runs the risk of being ruled incompatible with EU law, but rather a decision made by the government agency. Even though a case like this may not require any legis-lative reform, the instructions to the agency for how to apply the EU law may need to be changed, causing small political costs. As for the potential economic cost, in terms of tax revenues, these are arguably minor.

In a typical politically sensitive case the nature of the legal dispute is a conflict of laws, that is, it raises the question of whether or not the national law is compatible with the EU law. An example of a politically sensitive case is the dispute between a British citizen, Cyril Richardson, and the Sec-retary of State for health in United Kingdom concerning equal treatment of men and women.30 Mr. Richardson claimed that the national regulation, ac-cording to which women are exempt from prescription charges in connection with the provision of free medicine when they attain the age of 60, whereas men are only exempt from prescription charges when they attain the age of 65, was incompatible with EU law. This required an interpretation of a Council Directive on the progressive implementation of the principle of equal treatment for men and women in relation to national regulations. In a case like this and in the other cases belonging in the politically sensitive category, political cost for the member state clearly exists. The member state may be required to initiate legislative reforms. Also, if men would be exempt from prescription charges at the same age as women, it would mean in-creased economic costs for the member state.

Fourth, in a typical highly politically sensitive case the nature of the le-gal dispute is a conflict between national law and EU law. However, the conflict is not just about any national legislation potentially being incompat-ible with EU law. Instead, the issue being challenged is one of the member state’s core national features. This includes the organization of welfare, labor market regulations and taxation at the systemic level. The potential impact

29 Judgment of the Court (Fifth Chamber) of 22 May 2008. Skatteministeriet v Ecco Sko A/S. C-165/07. EU:C:2008:302 30 Judgment of the Court (Sixth Chamber) of 19 October 1995. The Queen v Secretary of State for Health, ex parte Cyril Richardson. C-137/94. EU:C:1995:342

78

makes it a salient case. An example of a case in this category is the dispute between Mr. De Vriendt and the National Pensions Office in Belgium con-cerning different calculations of pensions.31 In Belgium at the time pensions were calculated differently for men and women on the basis of an assumed employment record of 45 or 40 years, respectively. In the case of a shorter period of employment, the pension was calculated in fractions of fortieths (women) or forty-fifths (men), making it less beneficial for men to retire at a younger age than for women. Mr. De Vriendt claimed that the pension should, on grounds of equality, be calculated the same way for both men and women. The question that needed to be answered was whether the difference in the method of calculation was compatible with Article 4 of the Directive of equal treatment of men and women. Here, and in the other cases belong-ing to the category of highly politically sensitive cases, the political cost for the member state is evident. In the case of a ruling in favor of the plaintiff, the member state may be required to initiate legislative reforms in one of its core national policies, the pension system. With regard to the potential eco-nomic cost, making it more beneficial for men to retire at an earlier age would mean increased economic costs for the member state.

To ensure the reliability of the analysis, a researcher with judicial training performed an inter-coder reliability test on parts of the material.32 The result showed a high degree of inter-coder agreement. In 78 percent of the cases there was an agreement between the author’s classification and the one made by the legal researcher. The difference in the remaining 22 percent mainly related to how to determine whether a case is ‘politically sensitive’ or ‘high-ly politically sensitive.’ The author tended to be somewhat less likely to identify a case as being salient (highly politically sensitive) than the legal researcher. But this is not a major source of concern. In most of the analysis carried out in chapter 4 and 5, the purpose is to capture the main divide be-tween cases in which there is a conflict between national legislation and EU law and those where there is no such conflict. This means that the politically sensitive cases are merged with the highly sensitive cases into the category cases with a high degree of political sensitivity. Similarly, the politically non-sensitive cases are merged with the somewhat politically sensitive cases into the category cases with a low degree of political sensitivity.

31 Judgment of the Court (Sixth Chamber) of 30 April 1998 , August De Vriendt v Rijksdienst voor Pensioenen, C-377/96, EU:C:1998:183 32 The author is very grateful for the help provided by Lars Karlander, doctoral candidate in the Faculty of Law, Uppsala University.

79

Understanding the behavior of individual judges: A qualitative approach In order to shed light on the role of national courts as gatekeepers in EU legal integration it is necessary to take into account the behavior of individu-al judges. This is the focus of the third research question: What type of rea-sons are individual national judges guided by when making choices in the preliminary ruling procedure? The question is part of the broader debate about what drives the actions of judicial actors. Much previous research on EU legal integration rests on the assumption that national courts care about becoming empowered (Alter 2001; Burley and Mattli 1993; Tridimas and Tridimas 2004). This, in turn, follows from the dominant theoretical outlook of rational choice in the US judicial politics literature, which assumes that considerations about expected outcomes drive judicial behavior. However, few studies have tried to corroborate these claims empirically in the Europe-an context (Rehder 2014: 394). This thesis addresses this lacuna in the litera-ture by trying to capture the judges’ beliefs, thoughts and views about their different choices in the preliminary ruling procedure.

When the aim is to understand how actors perceive their world and their role in it, semi-structured interviews are a commonly used and well-suited technique (Marsh and Stoker 2010: 199). The researcher develops an inter-view guide based on the research question, usually including several differ-ent themes. This provides an overall structure for the interview but it is at the same time flexible enough to handle unexpected answers (Kvale and Brinkmann 2009: 130). The semi-structured interview type thereby makes use of open-ended questions aimed at exploring how the interviewees expe-rience their world and what motivations they have for making different deci-sions (King and Horrocks 2010; Kvale and Brinkmann 2009: 1).

Given the research question and the proposed method, semi-structured in-terviews, what is a suitable research design in terms of selecting respond-ents? In this type of qualitative study it is not possible to employ the same logic as the one used in the quantitative study, that is, to interview a random sample of European judges. Instead, a case study is the only feasible option. The question is then what case to select. This decision has in part been guid-ed by pragmatic concerns relating to how to get access to the relevant actors and, by extension, to be able to access the information necessary for answer-ing the research question.

With regard to the purpose of the interviews, national judges as a group can be understood as a type of ‘elite’. Elite respondents are experts on the topic being discussed during the interview (Burnham 2008: 231; Kvale and Brinkmann 2009: 147; Leech 2002b: 663). When it comes to elite inter-viewing, getting access to the relevant respondents may prove difficult (Goldstein 2002; Kvale and Brinkmann 2009: 147). It all comes down to whether the researcher is able to get in contact with the respondent and con-

80

vince him or her to participate in the study. In general it is more likely that a researcher who is knowledgeable about the topic of concern will be able to secure an interview compared to a researcher who lacks an in-depth under-standing of the context (Kvale and Brinkmann 2009: 147). This circum-stance points to the importance of being familiar with the context in which the respondent is situated. In this case, this means being knowledgeable about the member state’s political and legal context as well as having con-tacts within the member state that can vouch for the integrity of the research-er.

To be able to actually talk to the judges is a prerequisite for answering the research question. Therefore, there are two practical but important reasons why the Swedish judiciary is considered to be a good choice for a case study. First of all, Swedish judges are accessible. In Sweden it is in fact relatively easy to get access to various elites, including legislators, civil servants, and judges. Since the author of this thesis is based in Sweden and has extensive knowledge about the country’s political context as well as contacts within the judicial system, interviewing Swedish judges was deemed the most fea-sible alternative. In addition, the choice of studying Swedish judges has meant that it has been possible to meet the judges in person when conducting the interviews. This is believed to have contributed to establishing an atmos-phere of mutual trust during the interview (Burnham 2008:242), which in turn is expected to have made the interviewees more likely to give informa-tive answer to the questions asked. The second reason is that since the author speaks Swedish it has been possible to conduct the interviews in the judges’ mother tongue without an interpreter present. This circumstance is likely to contribute to uncensored and comprehensive answers from the respondents (Fujii 2013: 147).

There is only limited previous research on the behavior of individual Eu-ropean judges (Hönnige 2011: 350; Rehder 2014: 386) including the behav-ior of the Swedish judiciary. This means that studying how Swedish judges make decisions in the preliminary ruling procedure will make an empirical contribution to the research field. Moreover, although valid pragmatic issues such as access to respondents guided the case selection (Gerring 2007: 149-150), the choice of studying the Swedish judiciary can also be discussed in relation to how it informs our understanding of how European judges in gen-eral make decisions in the preliminary ruling procedure.

Although some researchers argue that interviews should not be used with the purpose of identifying generalized patterns of behavior (King and Horrocks 2010: 26) others think that making such generalizable claims is one of the basic goals of research (Goldstein 2002: 669). This thesis sides with the latter perspective and argues that it is indeed important to try to discern common behavioral patterns. Therefore, it is necessary to elaborate on how the case of the Swedish judiciary is to be understood in relation to other European judiciaries. This means that two questions must be answered.

81

To begin with, whether or to what extent can we expect that the Swedish judges’ views on the preliminary ruling procedure are representative of other judiciaries? This leads to the follow-up question: representative with regard to what?

While there is extensive research on what accounts for the variation in the national courts’ aggregated behavioral patterns (for an overview, see, e.g., Alter 2000; Conant 2007: 54) less is known about how individual judges behave (Hönnige 2011: 350; Rehder 2014: 386). It has been claimed that differences in legal cultures (Bell 2006) and differences in democratic tradi-tions (Wind 2010; Wind et al. 2009) may account for variations in judicial behavior. However, these accounts provide little insight as to how different legal cultures matter for how individual judges are reasoning in the prelimi-nary ruling procedure and what substantive reasons they have for making different decisions. In fact, there are no strong theoretical arguments or em-pirical data suggesting the existence of systematic cross-national variation when it comes to what type of considerations individual judges make in the preliminary ruling procedure. Hence, there is no reason to believe that Swe-den is a deviant case in the context of understanding how judges from EU member states reason when making decisions in the preliminary ruling pro-cedure. Studying the Swedish judiciary can therefore be seen as an important first step in gaining further knowledge about judicial behavior in EU legal integration. An in-depth study of how Swedish judges are reasoning allows us to identify hypotheses about what drives the judges’ actions in the prelim-inary ruling procedure. Subsequent research can examine whether the hy-potheses also hold true in other European contexts.

So far, the choice of studying Swedish judges has in part been motivated with reference to the feasibility of accessing respondents. It has also, in part, been motivated with reference to our limited knowledge about variations in the judges’ mode of reasoning, which means that we have little reason to doubt the representativeness of the Swedish judiciary. Lastly, the choice of studying Swedish judges may also be viewed in light of the literature on US judicial politics and EU legal integration and its expectations about judicial decision-making.

As was elaborated on in chapter 2, previous research that shows an inter-est in the micro foundations of judicial behavior has assumed a rational choice model of judicial decision-making (Dunoff and Pollack 2017: 241; Rehder 2014: 394). According to this literature, judges are, just like most actors driven by preferences over expected outcomes (Epstein et al. 2013; Epstein and Lindquist 2017; Segal and Spaeth 2002; Spiller and Gely 2008). In the European context, researchers inspired by the US judicial politics lit-erature have mainly characterized judges as rational actors interested in in-creasing their own power and influence vis-à-vis the political branches and other judges (Alter 2001; Burley and Mattli 1993) regardless of what mem-ber state they are working in.

82

How can the case of the Swedish judiciary be understood in relation to these theoretical assumptions? This needs to be elaborated on in relation to the differences between the US legal system and the Swedish legal system. Chapter 2 raised the question of how well the dominant theories about judi-cial behavior, developed with the intention to understand US judges and in particular US Supreme Court justices, travel to the European context. It was also discussed whether we can expect legal decision-makers to follow the same logics and dynamics as political decision-makers (Rehder 2014: 394). In the US, the political science research on courts has for long established that courts are important actors in American politics (Dahl 1957) and judges are to be studied the same way as any other political actor (Maveety 2009; Murphy 1964; Shapiro 2002: 4). In particular, the commonalities of courts and the legislative and executive branches of government are emphasized. Judicial action is thereby seen as a sub-type of political action which fits into the broader political process in the US political system (Rehder 2007: 10).

Given this context, it is not surprising that previous research makes the assumption that the judges’ decision-making in the US can be placed on a par with how other political actors are reasoning; they think about expected costs and benefits of different decisions (Spiller and Gely 2008: 34-35). But the questions is whether and to what extent can we expect Swedish judges to make decisions based on expected consequences for their self-regarding preferences, such as expanding their powers. The political and legal contexts that Swedish judges find themselves in are in many ways different from the contexts in which the US judges are placed. Bell has suggested that the self-image of a Swedish judge is as a member of the (impartial) public admin-istration (Bell 2006: 295). The Swedish system of government keeps the judiciary and the administration conceptually distinct from politics and em-phasizes that each branch has a specialized role. The boundaries for those roles are also clearly defined (Anton 1980: 162; Bell 2006: 273). Swedish judges are therefore not exposed to the same political mode of reasoning as the US judges are. Political reasoning has been described as interest-laden and occupied with issues of power (Shapiro 2002: 187). For example, many US states elect judges by popular vote and each judge’s political party affili-ation is made explicit. A judge’s expected goal is to be (re)-elected which makes them similar to politicians (Hall 1995). This is something that is like-ly to contribute to a more interest- or consequentialist-oriented mode of rea-soning among US judges. In contrast, the recruitment of judges in Sweden is meritocratic (Bell 2006: 244).

The judges working in US courts are in many ways most likely to consid-er expected outcomes on the basis of their personal preferences. However, the same is not true for the Swedish judiciary. The organization of the Swe-dish legal system is in important aspects the opposite of the US one. The Swedish judiciary can therefore be understood as a critical, least likely, case for testing whether one of the dominant assumptions in previous research

83

can be substantiated empirically. If the analysis would show that Swedish judges in fact employ a consequentialist mode of reasoning and strive to promote self-regarding preferences as they make decisions in the preliminary ruling procedure, it can be argued that this assumption is likely to also hold true in other European domestic judiciaries that are more similar to the US judicial branch than the Swedish judiciary is.

Selecting respondents and carrying out the interviews In order to prepare for the interviews with the Swedish judges a pilot study was conducted between the fall of 2014 and the spring of 2015. Four Swe-dish judges and three legally trained staff members of Judge Fernlund’s cab-inet at the CJEU were interviewed about the preliminary ruling procedure.33 The purpose of the pilot study was to gain practical knowledge about the inner workings of both Swedish courts and the preliminary ruling procedure. The informants in the pilot study provided detailed information about how the oral and written procedures are conducted in Swedish courts. Having this background knowledge made it easier to formulate follow-up questions and discuss different aspects of the judges’ work when conducting the actual interviews. In addition, mastering the ‘technical’ language and being knowl-edgeable about the topic of concern is believed to help the interviewer gain respect from elite respondents (Kvale and Brinkmann 2009: 147). Apart from providing an in-depth understanding of the judges’ work, these inter-views also made it possible to try out different interview questions. This turned out to be very valuable since some of the original interview questions that had been derived from previous research did not make sense to the in-formants included in the pilot study. Those questions were therefore refor-mulated.

For the actual interview study, 20 semi-structured interviews with Swe-dish judges were carried out during 2016. The respondents were selected first and foremost on the basis of having taken part in requesting preliminary rulings at least one time. The reason for this is obvious: if they had not taken part in the preliminary ruling procedure any discussion about their experi-ence of this procedure would have been purely hypothetical. The selection of respondents among those judges with experience of the procedure was then made with the intention to achieve variation in one aspect that previous re-search has identified as important (King and Horrocks 2010: 29). Drawing upon previous research, a reoccurring claim is that judges working at differ-ent levels in the judicial hierarchy have different goals or preferences (Baum 1997: 23). In particular, and in the European context, judges working in the lower courts are expected to be more likely than high court judges to em-

33 The judges had experience from working in Swedish high courts, one of them had experi-ence from academia, and one had worked at the Swedish judge’s cabinet at the CJEU.

84

brace EU legal integration since it gives them new powers (Alter 2001). The selection of respondents has thus been made with the intention to capture how judges from first instance courts compare to judges in the courts of ap-peal and judges in the final instances courts perceive the preliminary ruling procedure.

Based on a chronological list of all the 102 requests for preliminary rul-ings made by Swedish courts from 1995-2015 (Bernitz 2010; Bernitz 2016), the cases were sorted into two groups according to which court branch they belonged to (civil or administrative). Then, the cases were sorted by court level (courts of first instance, courts of appeal and courts of last instance). The cases were then chosen at random from each court level and branch. This selections process resulted in ten cases from the courts of final instance, six cases from the courts of appeal and four cases from the courts of first instance.

By reading the judgment pertaining to each case it was possible to deter-mine which judges that had been involved in requesting the preliminary rul-ing from the CJEU. Identifying these judges was easiest in the requests that had originated from high courts and the courts of appeal since those rulings can often be found online using legal databases.34 For the requests originat-ing from courts of first instance, it was necessary to contact the courts in question and ask for a copy of the judgment in which the judges’ names could be found. The judges were subsequently contacted.35

The respondents that had been selected all received information about the research project (see Appendix C). Almost all of the judges agreed to partic-ipate in the study. Three judges declined to be interviewed stating reasons such as “not interested,” “on vacation,” or simply, “do not remember.” Each of the 20 interviews lasted between 45 to 75 minutes and they have all been recorded and later transcribed by the author. The interviews were conducted either in the judges’ offices or in their homes. On one occasion the interview took place at a café. The respondents were granted anonymity and hopefully this made them more comfortable to speak freely about the topics brought up during the interview.

As previously mentioned, the interview guide36 consisted of a number of themes with open-ended questions. The first question posed was “Can you describe the case(s) you have referred to the CJEU?” in order to allow the judges to reflect upon the decision-making process using their own words. 34 For example: Infotorg juridik 35 In some cases more than one judge had been involved in requesting a preliminary ruling. In those cases the choice of which one of the three or five judges to contact was decided by lot. In other cases the judge who signed the final judgement was not the same as the one who had requested the preliminary ruling (since it takes approximately two years for the CJEU to answer the questions, the judge may no longer work in the same court). In those cases it was necessary to find the original decision to request a preliminary ruling in order to find out who the judge responsible for the reference was. 36 The interview guide can be found in Appendix C.

85

This type of “grand tour question” (Leech 2002a: 667) allowed the respond-ents to talk freely about something that they know about. Then, it was possi-ble to probe deeper into what specific decisions they had made in these cases and for what reasons. In general, this strategy worked well.

The purpose of asking the judges themselves about the choices they make in the preliminary ruling procedure was to come closer to how they reason and what considerations they make before deciding to refer a case or express opinions. This is information that is difficult if not impossible to infer from only studying patterns of judicial behavior at the macro level. However, building an analysis on the statements made by judges may give rise to other methodological problems, such as biased answers. The respondents may consciously or subconsciously try to put themselves in a favorable light even though they have been granted anonymity. A remedy to such problems is to discuss how the potential bias may influence the results. In this case, the research question concerns how the judges are reasoning. Can we expect these judges to give biased answers? Yes, because formally judges should base their decision on legal considerations (Baum 1997: 60), which in this case translates to the formal rules. There is therefore a risk that the analysis underestimates the occurrence of other types of considerations in the judges’ reasoning. This would be the case if, for example, the judges only referred to the formal rules when describing their choices in the preliminary ruling pro-cedure.

The problem of biased answers is, however, more likely to occur when the inquiry concerns the outcome of the case itself. That is, when the judges are asked about how they have been discussing different ways of interpreting the law during their conferences and why they chose a particular outcome over another (Marvell 1978). Fortunately, this type of judicial decision-making is not the focus of the present thesis. Instead, the aim is to under-stand how the judges make choices in the preliminary ruling procedure. This type of procedural decision-making is not surrounded by deliberation priva-cy. The judges are therefore more likely to speak openly about the decisions they have made in the preliminary ruling procedure than about their legal rulings. The concluding part of chapter 6 discusses how the result from the interview study is to be interpreted in the light of potential biases.

Further methodological considerations can be found in connection with the empirical analyses in chapters 4, 5, and 6. Chapters 4 and 5 are con-cerned with the national courts’ aggregated behavioral patterns while chapter 6 centers on the reasoning of individual judges.

86

4. The behavioral patterns of national courts: Supporting EU legal integration or protecting national law?

One of the starting points for this thesis is that what choices national courts make in the preliminary ruling procedure and the types of behavioral pat-terns these choices give rise to in aggregate have important implications for the pace and direction of EU legal integration. Despite being acknowledged as highly important decisions that are directly related to the CJEU’s ability to foster EU legal integration (Alter 2001:36; Conant 2013; Nyikos 2006; Wind 2010), the national courts’ key choices have not yet been systematical-ly examined in previous empirical research. The chapter therefore sets out to answer the first research question: What types of choices are national courts making in the preliminary ruling procedure?

The main empirical material consists of an original data set including 470 randomly selected cases referred from the national courts to the CJEU be-tween 1992 and 2012.37 The courts’ choices are analyzed 1) across member state groups, 2) over time and, 3) across court levels. The chapter also ex-plores the relationship between the two key choices and the aggregated be-havioral patterns that emerge. The results are compared to the theoretical propositions derived from the approaches of judicial empowerment and sus-tained resistance. The two approaches, which were chiseled out in chapter 2, denote the two main positions in the theoretical debate between those who portray the national courts as either being supporters of the supranational legal system or as being defenders of member state sovereignty (Conant 2013:595; Obermaier 2008; Pollack 2013).

What types of cases are national courts referring to the CJEU? The national courts’ first decision when confronted with a case that concerns EU law is to decide whether to refer the case to the CJEU or not. This deci-sion is very important since the CJEU is dependent on the willingness of 37 Detailed information about the data set can be found in chapter 3. If no other data sources are mentioned, this is the data set that has been used.

87

national courts to supply it with cases. Without a steady inflow of requests for preliminary rulings, the CJEU would lose its main tool for developing and expanding the reach of EU law in different policy areas. In particular, the CJEU is dependent on the influx of cases that can be considered to have a high degree of political sensitivity. This is so because these cases concern the removing of national policies that stand in conflict with the supreme EU law, something that is essential for legal harmonization between the member states and for the furthering of EU legal integration. As was discussed in chapter 3, EU law cases can vary in their degree of political sensitivity from a low degree of political sensitivity (politically non-sensitive and somewhat politically sensitive cases) to a high degree of political sensitivity (politically sensitive and highly sensitive cases). For the purpose of the present inquiry it is therefore important to assess to what extent national courts are allowing the CJEU to decide cases that have a high degree of political sensitivity.

From the two rival theoretical approaches on EU legal integration we can derive fundamentally different answers to the question of whether and to what extent national courts are allowing the CJEU to decide cases that have a high degree of political sensitivity. As was discussed in chapter 2, the judi-cial empowerment approach expects national courts to refer all types of cas-es to the CJEU, even the ones that have a high degree of political sensitivity. The reason for this is that national courts want to strengthen EU legal inte-gration and by extension increase their own influence over legal policy and doctrine (Mattli and Slaughter 1998; Weiler 1991). Conversely, the sustained resistance approach predicts that national courts are reluctant to request pre-liminary rulings when the case can be considered to have a high degree of political sensitivity. The reason for withholding such cases is that the courts want to protect the integrity of the member states’ legislative competences from further EU intrusions as well as uphold the coherence of the domestic legislation (Golub 1996; Wind 2010).

We know from previous research and the official statistical reports from the CJEU that courts from different member states differ in regard to one type of court behavior, namely, how many requests for preliminary rulings they make each year (Broberg and Fenger 2013; Court of Justice of the European Union 2017). It is therefore important to examine whether there are significant differences between member state courts in regard to the types of cases they send to the CJEU.

In this study, the small share of cases from each country does not allow for general comparisons between courts in different member states.38 How-ever, it is possible to compare the behavior of courts in different groups of EU member states. Previous research has pointed out that the behavior of courts in the founding member states differs from the courts in the newer member states when it comes to the number of requests being made. For 38 For information about the share of cases from each member state, see chapter 3.

88

instance, courts in the founding member states have made a greater number of references than courts in Greece, Portugal, and Spain (Wind et al. 2009:81). The expected variation in behavior between courts in the old and the new member states has therefore been examined in the analyses of what types of cases national courts refer to the CJEU.

What types of cases are national courts referring to the CJEU? Figure 4.1 shows the average share of different types of cases that all member state courts refer to the CJEU. It also displays the share of different types of cases that courts in the founding EU member states (EU 6: Belgium, France, Ger-many, Italy, Luxembourg, and the Netherlands) and courts in the newer member states (EU 9: Austria, Denmark, Finland, Greece, Ireland, Portugal, Spain, Sweden, and the UK) send to the CJEU. 39

Figure 4.1. The shares of different types of cases referred to the CJEU between 1992 and 2012. Number of observations: EU 6 = 293, EU 9 = 177, EU average = 470

From Figure 4.1 we learn that for all courts (EU average), the most frequent-ly referred type of case is the politically sensitive one with a share of 50 percent. The somewhat politically sensitive cases make up 35 percent of the referrals and the politically non-sensitive cases make up 10 percent. Finally,

39 It is not yet possible to make a separate analysis of the countries that joined the EU most recently (2004 and later) due to the small share of cases referred from the courts in those countries.

10

33

52

5

13

36

47

4

10

35

50

4

010

2030

4050

Shar

e of

cas

es (p

erce

nt)

EU 6 EU 9 EU average

Politically non-sensitive Somewhat politically sensitivePolitically sensitive Highly politically sensitive

89

4 percent of the referrals consist of highly politically sensitive cases.40 This result shows that the majority of the referred cases (54.5%)41 have a high degree of political sensitivity. Turning to the types of cases referred from different groups of member states, there are only minor differences in behav-ior between the average national court and the courts in the founding mem-ber states (EU 6) and the courts in the newer member states (EU 9).

Which member states’ courts are referring cases with a high degree of po-litical sensitivity to the CJEU? Since the share of cases from each country in the data set is low, it is not possible to compare court behavior between member states. However, courts from eight member states (Austria, Bel-gium, France, Germany, Italy, the Netherlands, Spain, and the UK) are rela-tively well represented in the sample42, making it possible to compare their shares of cases with a high degree of political sensitivity. These eight mem-ber states are: Austria, Belgium, France, Germany, Italy, the Netherlands, Spain, and the UK. The results are reported in Figure 4.2.

40 Confidence intervals (EU average/full sample) at the 95% level: Highly politically sensi-tive: 2.8-6.5%, politically sensitive: 45.7-54.7%, somewhat politically sensitive: 30.9-39.5%, politically non-sensitive: 7.9-13.6%. 41 Confidence intervals at the 95% level for the share of cases with a high degree of political sensitivity (approx. 55%, n=470, politically sensitive and highly politically sensitive cases): 50-59%. 42 Member states that are represented with 23 references or more

90

Figure 4.2. The share of cases with a high degree of political sensitivity from the eight member states that are represented with 23 cases or more in the sample. Note: The solid line denotes the mean share of cases with a high degree of political sensi-tivity (54.5%) in the entire sample (EU average). The dashed lines denote the lower (50%) and upper (59%) bound of the confidence interval for the EU average at the 95% level. Numbers in parentheses show the number of cases with a high degree of political sensitivity divided by the total number of references.

Based on the information in Figure 4.2, it appears as if the courts in Spain, Italy, France, and Austria refer a greater share of cases with a high degree of political sensitivity than the EU average. However, none of these differences are statistically different from the EU average. This finding further reinforc-es the interpretation of the result reported in Figure 4.1, namely, that there are no major differences in behavior between courts in different member states when it comes to the referral of cases that have a high degree of politi-cal sensitivity to the CJEU.

Previous research has suggested that national court behavior may change over time as courts become more acquainted with the EU legal system (Golub 1996). For this reason, it is interesting to examine whether national courts have become more or less inclined to refer cases with a high degree of political sensitivity to the CJEU over the studied time period. Figure 4.3 below shows the total number of requests for preliminary rulings between

51% (22/43)

52% (53/101)

55% (26/47)

57% (28/49)

60% (25/42)

60% (18/30)

63% (37/59)

65% (15/23)

UK

Germany

Belgium

Netherlands

Austria

France

Italy

Spain

20 40 60 80

Share of cases with a high degree of political sensitivity (percent)

91

1992 and 2012 and the share of preliminary rulings that have a high degree of political sensitivity between 1993 and 2011.43

Figure 4.3. The share of preliminary rulings with a high degree of political sensitivi-ty between 1993 and 2011 and the total number of requests for preliminary rulings between 1992 and 2012. Note: Confidence intervals at the 95% level for the share of preliminary rulings with a high degree of political sensitivity. Source: Author’s data set and Court of Justice of the European Union, (2017).44

As Figure 4.3 shows, it has become more common for national courts to refer cases with a high degree of political sensitivity to the CJEU over the years, from 47 percent45 in 1993 to 60 percent46 in 2011. The increase has been fairly stable but there have been fluctuations. For instance, the share of

43 The estimates for each year have been calculated by using a moving average mean to show the long-term trend over the 20 year long time period. Since a moving average mean (with data from three years) has been used, the numbers for the years 1992 and 2012 are omitted. For example, to calculate the mean for 1993, observations from 1992, 1993, and 1994 have been used. 44 The number of preliminary rulings with a high degree of political sensitivity in the data set: 256, the number of preliminary rulings in the population: 5509. A table displaying the number and share of preliminary rulings by year (in data set and in the population) can be found in Appendix A. 45 Confidence interval 41-53% 46 Confidence interval 58-62%

0

50

100

150

200

250

300

350

400

450

0%

10%

20%

30%

40%

50%

60%

70%

80%

19921994199619982000200220042006200820102012

Prel

imin

ary

rulin

gs (f

requ

ency

)

Shar

e of

pre

limin

ary

rulin

gs w

ith a

hig

h de

gree

of

polit

ical

sens

itivi

ty (p

erce

nt)

Year

Share of preliminary rulings with a high degree of political sensitivity(percent)

Preliminary rulings (frequency)

92

cases with a high degree of political sensitivity peaked in 2001 with 64 per-cent47 before decreasing to 49 percent48 in 2008. Interestingly, these varia-tions in type of case do not fully correspond to the variations in the total number of requests for preliminary rulings. This means that it does not seem to be the case that the increase in the share of preliminary rulings with a high degree of political sensitivity in the sample is driven by an overall increase in the number of cases sent to the CJEU. At least not between the years 1998 and 2001, when the number of preliminary rulings instead decreased while the share of cases with a high degree of political sensitivity remained fairly constant. Although an examination of the cause of these fluctuations in the share of references with a high degree of political sensitivity is beyond the scope of the present inquiry, two explanations spring to mind. It is possible that the share of cases with a high degree of political sensitivity is influenced either by the expansion of EU competence or by the accession of new mem-ber states.

How can the expansion of EU competence influence the national courts’ behavior? The upturn in the share of cases with a high degree of political sensitivity starting in 1997 follows in the wake of the Maastricht Treaty. This Treaty entered into force in 1993 and fundamentally transformed the European Community into the European Union. The Council’s use of quali-fied majority voting was extended to new policy areas, the European Parlia-ment’s influence increased, and the EU was entrusted with new legislative competences related to the Social Protocol. For example, the EU came to have a greater influence over working conditions, equal treatment of men and women, and social security (Hix and Høyland 2011: 256-257). It can be hypothesized that this transfer of legislative competence from the member state to the EU gives rise to new legal conflicts related to the relationship between national law and EU law, that is, the type of legal conflict that has a high degree of political sensitivity. The reason for this is that member states cannot predict what specific rules and regulations the EU will enact based on its new competence. It is therefore impossible for a member state to adjust all of its legislation to the anticipated EU law in advance, meaning that there will be periods during which member state law is incompatible with the new EU rules and regulations. Throughout those periods legal conflicts involving the validity of national law are bound to arise in the member state and end up in the courts. In turn, this rise in the overall share of cases with a high degree of political sensitivity in the national courts’ caseload is likely to also be reflected in their references made to the CJEU. After some time, however, the legal conflicts generated by the Treaty changes will have been resolved and that might explain the temporary decrease in the share of cases with a high degree of political sensitivity between 2002 and 2005. It should be the

47 Confidence interval 60-68% 48 Confidence interval 46-53%

93

task of future research to investigate whether Treaty changes can account for the fluctuations in the types of cases national courts refer to the CJEU.

The other event that occurred around the same time as the increase in 1997 of references with a high degree of political sensitivity is that three new member states joined the EU (Austria, Finland, & Sweden). How can the accession of new member states influence the courts’ behavior? When a country joins the EU, it takes time for it to align its policies with EU law (Falkner and Treib 2008; Haverland 2000). It is therefore likely that during those first years of membership, when the member state is still adjusting to EU law, many legal conflicts between national law and EU law will arise. This may have contributed to the increase in the share of cases with a high degree of political sensitivity between 1996 and 2001 that can be seen in Figure 4.3. Similarly, from 2008 onwards, following the accession of a total of twelve new EU member states in 2004 and in 2007, there is also an in-crease in the share of cases having a high degree of political sensitivity.

The question is whether or not the new member state courts drove the in-crease in the share of cases with a high degree of political sensitivity. In order to properly test this hypothesis one needs data on the share of different types of cases referred from the courts in each member state. However, since this information is unavailable we have to resort to another measure of court activity, namely, the overall number of references from each member state. These data are gathered from the annual report on judicial activity before the CJEU (Court of Justice of the European Union 2017). This information makes it possible to assess whether courts from new member states actually referred a greater number of cases during the first years following the coun-try’s accession to the EU than they did later on. If they requested a greater number of cases during the first years, it gives preliminary support for the hypothesis that these member state courts contributed to the increase in the share of references with a high degree of political sensitivity that is dis-played in Figure 4.3. In other words, it suggests that many conflicts between member state law and EU law arise in new member states, which in turn leads the national courts in these countries to request a greater number of preliminary rulings in cases with a high degree of political sensitivity.

Turning to the statistical report, the behavior of the courts in the member states that joined the EU in 1995 seems to support the proposed hypothesis. These member state courts referred a higher share of cases between 1997 and 2001 than between 2002 and 2006, just as we would have expected. However, it turns out that this was only due to the references made by the Austrian courts. The report shows that, between 1997 and 2001, Austrian courts referred 195 cases to the CJEU. During the subsequent five years the references made by Austrian courts dropped to 85 cases. In contrast, the courts in Finland and Sweden together referred almost the same number of cases between 1997 and 2001 (46) as they did between 2002 and 2006 (52). Moreover, when it comes to the references made by the courts in the ten

94

member states that joined the EU in 2004, their behavior is clearly incon-sistent with the hypothesis. During the first five years of membership, these courts made 46 references, a figure that increased to 157 in the subsequent five years (Court of Justice of the European Union 2017: 106). These find-ings suggest that it takes time for the national courts in new member states to learn how to engage in the preliminary ruling procedure (Wind 2010: 1056). It may also be the case that legal conflicts do not arise immediately after a country’s accession to the EU but rather after a while.

In sum, the Austrian courts’ behavior corroborates the proposed hypothe-sis that many legal conflicts arise during the first years of EU membership and that this, in turn, might lead to an increase in the share of references with a high degree of political sensitivity. However, the national courts in the other twelve member states instead sent fewer cases to the CJEU in the first five years than they did later on. Although we lack information about the member state courts’ references of cases with a high degree of political sen-sitivity, these findings suggest that the accession of new member states does not matter for what types of cases national courts send to the CJEU.

The final step in the analysis of the types of cases national courts refer to the CJEU requires a look at the variations in behavior between courts at dif-ferent levels in the judicial hierarchy. A pertinent question in the literature is whether lower national courts are more supportive of European legal integra-tion than courts of final instance. Refining the argument made by the judicial empowerment approach, Alter (2001) has suggested that it is predominantly lower national courts, often defined as first instance courts, which stand to gain new powers from EU legal integration and from participating in the preliminary ruling procedure. In contrast, courts of final instance are less inclined to embrace the EU legal system because the CJEU challenges the finality of their rulings (Alter 1998b; Alter 2001). How does the inter-court competition hypothesis apply to the national courts’ first key choice? Draw-ing upon the logic of competition, first instance courts are expected to be more likely than courts of final instance to refer cases with a high degree of political sensitivity to the CJEU. This is simply because the referral of cases with a high degree of political sensitivity provides the CJEU with opportuni-ties to expand the reach of EU law, something that is believed to benefit the courts of first instance but not the courts of final instance. Figure 4.4 pre-sents the share of cases with a high degree of political sensitivity, and the share of cases with a lower degree of political sensitivity, across three court levels: first, appeal, and final.

95

Figure 4.4. The share of cases with, respectively, a low or high degree of political sensitivity across court levels. Number of observations: First instance = 185, Courts of appeal = 116, Final instance = 169.

Are first instance courts also more likely to refer cases with a high degree of political sensitivity compared to courts of final instance as suggested by the inter-court competition hypothesis? Figure 4.4 shows that first instance courts refer a greater share of politically sensitive cases (60%) than courts of final instance (52%) but this difference of eight percentage points is not sta-tistically significant.49 This result means that the inter-court competition hypothesis is not supported. However, there might be other factors that in-fluence the relationship between court level and the national courts’ behav-ior. Further tests of this hypothesis are therefore conducted in chapter 5.

Results How can these empirical results from investigating the national courts’ first key choice be understood in light of the theoretical expectations derived from the approaches of judicial empowerment and sustained resistance? The theoretical approaches do not offer a set standard for what constitutes a low or a high share of different types of cases. For instance, there is no absolute 49 The confidence interval for the difference between two proportions (60% and 52%) is -0.03 – (+) 0.18 or -3 – 18 percentage points. Since this interval includes the value zero the null hypothesis is not rejected and we therefore conclude that there is no statistically significant difference between the two proportions at the 95% level.

40

60

5149 48

52

020

4060

Shar

e of

cas

es (p

erce

nt)

First instance Courts of appeal Final instance

Low degree of political sensitivity High degree of political sensitivity

96

figure for what share of cases with a high degree of political sensitivity would confirm the thesis of the judicial empowerment approach.50 However, even without a distinct point of comparison, what we learn from the present study is that a majority of the referred cases have a high degree of political sensitivity (54.5%). This finding, that more than half of the requests are ei-ther politically sensitive or highly politically sensitive, is deemed to be in-consistent with the prediction derived from the sustained resistance ap-proach. The sustained resistance approach expects national courts to be re-luctant to refer these types of cases to the CJEU. But, as the results show, cases with a high degree of political sensitivity are the most commonly re-ferred type of case. This cannot be interpreted as reluctant behavior on the part of the national courts.

In light of the theoretical propositions the results from this study show that there is no widespread reluctance among the national courts to refer cases with a high degree of political sensitivity to the CJEU. On the contrary, the referral of cases with a high degree of political sensitivity is a common practice and one that has been increasing over time. It means that the empiri-cally most common court behavior, to refer cases with a high degree of polit-ical sensitivity, corroborates the expectation derived from the judicial em-powerment approach.

However, one minor caveat must be added. The present study provides no information about the share of cases with a high degree of political sensitivi-ty that national courts could have referred to the CJEU but did not. There is a small risk that only a limited share of all those cases is being sent to the CJEU. It would imply that national courts are in fact withholding cases with a high degree of political sensitivity from the CJEU, which is contrary to the expectations of the judicial empowerment approach. However, a pilot study on Swedish court behavior suggests that is not the case (Hessel 2017:33). This pilot study reveals that among the EU law cases that are not referred to the CJEU there is no overrepresentation of cases with a high degree of polit-ical sensitivity.51 In fact, the Swedish courts even refer a slightly higher share of cases with a high degree of political sensitivity to the CJEU than they withhold (Hessel 2017:33). Hence it appears as if national courts are not withholding the lion’s share of the cases with a high degree of political sen-sitivity from the CJEU although further research is needed to confirm that 50 The literature does, however, provide one empirical point of comparison. Researchers who have used a similar albeit not identical definition of a high degree of political sensitivity report close to no such referrals from Danish courts (Wind et al. 2009, Wind 2010). The share of cases with a high degree of political sensitivity is thus higher in this study compared to previous findings from Denmark. However, these authors define cases with a high degree of political sensitivity as public prosecution cases. In those cases, it is common for EU law to stand in conflict with national legislation. It is difficult to say whether this is a wider defini-tion compared to the definition used in the present study. 51 For all types of cases, the shares of different types of cases that are referred are almost identical with the shares of cases that are being withheld.

97

this result also holds in other member states. But regardless of any reserva-tions, what the present study shows us is that national courts are in fact re-questing preliminary rulings from the CJEU in cases that have a high degree of political sensitivity and these cases make up the majority of the total number of references. This is a type of behavior that clearly comes much closer to the expectations formulated by the judicial empowerment approach than the ones formulated by the sustained resistance approach.

What types of opinions are national courts expressing in the requests for preliminary rulings? The national courts’ second key decision is to decide what types of opinions, if any, to include in the requests for preliminary rulings. As was discussed in previous chapters, scholars have described the national courts’ practice to express opinions as part of the judicial dialogue between the national courts and the CJEU (Jacobs 2003:548; Rosas 2007:126). How can the practice of expressing opinions be understood theoretically? According to Scharpf, the CJEU lacks knowledge about the legal particularities of each member state, which leads it to frequently make decisions that threaten the coherence of national legal policies (Scharpf 2009:186-187). In a worst-case scenario the CJEU’s rulings may have a disruptive effect on the national legal system, effectively contributing to disintegration (Alter 2001:48; Dehousse 1998:173). It has therefore been theorized that national courts include opin-ions in order to show the CJEU what interpretations of EU law are accepta-ble in their member state’s legal and political context. If national courts are not expressing opinions the risk is greater that the CJEU delivers a ruling that contravenes national legal and political traditions. It is further assumed that the CJEU sometimes takes the national courts’ opinions into account when deciding the outcome of a case (Nyikos 2006).

This leads to the question of what is the content of those opinions. As was discussed in chapter 3, national courts may express three different types of opinions: opinions that in some way support EU law and practice, opinions that in some way support national law and practice, and opinions that sup-port a position unrelated to the national-EU dimension (Leijon and Karlsson 2013).52 The theoretical expectation derived from the sustained resistance approach is that national courts mainly express support for national law. One of the suggested reasons for this is that national courts want to protect the integrity of the member states’ legislative competences from further EU intrusions (Wind 2010:1057) and maintain control over specific policy out-comes (Golub 1996:381). Another proposed explanation is that national

52 Examples of all three types of opinions can be found in chapter 3.

98

courts want to protect the quality and coherence of domestic legislation (Alter 2001:48; Dehousse 1998:173). Conversely, the judicial empowerment literature would instead claim that national courts mostly express support for EU law. This is because national courts want to show the CJEU that they accept and welcome an expansion of EU legal integration, which by exten-sion increases their own powers (Burley and Mattli 1993:63; Mattli and Slaughter 1998:190; Weiler 1994:523).

Turning to the empirical analysis, what one first would like to know is how common it is for national courts to express any type of opinion. Figure 4.5 below presents the average share of opinions that national courts include in their requests for preliminary rulings. Since previous research has sug-gested that national court behavior might vary between groups of member states (Wind et al. 2009) the figure also shows the share of opinions coming from the courts in the founding member states (EU 6): Belgium, France, Germany, Italy, Luxembourg, and the Netherlands, and the newer member states (EU 9): Austria, Denmark, Finland, Greece, Ireland, Portugal, Spain, Sweden, and the UK.

Figure 4.5. The share of cases with/without opinions. Number of observations: EU 6 = 293, EU 9 = 177, EU average = 470.

Figure 4.5 shows that for all courts (EU average) opinions are expressed in 48 percent of the requests for preliminary rulings. Similar shares are found in

5149

54

46

52

48

020

4060

Shar

e of

cas

es (p

erce

nt)

EU 6 EU 9 EU average

No opinion included Opinion included

99

EU 6 and in EU 9.53 Is this considered to be a high or low share of opinions? A previous study of national court behavior between the years 1966 and 1994 provides an empirical point of reference (Nyikos 2006). When compar-ing the behavior of the member state courts that have been included in both the present study and in the previous study, namely the courts in Belgium, France, Germany, Italy, the Netherlands, and the UK, we find that there has been an increase in the share of opinions of eight percentage points.54 This is a statistically significant difference, which suggests that the national courts from these six member states have become somewhat more likely to include opinions over time.55 Overall, the finding that national courts include opin-ions in close to half of all their requests for preliminary rulings is slightly higher compared to previous research.

Just as in the analysis of the national courts’ first key choice, the small share of cases from some member states means that it is not possible to compare the behavior of courts between countries. However, courts from eight member states are relatively well represented in the sample, making it possible to compare how common it is for them to include opinions in the requests for preliminary rulings. The shares of opinions from courts in Aus-tria, Belgium, France, Germany, Italy, Spain, the Netherlands, and the UK are presented in Figure 4.6.

53 Confidence interval at the 95% level: on average, the national courts include opinions in 43-52% of the requests for preliminary rulings. 54 The previous study reported opinions in 41 percent of the 574 cases investigated (Nyikos 2006), while the result from the present study shows that courts in these six countries have included opinions in 49 percent of the cases referred (See Appendix A). National courts from Belgium, France, Germany, Italy, The Netherlands, and the United Kingdom have together referred 329 of the cases included in this sample and opinions have been included in 162 of those requests. The share of opinions from those courts is 162/329 = 49.2 percent. 55 The increase is small but statistically significant. At the 95% level of confidence, the in-crease in the share of opinions ranges between 1 – 15 percentages point. The data from Nyi-kos’ study: n=574, share of opinions 41.3 percent. Furthermore, Nyikos (2006) uses a broader definition of opinions than the one used in the present thesis. About ten percent of the opin-ions Nyikos found are not explicit opinions (as used in the present study) but opinions “for-mulated as questions.”. This means that Nyikos found explicit opinions in only 37% of the cases (213 explicit opinions, n=574) while the present study found opinions in 49% of the cases from courts in the same six member states. The difference in the share of opinions between the present study and Nyikos’ is thus even greater (11 percentage points) and at the 95% level of confidence the increase in the share of opinions ranges between 5-19 percentage points.

100

Figure 4.6. The share of preliminary rulings with an opinion from the eight member states that are represented with 23 cases or more in the sample. Note: The solid line denotes the mean share of cases with opinions (48 %) in the entire sample (EU aver-age). The dashed lines denote the lower (43 %) and upper (52 %) bound of the con-fidence interval for the EU average at the 95% level. Numbers within parentheses show the number of cases with opinions divided by the total number of preliminary rulings. 56

As can be seen in Figure 4.6, there are some variations between courts in different member states when it comes to the practice of including opinions. The Spanish courts appear to include more opinions than the average court, but this difference is not statistically significant.57 However, the courts in Germany are significantly more likely to include opinions in their requests for preliminary rulings (60%) than the average national court (48%).58 Con-versely, the French courts include significantly fewer opinions in their re-quests for preliminary rulings (27%) than the average court.59 56 A table showing the share of cases with an opinion for each member state can be found in appendix A 57 The confidence interval for the difference between the two proportions (70% and 48%) is between -0.01 – 0.4 or -1 – 40 percentage points. Since this interval includes the value zero, the difference is not statistically significant at the 95% level. 58 The confidence interval for the difference between the two proportions (60% and 48%) is 0.01 – 0.23 or 1 – 23 percentage points. Since this interval does not include the value zero we reject the null hypothesis and conclude that there is a statistically significant difference be-tween the two proportions at the 95% level. 59The confidence interval for the difference between the two proportions (48% and 27%) is 0.01 – 0.36 or 1 – 36 percentage points. Since this interval does not include the value zero we

27% (8/30)

45% (22/49)

47% (20/43)

47% (22/47)

48% (20/42)

49% (29/59)

60% (61/101)

70% (16/23)

France

Netherlands

UK

Belgium

Austria

Italy

Germany

Spain

20 40 60 80

Share of cases with an opinion (percent)

101

A result that stands out in Figure 4.6 is the statistically significant differ-ence (33 percentage points) in behavior when comparing French and Ger-man courts.60 This difference is interesting because both France and Germa-ny have been part of the EU legal community since the start in the 1950s. A reasonable expectation is that the courts in both countries have about the same level of knowledge and experience of the preliminary ruling procedure and the option to include opinions. French courts are thus expected to in-clude opinions in their requests just as frequently as the German courts. Yet, we find that French courts are less likely than the German courts to include opinions in the requests for preliminary rulings. How can this difference be explained? The difference in behavior between French and German courts might have something to do with the courts’ approaches to the use of so-called ‘dissenting opinions’. Domestic legal systems that have a tradition of dissenting opinions allow judges to express their disagreement with the ma-jority opinion of the court that forms the basis of its judgment. In regard to dissenting opinions, French judges are not allowed to issue dissenting opin-ions while the German judges are allowed to express such separate opinions (Raffaelli 2012:18, 22). It may therefore be theorized that French courts are reluctant to express opinions in the requests for preliminary rulings since the practice of writing dissenting opinions is prohibited in the French legal sys-tem. Future research should pay more attention to this type of variation since it shows that German courts are more likely than the French courts to engage in a judicial dialogue with the CJEU. This suggests that the French courts might have less influence on the CJEU’s rulings and the development of EU law than courts from Germany.

Since previous research has suggested that national court behavior chang-es over time (Golub 1996; Wind et al. 2009), it is interesting to take a closer look at how the share of opinions in the requests for preliminary rulings might vary over time. Figure 4.7 below presents the share of cases with opin-ions for each year during the period 1993-2011.61

reject the null hypothesis and conclude that there is a statistically significant difference be-tween the two proportions at the 95% level. 60 The confidence interval for the difference between the two proportions (60% and 27%) is 0.12 – 0.50 or 12 – 50 percentage points. Since this interval does not include the value zero we reject the null hypothesis and conclude that there is a statistically significant difference between the two proportions at the 95% level. 61 The share of opinions for each year has been calculated by using the moving average mean in order to smooth out the short-term fluctuations and better describe the general trend for the entire time period.

102

Figure 4.7. The share of preliminary rulings with opinions between 1993 and 2011. Note: Confidence intervals at the 95% level.62

There has been an increase in the share of cases with an opinion from 33 percent63 in 1993 to 56 percent64 in 2011. However, there have also been fluctuations over time, which raises the question of how those variations in national court behavior can be understood. A comprehensive inquiry into the matter is beyond the scope of this thesis but what can be provided is an out-line of three potential explanations: Can the fluctuations have been caused by (1) the distribution of information about the practice of including opin-ions to national courts; (2) the expansion of EU competence; or, (3) the ac-cessions of new member states?

The national courts’ actual knowledge about the option to include opin-ions is the first proposed explanation for the two upturns in the share of opinions. The claim is simple: when the national courts become aware of the fact that opinions are welcome in the requests for preliminary rulings, they will be more likely to express them. In 1996, the CJEU sent out an infor-mation note concerning the preliminary ruling procedure to the member states. In 2005, the CJEU decided to update this note and publish it in the Official Journal of the European Union (European Court of Justice 2005). This note contains information about various things related to the prelimi-nary ruling procedure including guidelines about the option to express opin- 62 The number of preliminary rulings with opinions the data set: 225. A table displaying the number and share of preliminary rulings with opinions by year can be found in Appendix A. 63 Confidence intervals 26%-41% 64 Confidence interval 53%-58%

0%

10%

20%

30%

40%

50%

60%

70%

80%

1993 1995 1997 1999 2001 2003 2005 2007 2009 2011

The

shar

e of

case

s with

opi

nion

s (pe

rcen

t)

Year

103

ions. As Figure 4.7 shows, the circulation of these information notes in 1996 and 2005 is on both occasions followed by a rise in the share of cases with an opinion. Hence, it is hypothesized that when informed or reminded about the option to include opinions in the references to the CJEU, national courts will become more likely to actually express their views.

Turning to the second proposed explanation, Figure 4.7 shows that, after the Maastricht Treaty entered into force in 1993, there is a fairly stable in-crease in the share of cases with opinions, peaking at 71 percent in 2001.65 This increase almost coincides with the upturn in the share of cases with a high degree of political sensitivity referred to the CJEU (see Figure 4.3). As was discussed previously, it is likely that the expansion of EU legislative competence that followed from the Maastricht Treaty entering into force gave rise to an increase in the legal conflicts between national legislation and EU law. How may this development have influenced the national courts’ inclination to include opinions? The proposed hypothesis is that in the wake of such new conflicts between EU and member state law, national courts find it to be important to express their views on the relationship between the two sources of law. For instance, the courts may want to inform the CJEU about problems that will arise in the member state if a particular national policy is deemed incompatible with EU law. This rise in the share of politi-cally sensitive legal conflicts brought about by the Maastricht Treaty is thus expected to have led to an increase in the occurrence of opinions. This effect is also believed to diminish after awhile when the courts have sorted out the most pressing legal issues. Indeed, after 2001 the share of cases with opin-ions decreases and in 2004 the share of opinions is down to 32 percent.66

The third hypothesis suggests that it was the accession of new member states that contributed to an increase in the share of opinions. Three countries joined the EU in 1995, ten countries in 2004, and an additional two in 2007. As can be seen in Figure 4.7, opinions became more frequent both after 1996 and after 2004. Why would national courts in a new member state be more likely than courts in the older member states to include opinions in their request for preliminary rulings? The proposal is that they, as newcomers to the EU legal system, find it important to inform the CJEU about what inter-pretation of EU law is acceptable in their member state. National courts are after all experts on domestic law while the CJEU has little knowledge about the particularities of, for example, Slovakian contract law. Unless such opin-ions are included, the CJEU will be unaware of the potentially disruptive effects EU laws have on the national legal systems and legislative traditions. A rise in the share of opinions is therefore to be expected when new coun-tries join the EU. Future research should further investigate whether any of these hypotheses hold up to empirical testing.

65 Confidence interval 68%-75% 66 Confidence interval 25%-39%

104

So far, we have learned that national courts are including opinions in al-most half of all the cases they refer to the CJEU. It is now time to examine the substantive content of those opinions. As was discussed in chapter 3, national courts may express three different types of opinions: opinions that in some way support EU law and practice, opinions that in some way sup-port national law and practice, and opinions that support a position unrelated to the national-EU dimension (Leijon and Karlsson 2013).

Figure 4.8 below shows the average share of different types of opinions from all member state courts. It also includes the share of different opinions from the courts in EU 6 (Belgium, France, Germany, Italy, Luxembourg, and the Netherlands) and in EU 9 (Austria, Denmark, Finland, Greece, Ireland, Portugal, Spain, Sweden, and the UK).

Figure 4.8. The percentage of different types of opinions. Number of observations: EU 6 = 144, EU 9 = 73, EU average = 225.

Figure 4.8 shows that national courts most commonly express opinions in support of EU law (EU average 52%).67 Furthermore, there are no statistical-

67 Confidence interval at the 95% level: 46-59%

56

29

15

49

26 25

52

28

19

020

4060

Shar

e of

opi

nion

s (p

erce

nt)

EU 6 EU 9 EU average

Supporting EU law Supporting national law Other position

105

ly significant differences between the average court and the courts from the EU 6 and EU 9 when it comes to what types of opinions they express.68

What about differences between high and low courts when it comes to the types of opinions they include in the request for preliminary rulings? As previously discussed, the inter-court competition hypothesis suggests that lower national courts, mainly courts of first instance, stand to gain more from cooperating with the CJEU compared to courts of final instance (Alter 2001). The courts of first instance are therefore expected to be expressing support for the EU law to a higher extent than courts of final instance. Con-versely, the highest courts are expected to be more likely to express support for national legislation compared to courts of first instance. To investigate the overall differences between court levels, Figure 4.9 shows the share of the three different types of opinions across the three court levels.

Figure 4.9. The share of different types of opinions across court levels. Number of observations: First instance = 94, Courts of appeal = 51, Final instance = 80.

The results reported in Figure 4.9 show that support for EU law is most commonly expressed by the courts of first instance followed by the courts of appeal and the courts of final instance. Conversely, support for national law 68 For example, it is not statistically significant at the 95% confidence level that the courts in the founding member states (EU 6) are more likely to express support for the EU law than the average court (56% vs. 52%). The confidence interval for the difference between the propor-tions is: -0.0655 – (+) 0.1483. Since the value zero is included in the interval, the difference is not significant.

61

2118

55

2422

41 40

19

020

4060

Shar

e of

opi

nion

s (p

erce

nt)

First instance Courts of appeal Final instance

Supporting EU law Supporting national lawSupporting any other position

106

is most commonly expressed by the courts of final instance followed by the courts of appeal and the courts of first instance. Testing if the differences are statistically significant confirms the inter-court competition hypothesis: Lower courts (first instance courts and courts of appeal) are more supportive of EU law than the final instance courts69 and, conversely, the courts of final instance are more supportive of national law compared to the lower courts.70

Results How can the outcome of the national courts’ second key choice be under-stood in relation to the theoretical expectations? Since previous research and theory offer meager guidance regarding what constitutes a low or high share of opinions, the way forward is to compare the relative shares of different opinions. From the preceding analyses we have learned that the most com-mon behavior among the national courts is to behave as expected by the judicial empowerment approach; in 52 percent of the cases with opinions, courts express support for the EU law. Conversely, the behavior expected by the sustained resistance approach is less common. Only in 24 percent of the cases in which opinions have been included are national courts expressing support for national law. This means that the result corroborates the expecta-tions derived from the judicial empowerment approach. Moreover, when looking at the behavior of courts at different levels of the judicial hierarchy, it becomes apparent that the support for EU legal integration, in the shape of opinions supporting EU law, mainly comes from lower courts. In contrast, courts of final instance are just as likely to exhibit a sustained resistance behavior as they are to support legal integration.

Combining cases with opinions: four behavioral patterns To fully understand the behavior of national courts throughout the prelimi-nary ruling procedure it is necessary to analyze the two key choices in con-junction. This information will help us answer important empirical questions such as how common it is for national courts to express support for national law in cases with a high degree of political sensitivity. But what is more, it will also help resolve the theoretical controversy about the role of national courts in EU legal integration. To what extent are national courts actually exhibiting a behavioral pattern that either supports EU legal integration or

69 The confidence interval at the 95% level for the difference between two proportions (58% vs 41%) shows that lower courts (first instance and appeals) are between 3-31 percentage points more likely to express support for EU law than the courts of final instance. 70 The confidence interval at the 95% level for the difference between the two proportions of opinions (40% vs 22%) shows that courts of final instance are between 4-31 percentage points more likely to support national law compared to the lower courts.

107

one that protects member state sovereignty? Table 4.1 below shows how the 182 cases that include opinions supporting either the EU law or the national law are distributed across four possible behavioral patterns.

Table 4.1. The national courts’ combined behavioral patterns

Opinions supporting EU law

Opinions supporting national law

Cases with a high degree of political sensitivity71

Judicial empowerment

42.1 %

Compatibility defended

21.3 %

Cases with a low degree of political sensitivity72

Access contained

22.4 %

Sustained resistance

14.2 %

Note: Number of observations: 182 (the cases with opinions supporting either national law or EU law).

As can be seen in Table 4.1, the single most common behavior among the national courts is judicial empowerment. In 42 percent73 of the cases national courts provide the CJEU with opportunities to expand the reach of EU law by referring cases with a high degree of political sensitivity together with encouragement to do so in the form of opinions supporting EU law. In con-trast, the least common behavior is sustained resistance. This combination of cases with low political sensitivity together with opinions defending national policies occurs in only 14 percent74 of the cases. Returning to the theoretical controversy, it becomes clear that the role of national courts in EU legal integration mostly mirrors the expectations formulated by the judicial em-powerment approach. The conclusion is that by viewing the courts’ choices in conjunction, the behavioral pattern that emerges is one that furthers EU legal integration rather than one that protects member state sovereignty.

However, apart from showing that judicial empowerment is the main be-havioral pattern among national courts, Table 4.1 also highlights that their behavior actually deviates from previous expectations. As was discussed in chapter 2, combining the two key choices will result in a total of four differ-ent behavioral patterns. Yet, only two of them have been theorized in previ-ous research. What is striking is that 44 percent of the cases do not corre- 71 High = Politically sensitive cases/highly politically sensitive cases 72 Low = Non-/Somewhat politically sensitive cases 73 95% confidence interval: 35% to 50% 74 95% confidence interval: 9% to 20%

108

spond to the expectations derived from either the sustained resistance ap-proach or the judicial empowerment approach. This means that almost half of the national courts’ behavior has not been properly defined by previous theory and research.

The present thesis proposes that the two undefined behavioral patterns are best understood from the viewpoint that national courts are gatekeepers placed in the intersection between two political systems. This position ex-poses national courts to pressure and demands from both the member state and the EU. The pressure from the EU level consists of the rules regulating the preliminary ruling procedure that call on the national courts to send cases to the CJEU. Formally, national courts should refer all cases in which the interpretation of EU law is unclear to the CJEU, regardless of the cases’ degree of political sensitivity. Failure among the courts to abide by this trea-ty regulation may result in infringement proceedings against the member state.

But strictly following the EU rules and allowing the CJEU to decide cases with even a high degree of political sensitivity can be met with opposition in the member state. This is the pressure national courts face from the member state level. The CJEU has been known for frequently overturning national policies, something that the member states’ governments are not very fond of since they want to remain in control of sensitive policy matters. This does not mean that member states want to abolish the preliminary ruling proce-dure. It is rather about its scope: in what types of cases should the CJEU actually have the final say? The member states are assumed to be opposed to having politically sensitive questions removed from political decision-making and handed over to the courts ( Blauberger 2014: 460; Grimm 2015). The member states and the CJEU are likely to disagree about where to draw this line between EU and national competences, and between legal and polit-ical questions. In particular, it becomes an even bigger problem if the mem-ber state perceives that the CJEU is delivering controversial rulings that are insensitive to the member states’ legal and political traditions (Grimm 2015, Scharpf 2009).

However, the CJEU and the member state government have no say in whether a preliminary ruling should be requested, not even in a case with a high degree of political sensitivity. This is up to the national courts to decide, which leads us to the question of how they are handling the pressure from, on the one hand, the EU and, on the other hand, the member state, when making choices in the preliminary ruling procedure. As the data in Table 4.1 show, the national courts’ response to the conflicting demands placed on them can manifest itself in two different behavioral patterns.

In the first of these positions national courts respond by allowing the CJEU to decide cases with a high degree of political sensitivity. However, in those requests they are also including opinions that explicitly defend domes-tic law and national practices. This combined response to the two choices

109

enables national courts to deal with the demands stemming from both the EU level and the national level. In theoretical terms the national courts’ behavior can be described as compatibility defended. By requesting preliminary rul-ings even in cases that have a high degree of political sensitivity, national courts fulfill the Treaty requirement that all unclear EU law cases shall be referred to the CJEU. At the same time they are also adhering to the con-cerns of the member state government by including opinions defending na-tional practices in the requests for preliminary rulings. That is, they are pro-claiming the compatibility of national practices with EU law before the CJEU. This behavior makes up 21 percent75 of the cases in the empirical investigation (Table 4.1.).

What is characteristic of the behavioral pattern compatibility defended is that the national courts’ response to being gatekeepers is not to shun conflict but rather to engage in a dialogue with the CJEU and actively challenge its attempts to expand EU law. If the national courts want to avoid further EU legal integration they are taking a risk when allowing the CJEU to decide cases with a high degree of political sensitivity. Yet, they may believe that by expressing opinions supporting national policies and practices they can put up a defense against further expansion of EU law. Although the effect of opinions on the CJEU’s final ruling has not yet been systematically exam-ined, anecdotal evidence indicates that the CJEU sometimes has taken the views of national courts into account in its judgment (Alter 2001: 61-62; Nyikos 2006; Rosas 2007: 126). If successful in influencing the CJEU’s final ruling by expressing an opinion in support of national policy, the na-tional courts’ behavior may have implications beyond the case itself. It can contribute to setting precedents that limit the reach of EU law not only in the case at hand but also in adjacent cases and policy areas. Using Rytter & Wind’s (2011) terminology this behavior suggests that national courts are trying to be active co-producers of EU legal norms in politically sensitive cases rather than passive consumers (Rytter and Wind 2011: 488).

The second of the previously unexplored positions is defined by national courts that respond to the pressure from EU and the member state in a way that can be described as access contained. Here, the courts express opinions supporting EU law when requesting preliminary rulings in cases that have a low degree of political sensitivity. This is the second most common behav-ioral pattern with 22 percent.76 It is a different way of handling the conflict-ing demands compared to the first proposal but it can nevertheless be under-stood by applying the same theoretical perspective, that national courts are gatekeepers in EU legal integration who respond to the demands from two different levels. How then does this behavior fulfill the wishes of both the member state and the EU? To begin with, the courts are accommodating to

75 95% confidence interval: 16% to 28% 76 95% confidence interval: 17% to 29%

110

the EU level and the CJEU by including opinions in the requests for prelimi-nary ruling that support EU law. That is, the courts are pointing out that fur-ther expansion of EU law is indeed acceptable in the cases they refer. The member state governments for their part are expected to not mind this behav-ior since the cases in question have a low degree of political sensitivity. It means that the referral is unlikely to result in the overturning of any sensitive member state policy.

However, while this behavioral pattern opens up further EU legal integra-tion in some types of cases, the gate remains closed in others. The CJEU’s ability to expand the reach of EU law is thus contained in the sense that na-tional courts do not give it access to cases with a high degree of political sensitivity. It should be kept in mind that this behavioral pattern may lead to dissatisfaction at the EU level. The CJEU’s influence over legal develop-ment is in part circumscribed because some policy matters are out of its reach. In a worst-case scenario, the national courts’ behavior will result in sanctions against the member states and their courts. Still, access contained is a method that national courts can use if they want to balance the pressure from the two political systems. National courts that behave in this way are participating in a dialogue with the CJEU and are co-producers of EU law but only within a specific sphere, the sphere of cases with low political sen-sitivity.

In summary, viewing the national courts’ key choices in conjunction sheds new light not only on how national courts behave in EU legal integra-tion but also on what is missing in previous theory and research. While the expectations derived from the two main approaches provide fruitful insights, when the choices are studied separately it becomes clear that further theoriz-ing is necessary for understanding the courts’ combined behavioral patterns.

Summary of chapter This chapter has provided the first systematic analysis of what key decisions national courts make in the preliminary ruling procedure to answer the first research question: What types of choices are national courts making in the preliminary ruling procedure? Answering this question contributes to resolv-ing the theoretical debate between those who expect national courts to sup-port EU legal integration and those who expect national courts to defend national sovereignty. The first take away is that the national courts' aggre-gated behavioral pattern corroborates the expectations derived from the judi-cial empowerment approach. It means that national courts mainly exhibit a behavior that is contributing to further EU integration. The national courts’ single most common behavior is to refer cases that have a high degree of political sensitivity and, to express opinions in support of EU law. This is

111

what we see both when examining the national courts’ two key choices indi-vidually and together.

However, probing into the courts’ aggregated behavioral patterns nuances this picture. The second take away from this chapter is that unless the two key decisions are considered together one fails to accurately understand the national courts’ behavior in the preliminary ruling procedure. In around half of all cases that carry opinions, national courts are not behaving as predicted by either one of the two dominating theoretical approaches. The results have important implications for the research field. Previous research has not paid sufficient attention to the national courts’ key choices and it has resulted in a theoretical debate that lacks important insight into what the national courts’ position as gatekeepers in EU legal integration actually entails. This thesis suggests that a substantial part of the national courts’ behavior can be under-stood as an attempt to balance conflicting demands. By adopting the per-spective that national courts are gatekeepers, placed in the intersection be-tween the domestic political system and the EU, allows a new understanding of the national courts’ aggregated behavior beyond the characterization of national courts as being either supporters of EU legal integration or defend-ers of national policy. Instead, the courts’ behavior may also lead to access contained or compatibility defended.

In sum, what is required is a reformulation of the theoretical controversy regarding the role of national courts in EU legal integration. While it is true that the single most common behavior is one that supports further EU legal integration, national courts also display a behavior that can be understood as trying to strike a balance between member state autonomy and European integration. The implication of this finding is further elaborated on in the concluding chapter.

112

5. Explaining what choices national courts make in EU legal integration

The conclusion that could be drawn from the analyses in chapter 4 is that the national courts’ most common behavioral pattern is making choices that enable further EU legal integration. This pattern corresponds to the expecta-tions of the judicial empowerment approach. However, the descriptive anal-ysis also displayed considerable variations in the choices that courts make. For instance, in close to 30 percent of cases, national courts express support for national law, and in approximately 50 percent of cases, they express sup-port for EU law. It raises the question of why we see these differences. This chapter therefore seeks to address the second research question: What factors can explain the variations in the types of choices that national courts make in the preliminary ruling procedure?

Theoretical explanations The starting point of the present thesis is that the choices that national courts make in the preliminary ruling procedure have important implications for the pace and direction of EU legal integration. What we would like to know is why some courts are more inclined than others to allow the CJEU to decide cases with a high degree of political sensitivity, and why some courts are more likely than others to express opinions supporting national law in re-quests for a preliminary ruling. In theoretical terms, we want to know why some national courts make choices that help the CJEU to foster EU legal integration, while others make choices that defend the member states’ law.

With regard to the first key choice, what can be examined empirically are the types of factors associated with the referral of cases with a high versus low degree of political sensitivity.77 With regard to the second key choice, it is possible to analyze the factors that render national courts more likely to express support for national law than for EU law.

77 Although we lack information about the counterfactual outcome since there are no data on cases with a high degree of political sensitivity that have not be referred, it is still possible to analyze the factors that might account for the variation in the actual referrals between courts and member states.

113

Chapter 2 elaborated on how previous theory explains national court be-havior in EU legal integration. Building on these theoretical propositions, the next section forms hypotheses about what can explain the variations in the types of choices that national courts’ make in the preliminary ruling proce-dure.

Hypotheses derived from judicial empowerment approach

Inter-court competition The inter-court competition hypothesis is based on the neofunctionalist claim that the preliminary ruling procedure empowers national courts by giving them the opportunity to exercise judicial review. This assumption, which is the cornerstone of the judicial empowerment approach, has been further developed by subsequent research with the purpose of explaining why some national courts appear to support EU legal integration, while others do not. The argument is that not every national court becomes empowered by partic-ipating in the procedure. Instead, it is lower domestic courts that gain influ-ence at the expense of higher courts. By referring cases to the final authority on EU law, the CJEU, lower courts are able to deliver rulings that the high-est domestic courts will also have to accept. With the support of the CJEU’s answers to the legal questions posed in the requests for preliminary rulings, all national courts can create precedents and exercise judicial review – pow-ers that previously were reserved for the courts of final instance (Alter 1996; Alter 2001).

Based on the theoretical argument of the inter-court competition hypothe-sis, what types of cases can we expect lower or higher national courts to refer to the CJEU? High courts are believed to be reluctant to request prelim-inary rulings in cases with a high degree of political sensitivity since they do not want to surrender to the CJEU their control over the development of politically sensitive issues. In contrast, lower courts are expected to refer cases with a high degree of political sensitivity to the CJEU since doing so means that they will be able to exercise judicial review in important cases that otherwise would have been addressed by the high courts. The lower national courts are therefore expected to be more likely to refer cases with a high degree of political sensitivity than high courts since it means that they, for the first time, with the support of the CJEU’s rulings, can create prece-dents on politically important issues.

Expanding on this argument, it is possible to also formulate a hypothesis about the relationships between court levels and types of opinions. Since the EU legal system provides lower national courts with new powers that in-

114

crease their influence vis-à-vis high courts, they are expected to be more likely to express support for EU law, compared to the courts of final in-stance. Higher courts are, in contrast, more likely than other courts to stand up for and to defend national legislation against EU law intrusions. The rea-son for this greater likelihood is that domestic courts of final instance are hypothesized to have a skeptical attitude toward the preliminary ruling pro-cedure since the CJEU undermines the finality of their rulings and, by exten-sion, their legal influence (Alter 2009).

In summary, the courts’ place in the judicial hierarchy is believed to pro-vide them with different incentives in the preliminary ruling procedure. But what defines a low respectively high court? Often, the high courts are re-ferred to as courts of final instance, suggesting that this category only in-cludes courts against whose decisions there is no judicial remedy. Lower courts, conversely, are sometimes referred to as first instance courts (Alter 2009: 100), while at other times, they are referred to as all other courts ex-cept for the Constitutional court or the Supreme court (Alter 2001: 49). Whether the intermediate courts of appeal are to be understood as lower courts that become empowered by EU legal integration is not clear (Sweet and Brunell 1998b: 71).

To properly test the inter-court competition theory, two different sets of hypotheses about what choices national courts make are formulated. The original version treats the courts of appeal as lower courts since they are not at the top of the domestic legal hierarchy. They simply do not have as much power to lose to the CJEU, compared to the courts of final instance.

Inter-court competition: The final instance hypotheses - H1a: Courts of final instance are less likely to refer cases with a

high degree of political sensitivity to the CJEU, compared to courts of first instance and courts of appeal.

- H1b: Courts of final instance are more likely to express support for national legislation, compared to courts of first instance and courts of appeal.

In the alternative formulation of the inter-court competition hypothesis, it is primarily the first instance courts that become empowered by the EU legal system. We would therefore expect to see a difference in behavior between the first instance courts and the other, higher courts. The argument is that the preliminary ruling procedure has its greatest impact on the life of the first instance courts since they traditionally have had none or very limited influ-ence on domestic legal developments.

115

The alternative inter-court competition: The first instance hypothe-ses

- Alternative H1a: Courts of first instance are more likely to refer cases with a high degree of political sensitivity to the CJEU, compared to courts of final instance and courts of appeal.

- Alternative H1b: Courts of first instance are less likely to ex-press support for national legislation, compared to courts of final instance and courts of appeal.

Hypotheses derived from the sustained resistance approach

Public opinion One of the hypotheses that can be derived from the sustained resistance ap-proach is that public opinion influences the courts’ behavior in the prelimi-nary ruling procedure. In the EU context, it has been theorized that low pub-lic support for EU integration in a member state makes the courts less likely to refer cases to the CJEU (Carrubba and Murrah 2005: 405-406; Mattli and Slaughter 1998). Previous research has suggested that the rational behavior of a court confronted with public opinion opposed to its decisions is to modi-fy its behavior in order to protect the legitimacy of the court and its policy effectiveness. Otherwise, the risk is that the courts’ decisions are rejected (Giles et al. 2008: 295). The argument is that any court that cares about its perceived institutional legitimacy must attempt to anticipate whether its de-cisions will be respected and followed by the public (McGuire and Stimson 2004: 1019).

Based on previous research, it can be expected that widespread Euroscep-ticism among the public constrains the behavior of the national courts, influ-encing their decision-making in the preliminary ruling procedure. With re-gard to the first key choice, low support of EU integration is expected to render national courts less likely to refer cases with a high degree of political sensitivity to the CJEU. The argument is that, when national courts request preliminary rulings in politically sensitive cases, it indicates that the CJEU will have the opportunity to expand the reach of EU law, even in outright political matters that constitute the core of member states’ internal affairs. This condition is likely to create discontent among citizens. If citizens are skeptical of EU membership, they will not be fond of having their own courts sending cases to the CJEU, which has a track record of delivering rulings that strengthen European integration.

116

A negative view of the EU in a member state can also influence the con-tent of the national courts’ opinions, making the courts more likely to defend national law. The courts in these member states are assumed to know that referring cases to the CJEU often leads to rulings that strengthen European integration and that the citizens disapprove of this development. To preserve their legitimacy among the public, national courts might attempt to avoid the preliminary ruling procedure altogether. However, they are formally obliged to refer cases to the CJEU. As was theorized in chapter 4, one way for the national courts to balance these two conflicting demands is by expressing opinions that defend national legislation in the request for a preliminary rul-ing. In this manner, the national courts can show the public that they are protecting national practices although they refer cases to the CJEU. There are also examples of cases in which domestic courts have argued in support of national legislation and have noted that the public will react negatively to any rulings that undermine the existing national policy.78 The CJEU can con-sider the national courts’ arguments when deciding the case. In fact, the CJEU is likely to value this type of information about the interpretations of EU law that are acceptable in different member states (Nyikos 2006). The CJEU knows that it is dependent on the national courts’ acceptance of its interpretations and on the member states’ willingness to abide by its rulings. From time to time, oppositions and challenges from national courts have resulted in that the CJEU has practiced judicial restraint (Alter 2001: 61). Taken together, the proposition that public opinion influences judicial be-havior can be summarized in the following two hypotheses.

78 For example, a UK court informed the CJEU that the export of live calves from the UK to other member states with less rigid animal protection controls was “a topic of considerable public concern in the United Kingdom”. See: The Queen v Minister of Agriculture, Fisheries and Food, ex parte Compassion in World Farming Ltd.

117

Public opinion: The Euroscepticism hypotheses

- H2a: National courts in member states with low public support for EU integration are less likely to refer cases with a high degree of political sensitivity to the CJEU, compared to courts in mem-ber states in which the public is more positive toward further in-tegration.

- H2b: National courts in member states with low public support for EU integration are more likely to express support for national legislation, compared to courts in member states in which the public is more positive toward the EU.

Majoritarian democracy It has been suggested by previous research that the manner in which the member states have organized the relationship between the legal and politi-cal branches of government is likely to influence the willingness of national courts to refer matters to the CJEU (Wind et al. 2009: 71). The proposed hypothesis, which has its basis in the sustained resistance approach, is that national courts from member states that have a tradition of majoritarian democracy are reluctant to participate in the preliminary ruling procedure. According to Wind et al. (2009), a majoritarian democracy is characterized by a strong tradition of parliamentary sovereignty, while the role of courts, as a counter-majoritarian force, is restricted. The member states that have been identified as being organized according to this tradition are Denmark, Finland, Sweden and the UK (Wind et al. 2009: 72).

In majoritarian democracies, the courts’ behavior in the preliminary ruling procedure is believed to be formed by the societal agreement that courts should have a limited influence over public policy and that the legislature has the final say in political matters. These traditions make national courts hesitant to engage in a judicial cooperation with the CJEU, which has a track record of overturning decisions made by national parliaments (Nergelius. 2001: 88; Wind et al. 2009: 73). In addition to this implicit tradition, we have seen in chapter 2 that government representatives in two majoritarian democracies, Denmark and the UK, have instructed national courts to avoid the preliminary ruling procedure. Instead of sending cases to the CJEU, the courts are advised to settle questions of EU law themselves (Craig 1998: 205; Wind et al. 2009: 77-78). Such authoritative statements can also occur in other member states; however, the key difference lies in how national courts respond to this type of appeal. In a majoritarian democracy, courts are expected to defer to the political branches and to follow signals from the government. The courts’ understanding of the relationship between the polit-ical and judicial branches of government is that they should cooperate (Wind et al. 2009: 76).

118

What does this mean for the national courts’ behavior in the preliminary ruling procedure? The expectation is that courts in majoritarian democracies are less likely than other courts to refer cases with a high degree of political sensitivity to the CJEU. Cases in which sensitive national policies are at stake can be said to belong to the core of the member states’ interests. The courts’ understanding is that these types of matters are to be decided by the legislature, not the CJEU. This stance of national courts can also be rein-forced by signals from political actors and high court judges, for instance, the instructions sent from the Danish Ministry of Justice to the national courts about how and when to request a preliminary ruling (Wind 2010; Wind et al. 2009).

How does a tradition of majoritarian democracy influence the types of opinions that courts include in their requests? It is a commonly held view in the literature that the CJEU and the preliminary ruling procedure challenge the political power of the member states’ governments. When a case reaches the courts, member states cannot be certain that the CJEU will show consid-eration for the domestic policy or practice (Alter 1998a; Blauberger 2012; Wind 2010; Wind et al. 2009). In fact, the CJEU is known for frequently overturning national policies (Scharpf 2009). What is important for a mem-ber state government in the preliminary ruling procedure is therefore to de-fend its interests. In a majoritarian democracy, in which national courts are more likely to accommodate signals from political actors, we would expect courts to be more likely to express support for national legislation than for EU law.

The majoritarian democracy hypotheses

- H3a: National courts in member states with a majoritarian tradi-tion of democracy are less likely to refer cases with a high degree of political sensitivity to the CJEU, compared to other courts.

- H3b: National courts in member states with a majoritarian tradi-tion of democracy are more likely to express support for national law, compared to other courts.

The political sensitivity of the case As was discussed in chapter 2, the type of cases with which national courts are confronted might also matter for the type of opinion that they decide to include in the request for a preliminary ruling. Nyikos theorizes that national courts are more likely to express opinions when they are more certain that the CJEU will listen (Nyikos 2006: 532). The CJEU is more likely to listen when the case at hand is expected to result in a legal controversy due to clashes between EU law and national law (Nyikos 2006: 533). What can be

119

derived from these claims is that, when a case has a high degree of political sensitivity, national courts are more likely to express opinions, and the CJEU is more likely to consider these opinions when deciding cases.

What about the impact of cases with a high degree of political sensitivity on the type of opinion included in the request for a preliminary ruling? Alt-hough Nyikos did not specify the types of opinions that national courts can be expected to include in the references to the CJEU, all of the examples of opinions in the study show how national courts defend member state law (Nyikos 2006: 528, 530). A reasonable expectation is therefore that national courts are more likely to express opinions defending national legislation in cases with a high degree of political sensitivity, compared to cases with a low degree of political sensitivity. The reason for this expectation is that there is more at stake in the politically sensitive types of cases, for instance, the risk that national policy will be overturned. National courts that want to protect the coherence of the domestic legal order might find it especially important to express their views in this type of case.

The political sensitivity hypothesis:

- H4b: National courts are more likely to express support for na-tional law in cases that have a high degree of political sensitivity, compared to other cases

Hypotheses derived from the Europeanization literature

Experience of the preliminary ruling procedure The Europeanization literature suggests that differences in behavior between member states’ courts can be explained by how much time they have had to adapt to the EU political system (Börzel 2003). The claim is that national courts grow accustomed to the EU legal order through repeated interactions with the CJEU in the preliminary ruling procedure (Golub 1996; Ladrech 2010). Courts must learn about how the CJEU’s doctrines are to be under-stood and under what circumstances preliminary rulings should be requested, as well as how to formulate these requests.

At the outset, the preliminary ruling procedure might appear complicated and unpredictable to national courts. Courts with less experience with the procedures are expected to be cautious in regard to requesting preliminary rulings and particularly so in cases with a high degree of political sensitivity. For courts that are unfamiliar with the procedure, it is a large step to hand over domestic cases that concern matters of outright political significance

120

(Hirschl 2008: 2) to the discretion of the CJEU. However, over time and after repeated interactions with the CJEU, the national courts are believed to have adapted to EU legal norms and have started to view the procedure as the normal way to decide cases with a high degree of political sensitivity as well. The expectation is that national courts with relatively more experience with the procedure are more likely than other courts to refer cases with a high degree of political sensitivity to the CJEU.

The expectation that courts conform to EU practices over time also has implications for what types of opinions they include in requests for prelimi-nary rulings. Drawing on the Europeanization literature, national courts with more experience with the procedure are believed to be more likely to express support for EU law. The argument is that these courts have, through judicial dialogue with the CJEU, been socialized into viewing EU law and the CJEU’s rulings as necessary for achieving a functioning and coherent legal system. They therefore express that they find further EU legal integration acceptable when requesting preliminary rulings from the CJEU. The theoret-ical expectations can be summarized in two hypotheses.

The experience hypotheses

- H5a: National courts that have relatively more experience with the preliminary ruling procedure are more likely to refer cases with a high degree of political sensitivity, compared to courts with less experience.

- H5b: National courts that have relatively more experience with the preliminary ruling procedure are more likely to express sup-port for EU law, compared to courts with less experience.

Concrete judicial review How does a domestic tradition of concrete judicial review affect the courts’ referral of cases with a high degree of political sensitivity? Here, ‘concrete judicial review’ refers to the power of all courts, not only constitutional courts, to review the constitutionality of legislation in the actual cases brought before them (Mayoral 2012; Vink et al. 2009). For courts that are allowed to perform judicial review at the national level, there is nothing strange about acting as review courts; on the contrary, they will find the pre-liminary ruling procedure to be consistent with their domestic legal practic-es. These courts will also be accustomed to working in a system in which courts are allowed to decide politically salient issues (Hirschl 2008; Vallinder 1994).

121

In contrast, courts from member states in which judicial review is prohib-ited are expected to perceive the preliminary ruling procedure and the prac-tice of having courts deciding political matters as foreign to their legal sys-tems. The judicial review component in the preliminary ruling procedure is directly contrary to the domestic principles regarding the role that courts should play in society (Slaughter et al. 1998; Sweet and Brunell 1998b). For courts that are unfamiliar with the practice of judicial review, the referral of cases with a high degree of political sensitivity to the CJEU goes against their understanding of what they as courts are allowed to do. It can be hy-pothesized that national courts not accustomed to concrete judicial review will have reservations in regard to referring cases with a high degree of polit-ical sensitivity to the CJEU.

Moreover, national courts that exercise concrete judicial review in their domestic legal systems are used to evaluating legal problems in relation to the norms enshrined in the constitution (Alter 1996; Sweet and Brunell 1998b: 68; Vink et al. 2009: 10). This fact accustoms them to arguing in favor of the higher legal norms in cases of a conflict between different sources of law. How does this fact impact the opinions that these courts in-clude in requests for preliminary rulings? In the EU legal system, the CJEU has established that EU law is supreme, which means that it takes precedent over any piece of national legislation. Since courts from member states that have judicial review are fostered in a legal tradition in which the highest legal norm should prevail, they are expected to also apply this rule in the preliminary ruling procedure. That is, a national tradition of concrete judicial review renders national courts more likely to include opinions supporting supreme EU law.

The concrete judicial review hypotheses:

- H6a: National courts in member states in which concrete judicial

review is allowed are more likely to refer cases with a high de-gree of political sensitivity to the CJEU, compared to other courts.

- H6b: National courts in member states in which concrete judicial review is allowed are more likely to express support for EU law, compared to other courts.

122

Overview of the hypotheses Table 5.1 below provides an overview of the hypotheses.

Table 5.1 The expected relationship between the explanatory factors and the two dependent variables

Hypotheses Type of case: High degree of political

sensitivity

Type of opinion: Supporting nation-

al law

Orig.H1a/b Inter-court competition: Final instance

Less likely to refer More likely

Alt.H1a/b Inter-court competition: First instance

More likely to refer Less likely

H2a/b Euroscepticism Less likely to refer More likely

H3a/b Majoritarian democracy Less likely to refer More likely

H4a/b Case: High degree of political sensitivity

N/A More likely

H5a/b Experience of the prelim-inary ruling procedure79

More likely to refer Less likely

H6a/b Concrete judicial review More likely to refer Less likely

Data and measurement To test these seven hypotheses, this chapter examines the national courts’ choices in the preliminary ruling procedures from 1992 to 2012. The units of analysis are the individual legal cases that national courts have sent to the CJEU within the framework of the preliminary ruling procedure. Descriptive information about all of the variables can be found in Appendix B.

Dependent variables In the first analysis, the question regards the factors that can explain the var-iations in types of cases national courts allow the CJEU to decide. This vari-ation is measured by the dependent variable Type of case. It is coded 1 if the

79 Since the experience variable is measured as the number of cases that national courts have referred it also functions as a control variable by filtering away the potential impact of being a court in a member state that has been an EU member for a longer or shorter time period.

123

case referred by the national court has a high degree of political sensitivity (politically sensitive and highly politically sensitive cases) and 0 if the case has a low degree of political sensitivity (non-sensitive and somewhat politi-cally sensitive cases). A case that has a high degree of political sensitivity has been defined as a case in which the nature of the legal dispute is a con-flict between national law and EU law, possibly concerning a salient issue, such as social rights. In contrast, a case with a low degree of political sensi-tivity is a case in which the nature of the legal dispute concerns different interpretations of EU law.80

In the second analysis, the aim is to explain the variation in the types of opinions that national courts express in their requests for preliminary rulings. The theoretically most interesting types of opinions are those in which the national court supports national law and those in which the courts support EU law.81 The dependent variable, Type of opinion, is therefore binary, and it is coded 1 if the national court has included an opinion expressing support for the national legislation in the request for a preliminary ruling and 0 if the court instead has expressed support for EU law. The original data set con-sists of a simple random sample of 470 cases. In this sample, these two types of opinions are found in 182 cases.

Independent variables Testing the hypotheses Inter-court competition: Final instance (Original H1a/b) requires the creation of a variable that indicates whether the referring court is a court of final instance or not.82 The resulting dummy variable is called Final instance and is coded 1 if the referring court is a court of final instance and 0 if it is not (that is, the referring court is a court of appeal or a court of first instance). A negative relationship in the first analysis means that courts of final instance are less likely than lower courts to refer cases with a high degree of political sensitivity to the CJEU, which is consistent with hypothesis H1a. Conversely, according to hypothesis H1b, this variable is expected to have a positive impact in the second analysis. That is, courts of final instance are more likely than lower courts to express opinions that support national law.

To test the hypotheses Alternative Inter-court competition: First instance (Alt.H1a/b), the variable First instance was created. By drawing on the same 80 For a detailed discussion about the definitions of different types of cases, see chapter 3. Descriptive information about the two dependent variables can be found in chapter 4 and in Appendix B. 81 This fact indicates that cases in which national courts express support for any “other posi-tion” are not included in the analysis. 82 The name of the referring court can be found in the CJEU’s judgment and information regarding what court level the referring court belongs to, which are obtained from the over-views of the national legal systems that were provided by each member states’ Ministry of Justice.

124

information concerning the court level that the referring court belongs to, the resulting dummy variable takes the value of 1 if the referring court is a court of first instance and 0 if it is not (that is, the referring court is a court of final instance or a court of appeal). A positive relationship means that courts of first instance are more likely than higher courts to refer cases with a high degree of political sensitivity to the CJEU, which is consistent with hypothe-sis Alternative H1a. According to Alternative H1b, the variable First in-stance is expected to have a negative impact on the dependent variable in the second analysis. In other words, courts of first instance are less likely than all higher courts to express opinions that support national law.

The Euroscepticism hypotheses (H2a/b) are tested by creating a variable that indicates the proportion of the population in the member state of the referring court that sees the EU membership as “a bad thing” in the same year as the referral was made.83 The variable is called Euroscepticism and is a continuous variable that ranges from 0.03 to 0.38.84 A negative relationship in the first analysis is consistent with hypothesis H2a. A higher degree of Euroscepticism (low support for the EU) among the public decreases the likelihood that the national courts refer cases with a high degree of political sensitivity to the CJEU. Conversely, a positive relationship in the second analysis is consistent with hypothesis H2b, which means that a high degree of Euroscepticism increases the likelihood that the national courts express opinions that support national law.

Testing the majoritarian democracy hypotheses (H3a/b) requires the crea-tion of a dummy variable called Majoritarian democracy. This variable as-sumes the value of 1 if the member state of the referring court has a tradition of majoritarian democracy and is 0 otherwise. According to Wind (2010) and Wind et al. (2009), Denmark, Finland, Sweden and the UK have been identified as having a tradition of majoritarian democracy. In the first analy-sis, a negative relationship is consistent with hypothesis H3a. The courts in member states that have a tradition of majoritarian democracy are less likely than other courts to refer cases with a high degree of political sensitivity to the CJEU. Conversely, in the second analysis, a positive relationship is con-sistent with hypothesis H3b, which means that the courts in majoritarian democracies are more likely than other courts to express opinions that sup-port national law.

83 The support in percentage was divided by 100. Range: 0=0% of the population considered EU membership to be a bad thing, and 1=100% of the population considered EU membership to be a bad thing. The data come from the Eurobarometer series 1992-2012. See also Carrub-ba & Murrah. (2005) and Mayoral (2012). 84 In Sweden, 38% viewed the EU as a bad thing in 1998. In contrast, only 3% of the Irish respondents answered that the EU membership was a bad thing in 1997. For an overview of the average share of the population who answered that the “EU is a bad thing,” see Appendix B.

125

To test the hypothesis political sensitivity (H4), a dummy variable is cre-ated. This variable is called Political sensitivity and is coded 1 if the referred case has a high degree of political sensitivity and is 0 otherwise. This varia-ble is the same as the dependent variable Type of case in the first analysis. A negative relationship is consistent with hypothesis H4, which means that the national courts are more likely to express support for national law in cases with a high degree of political sensitivity than in other types of cases.

Next, the Experience hypotheses (H5a/b) are tested by creating the con-tinuous variable Experience. This variable indicates the number of cases that the national courts in each member state have referred to the CJEU at the time of requesting a preliminary ruling in the case at hand, and it ranges from 0 to 1,885. The data come from the CJEU’s database. A negative rela-tionship in the first analysis is consistent with hypothesis H5a. Courts in a member state in which the judiciary has more experience with the prelimi-nary ruling procedure are more likely to refer cases with a high degree of political sensitivity to the CJEU. In the analysis of the type of opinion that is included in the referral, a negative relationship is consistent with hypothesis H5b, which means that courts in a member state in which the judiciary has more experience with the preliminary ruling procedure decreases the likeli-hood that these courts express opinions that support national law.

Testing the Concrete judicial review hypotheses (H6a/b) requires the creation of a dummy variable called Concrete judicial review. This variable assumes the value of 1 if concrete judicial review is allowed in the member state of the referring court and is 0 otherwise. According to the Maddex clas-sification, Finland, France, The Netherlands and the UK do not allow con-crete judicial review over legislation, while all other EU member states al-low concrete judicial review (Maddex 2007). In the first analysis of the type of case that is referred, a positive relationship is consistent with hypothesis H6a. Courts that are allowed to perform concrete judicial review are more likely than other courts to refer cases with a high degree of political sensi-tivity to the CJEU. Conversely, a negative relationship in the second analysis is consistent with hypothesis H6b, which means that courts that are allowed to perform concrete judicial review are less likely than other courts to ex-press opinions that support national law.

Models and analysis The hypothesized relationships are estimated with a logistic regression anal-ysis. An ordinary least square (OLS) regression85 has not been used because 85 When the dependent variable is binary, the OLS regression coefficient shows the decrease or increase in the predicted probability of having a characteristic because of a one-unit change in the independent variable. The estimations of the hypothesized relationships with the OLS models yield only minor differences in the results and these differences were not found to be of substantial importance. These models can be found in Appendix B.

126

the dependent variables are dichotomous. Using OLS when the dependent variable is binary leads to various problems such as heteroscedasticity,86

which means that the estimation is inefficient (Pampel 2000: 9). Moreover, all models have been estimated with country-clustered robust standard er-rors.87 This procedure allows the error terms to be correlated within countries (Pepper 2002). The reason for using standard errors clustered by country is that the dependent variables Type of case and Type of opinion vary on the individual case level, while the independent variables Majoritarian democ-racy, Concrete judicial review, Experience and Euroscepticism vary by member state. The observations within a member state may be correlated in some unknown way, which in turn leads to correlated errors within the member state. Failure to control for this expected within-cluster correlation can lead to misleadingly small standard errors (Cameron and Miller 2015).

Analysis I: The referral of cases with a high degree of political sensitivity What factors explain the variation in the types of cases that the national courts refer to the CJEU? To answer this question, the hypothesized relation-ships are first estimated in six bivariate logistic regression models, one for each hypothesis. Then, all of the relevant hypothesized relationships are estimated in a multiple logistic regression model. Because none of the hy-potheses have previously been empirically investigated, this approach is used to explore the relationships among the independent variables. For in-stance, this method makes it possible to identify potentially spurious rela-tionships among the variables.

The results from the analyses are presented in Table 5.2. Odds ratios that are greater than one indicate a positive relationship between the independent variable and the probability that the national courts refer a case with a high degree of political sensitivity to the CJEU rather than a case with a low de-gree of political sensitivity. Conversely, odds ratios that are smaller than one indicate a negative relationship. If the odds ratio is equal to one, there is no relationship between the variables.

86 Heteroscedasticity means that the sample estimates of the standard errors are biased, and this results in invalid tests of significance. See, e.g.,Pampel (2000). 87 The model errors for the cases in the same member state are assumed to be correlated, while the model errors for the cases in different countries are assumed to be uncorrelated. In addition, a multilevel model has not been used because of the relatively small sample size (470 observations). The initial tests of this model did not yield any substantive differences in the results. In addition, because the exploration of the data did not reveal any substantive variation over time (see Chapter 4), the models did not consider variation over time.

127

Table 5.2. The referral of cases with a high degree of political sensitivity (logistic regression, odds ratio)

(1) (2) (3) (4) (5) (6) (7) Final instance 0.86 (0.12) First instance 1.45*** 1.50*** (0.18) (0.20) Euroscepticism 0.14* 1.18 (0.16) (1.89) Majoritarian democracy 0.55** 0.44** (0.16) (0.15) Experience 1.00 1.00* (0.00) (0.00) Concrete judicial review 1.03 0.81 (0.15) (0.11) Constant 1.26*** 1.04 1.60*** 1.32*** 1.18 1.17 1.49* (0.10) (0.10) (0.21) (0.09) (0.18) (0.13) (0.34) Prob > chi2 0.262 0.003 0.084 0.045 0.911 0.835 0.002 Pseudo R2 0.00094 0.00583 0.00375 0.00862 1.89e-

05 3.45e-05

0.0172

Countries 23 23 23 23 23 23 23 Observations 470 470 470 470 470 470 470

Robust standard errors in parentheses (clustered at the member state level) *** p<0.01, ** p<0.05, * p<0.1

Models (1) to (6) report the bivariate relationship between each of the inde-pendent variables and the referral of cases with a high degree of political sensitivity. Model (1) shows that the variable Final instance is not statistical-ly significant.88 However, as can be seen in Model (2), the variable First instance is statistically significant (p<0.01) and has the expected positive sign. Moreover, the variable Euroscepticism in Model (3) and the variable Majoritarian democracy in Model (4) are also significant. Both of these variables have the expected negative sign. Finally, the results reported in Models (5) and (6) show that the variables Experience and Concrete judicial review are not statistically significant. Based on the bivariate results alone, the alternative inter-court competition hypothesis (Alt. H1a), the Euroscepti-cism hypothesis (H2a) and the majoritarian democracy hypothesis (H3a) find preliminary support. To investigate whether these bivariate results hold, however, we must conduct a multiple logistic regression analysis, which is reported in Model (7).

88 The variable Final instance is not statistically significant in any of the models that have been estimated. For example, the multiple logistic regression model including the variable Final instance (which measures the original inter-court competition hypothesis) and all other independent variables apart from the variable First instance, shows no statistical significance.

128

An examination of the results that are reported in Model (7) reveals that the member state’s tradition of majoritarian democracy still has a significant negative effect (p<0.05) when controlling for the other variables. This result corroborates the majoritarian democracy hypothesis (H2a): courts in member states that have a majoritarian tradition of democracy are less likely than the courts from other member states to refer cases that have a high degree of political sensitivity to the CJEU. In practice, this means that courts in Den-mark, Finland, Sweden or the UK are less inclined than courts from other member states to send cases with a high degree of political sensitivity to the CJEU.89

Model (7) also shows that the variable First instance remains positive and statistically significant (p<0.01). This result supports the alternative inter-court competition hypothesis (Alt H1a): courts at the lowest level of the le-gal hierarchy are more likely than higher courts to refer cases with a high degree of political sensitivity to the CJEU.90

The level of Euroscepticism, in contrast, is no longer statistically signifi-cant in Model (7). How can this result be understood? It can be theorized that a tradition of majoritarian democracy increases the degree of Euroscep-ticism in the member state. Indeed, regressing the variable Euroscepticism on the variable Majoritarian democracy shows a positive and significant relationship.91 This finding may explain why the variable Euroscepticism in Model (3) is no longer statistically significant when the majoritarian democ-racy variable is added to Model (7). The hypothesized explanation is that a tradition of majoritarian democracy in the member state decreases the likeli-hood that the national courts refer cases with a high degree of political sensi-tivity to the CJEU and, that a tradition of majoritarian democracy increases the level of Euroscepticism among the citizens. The result suggests that the negative bivariate relationship between the variable Euroscepticism and the referral of cases with a high degree of political sensitivity in Model (3) was spurious and driven by the member states’ traditions of majoritarian democ-racy. The Euroscepticism hypothesis (H2a) is thus not supported. Finally, Model (7) also shows that the variable Experience is statistically significant (p<0.1). However, because the odds ratio equals 1 there is no substantially interesting relationship between the variables. The experience hypothesis (H5a) is therefore not supported.

89 For a unit change in the variable Majoritarian democracy (from a tradition of constitutional democracy to a tradition of majoritarian democracy), the odds of referring a case that has a high degree of political sensitivity decrease by a factor of 0.44, holding all other variables constant. 90 For a unit change in the variable First instance (from a high court to a first instance court) the odds of referring a case that has a high degree of political sensitivity increase by a factor of 1.5, holding all other variables constant. 91 b: 0.12, SE: 0.02. This regression model can be found in Appendix B.

129

When summarizing the main empirical findings thus far, two factors mat-ter to the type of case that the national courts refer to the CJEU: the court’s place in the judicial hierarchy and a majoritarian tradition of organizing the relationship between law and politics in a member state. Courts of first in-stance are more likely than both courts of appeal and courts of final instance to refer cases with a high degree of political sensitivity to the CJEU. Moreo-ver, a tradition of majoritarian democracy is associated with a lower referral rate of cases with a high degree of political sensitivity when controlling for other relevant variables. This result means that the majoritarian democracy hypothesis (H3a) and the alternative version of the inter-court competition hypothesis (Alt.H1a) are supported. The hypotheses concerning national court experience, concrete judicial review and the level of Euroscepticism among the public were not supported.

However, one caveat must be added. Because the data set only includes information on the cases that actually have been referred to the CJEU, we lack information on what factors may be associated with a court’s decision to not refer a case with a high degree of political sensitivity to the CJEU.92 We simply do not know how many cases with a high degree of political sen-sitivity the national courts withhold from the CJEU since accurate data are difficult to obtain. Despite this limitation, this analysis provides an important first examination of which factors are positively or negatively associated with the referral of cases with a high or low degree of political sensitivity. To examine whether these results are robust, several different models have been tested; however, none of these tests challenge the above presented find-ings.93

Analysis II: Explaining the variation in the types of opinions expressed The second analysis centers on the national courts’ choices regarding what type of opinion to express in the request for a preliminary ruling. The binary dependent variable Type of opinion is coded 1 for opinions in support of national legislation and 0 for opinions in support of EU law.94 In the sample of 470 observations, these two types of opinions have been found in 182 cases. Each of the hypothesized relationships is first estimated in bivariate

92 This type of data is not readily available because of variations in how different courts report what types of cases they decide. However, for an attempt to assess this matter, see: Hübner, Denise Carolin (2016) The ‘National Decisions’ Database (Dec. Nat): Introducing a Database on National Courts’ Interactions with European Law. European Union Politics, 17, 324-339. 93 As an example of these tests, excluding the observations from German courts did not change the results. Results from the robustness checks are available in Appendix B. 94 The cases in which national courts express support for any other position are not included in the analysis.

130

logistic regression models. Then, all hypotheses are tested in two multiple logistic regression models. The first of these models includes the variable Final instance (the original inter-court competition hypothesis, Orig. H1b) and the second model includes the variable First instance (the alternative inter-court competition hypothesis, Alt H1b).

The results from the logistic regression analyses are presented in Table 5.3 below.

Table 5.3. Support for national legislation (logistic regression, odds ratios) (1) (2) (3) (4) (5) (6) (7) (8) (9) Final instance 2.58**

* 2.58**

*

(0.74) (0.87) First instance 0.49* 0.48* (0.19) (0.19) Euroscepticism 1.21 0.28 0.25 (2.02) (0.68) (0.55) Majoritarian democracy 1.12 1.40 1.03 (0.31) (0.74) (0.42) High degree of pol. sensitivity

0.83

0.85 0.81

(0.30) (0.31) (0.32) Experience 1.00 1.00 1.00 (0.00) (0.00) (0.00) Concrete judicial review 0.82 1.05 0.97 (0.18) (0.22) (0.17) Constant 0.38**

* 0.72* 0.53** 0.53**

* 0.61* 0.67** 0.63** 0.51 1.27

(0.05) (0.13) (0.15) (0.06) (0.16) (0.14) (0.12) (0.24) (0.61) Prob > chi2 0.0011 0.0598 0.9095 0.6755 0.6085 0.3680 0.1095 0.0079 0.0352 Pseudo R2 0.0367 0.0213 3.30e-

05 0.000278

0.00141

0.00475

0.00129

0.0405 0.0277

Countries 18 18 18 18 18 18 18 18 18 Observations 182 182 182 182 182 182 182 182 182

Robust standard errors in parentheses (clustered at the member state level) *** p<0.01, ** p<0.05, * p<0.1

Models (1) to (7) in Table 5.3 report the bivariate relationship between each of the independent variables and the dependent variable Type of opinion. As shown in the table, the variable Final instance in Model (1) is positive and statistically significant (p<0.01). This variable remains significant when controlling for the other variables in Model (8).95 This result is consistent

95 For a unit change in the variable Final instance (from a lower court to a court of final in-stance), the odds of expressing an opinion in support of national law increase by a factor of 2.58, holding all other variables constant.

131

with the original inter-court competition hypothesis (Orig. H1a) that expects courts of final instance to be more likely than lower courts to express support for national legislation. The other variable that is statistically significant, albeit at a lower level (p<0.1), is First instance in Model (2). The detected negative relationship holds when controlling for the other variables in Model (9).96 This finding corroborates the alternative inter-court competition hy-pothesis (Alt. H1a): courts of first instance are less likely than all higher courts to express support for national law. None of the other five hypotheses are supported.

The results that are reported in Table 5.3 thus provide empirical support for both the original inter-court competition hypothesis and the alternative inter-court competition hypothesis. How is this finding to be understood? To begin with, courts of final instance are more likely than all lower courts to argue in favor of national legislation in the request for preliminary rulings. By drawing on the judicial empowerment approach, this finding is explained by the fact that the highest courts stand to lose the most from further EU legal integration in terms of power over the development of the law. There-fore, courts of final instance argue in favor of preserving national law. Moreover, courts of first instance are more likely than all higher courts to express support for EU law. The lowest national courts are assumed to expe-rience empowerment by sending cases to the CJEU and thereby circumvent-ing the highest domestic courts. Because this empowerment is made possible by the EU legal norms, the lower national courts are expected to be more likely to express support for EU law.

However, what about the behavior of courts of appeal? In regard to the types of opinions that courts express, there is no significant difference be-tween courts of appeal and the other courts.97 Accordingly, courts of appeal are not less likely than courts of final instance to express support for national law. Similarly, courts of appeal are not more likely than first instance courts to express support for national law. These findings suggest that the main difference in behavior can be found when comparing the choices that are made by courts of final instance and the choices that are made by courts of first instance. In terms of inter-court relations in EU legal integration, this result means that the key competition occurs between the courts at the top and the bottom of the domestic legal hierarchy.

96For a unit change in the variable First instance (from a higher court to a court of first in-stance), the odds of expressing an opinion in support of national law decrease by a factor of 0.48, holding all other variables constant. 97 These models can be found in Appendix B.

132

Summary of chapter What explains the variations in national court behavior in the preliminary ruling procedure? Concerning the first key choice the member states’ tradi-tions of majoritarian democracy, characterized by a strong tradition of par-liamentary sovereignty, turned out to have a negative association with the national courts referral of cases with a high degree of political sensitivity to the CJEU. The national courts from Finland, Denmark, Sweden and the UK are found to be less inclined than other courts to provide the CJEU with op-portunities to foster EU legal integration by referring politically sensitive cases. This result is consistent with the majoritarian democracy hypothesis that was derived from the sustained resistance approach.

Moreover, the position of the court in the national legal hierarchy plays a role for both the type of cases that national courts refer to the CJEU and the type of opinions that they decide to include in the requests for preliminary rulings. In particular, this inter-court competition is mainly expressed when examining the relationship between the courts at the top and at the bottom of the domestic legal hierarchy. Courts of first instance clearly make choices that support EU legal integration, that is, courts of first instance are more likely than all other courts to allow the CJEU access to cases with a high degree of political sensitivity. In addition, courts of first instance are also more likely than the highest courts to express support for EU law.

This difference in behavior between low and high courts is what the inter-court competition hypothesis expects us to find. Courts of first instance be-come judicially empowered through the preliminary ruling procedure and thus have an incentive to open the gate and allow EU law to enter the domes-tic legal system. In contrast, because the preliminary ruling procedure partly deprives the high courts of their powers, the high courts are instead reluctant to allow the CJEU to influence the domestic legal landscape. In general, courts of first instance make choices that facilitate further EU legal integra-tion, while the courts of final instance make choices that defend national sovereignty. Furthermore, the share of unexplained variation in the models98 indicates that there may be other factors that can account for the variations in the courts’ choices that have not yet been identified. Thus, there is a need for further theorizing on the factors that may influence the national courts’ choices in the preliminary ruling procedure.

98 The Pseudo R2 in the models is at best approximately 0.04 (on a scale 0-1), which indicates that there is variation that is unaccounted for.

133

6. The behavior of individual national judges in the preliminary ruling procedure

This thesis has so far provided a unique and comprehensive picture of the national courts’ aggregated behavioral patterns in the preliminary ruling procedure. However, in order to fulfill the overall aim of the thesis one more question must be answered and it concerns the behavior of individual nation-al judges. How national judges make decisions has important implications for our understanding of the driving forces of European legal integration. If we do not pay attention to how the judges’ themselves experience the pre-liminary ruling procedure we run the risk of misinterpreting the aggregated behavioral patterns and how the national judges take on their responsibility as gatekeepers in EU legal integration.

Previous research on European integration often portrays national courts and judges as rational actors, primarily interested in increasing their own powers or in defending national law and sovereignty (Burley and Mattli 1993; Carrubba and Murrah 2005; Conant 2013; Golub 1996; Nyikos 2006). Yet, few studies have tried to corroborate these claims by studying how in-dividual judges are behaving. Whether, and to what extent, national judges are making choices in order to maximize such self-regarding preferences needs to be investigated empirically. First of all, it means finding out what substantive reasons national judges have for making different types of choic-es in the preliminary ruling procedure. The second important aspect is to examine what guides the judges’ mode of reasoning. Testing whether the theoretical assumptions of the self-regarding mode of reasoning are support-ed empirically provides a key piece to the puzzle of how we should under-stand national court behavior in EU legal integration.

The present chapter therefore sets out to analyze the micro foundations of the national courts’ macro behavior in EU legal integration. It does so by answering the third and final research question: What type of reasons are individual national judges guided by when making choices in the prelimi-nary ruling procedure?

134

Identifying the driving forces of judicial behavior In order to investigate the judges’ mode of reasoning a total of 20 semi-structured interviews with Swedish judges were carried out in 2016. As was discussed in chapter 3, the limited research on how individual judges make decisions in the preliminary ruling procedure means that we have no reason to believe that the Swedish judges’ mode of reasoning deviates systematical-ly from the average European judge. However, the Swedish case can also be viewed in light of previous research on US judicial decision-making. The dominant theoretical outlook in the US judicial politics literature assumes that considerations about expected outcomes drive judicial behavior (Epstein et al. 2013; Epstein and Lindquist 2017; Segal and Spaeth 2002; Spiller and Gely 2008). Research on EU legal integration is heavily inspired by the US judicial politics literature and judges have been characterized as rational actors interested in increasing their own power and influence vis-à-vis the political branches and other judges (Alter 2001; Burley and Mattli 1993), that is, judges are expected to be guided by self-regarding preferences. Given the differences between Sweden and the US in terms of institutional and political tradition, the Swedish judiciary can be understood as a critical, least likely, case for exhibiting a self-regarding mode of reasoning. If the analysis shows that the Swedish judges in fact strive to promote self-regarding pref-erences, it can be argued that this assumption is likely to hold true in other European judiciaries that are more similar to the US judicial branch.

The respondents have been randomly selected from the three different court levels: courts of first instance, courts of appeal, and courts of final instance. Importantly, the selection of respondents from each court level was made from a list of judges who had experience with requesting a preliminary ruling from the CJEU in at least one case.99 The full theoretical motivation for interviewing Swedish judges and the details about the selection process of the respondents have been elaborated on in chapter 3.100

The purpose of the interviews was to find out how the judges are reason-ing in the preliminary ruling procedure, which means that the empirical fo-cus is slightly different in this chapter compared to chapters 4 and 5. The analysis still centers on the national courts’ choices in the procedure but on a general level. The questions are not about how the judges are reasoning when deciding whether or not to refer cases with a high degree of political sensitivity, or whether or not to include opinions in support of EU law/national law in the request. Instead, the focus is on how judges make decisions when confronted with EU law cases: should they be referred to the CJEU or not and, should an opinion be included in the request or not. The analysis of those considerations is carried out with the help of the four theo- 99 Around 120 Swedish judges have requested at least one preliminary ruling. 100 Further information about the respondents and the interview guide can be found in Appen-dix C.

135

retical modes of reasoning which were elaborated on in chapter 2: conse-quentialism based on self-regarding preferences or other-regarding prefer-ences and, appropriateness based on formal codified rules or informal pro-fessional practices. The next sections outline the operationalization of these four different modes of reasoning.

The dominant mode of reasoning: Self-regarding preferences When interviewing judges about the choices they make in the preliminary ruling procedure, how is a self-regarding mode of reasoning identified em-pirically? First of all, national judges are expected to describe how they have considered which decision is best for achieving a particular outcome. Se-cond, what is important is that the outcome in question benefits the judges themselves. That is, how the judge thinks the decision affects his or her per-sonal situation and well-being.

There are plenty of examples in previous research of what constitutes such self-regarding preferences. It has been theorized and empirically demonstrated that judges want to advance their careers and be respected in the legal community (Baum 1997: 17; Dunoff and Pollack 2017: 241; Heumann 1978; Miceli and Coşgel 1994; Schauer 2000). This includes avoiding criticism from colleagues, other courts, and political actors (Epstein et al. 2013: 34). Others researchers have conceptualized policy preferences as a type of self-regarding preference (Landes and Posner 1975: 887; Posner 1993; Segal and Spaeth 2002). Moreover, in theories on EU legal integration it is commonly assumed that national judges want to increase their own in-fluence over legal and political issues. European integration and the prelimi-nary ruling procedure are in this research understood as means for national judges to increase their powers in relation to other courts or other branches of government (Alter 2001; Burley and Mattli 1993; Weiler 1994: 523).

In a stylized example of a mode of reasoning based on self-regarding preferences the judges motivate their decision to refer cases to the CJEU with reference to how it increases their powers vis-à-vis other courts. An empirical example can be found in Alter’s interviews with lower tax court judges in Germany in 1994. They said that they looked for cases to refer to the CJEU in order to challenge the doctrine of the higher Federal Tax Court (Bundesfinanzhof) (Alter 2001: 100, note 74). Self-regarding preferences may theoretically include many different things other than the ones present-ed in the preceding section. The key point is that they should benefit the judges themselves. In other words, the judges’ reasons for making different decisions are personal.

136

The challenge from reasoning based on other-regarding preferences Judges who are guided by other-regard are also expected to motivate their decisions with reference to intended consequences. However, in contrast to the mode of reasoning based on self-regard, other-regarding considerations center on how particular decisions might benefit other actors (Dunoff and Pollack 2017; Fehr and Fischbacher 2002; Fowler and Kam 2007; Mansbridge 1990; Stout 2001). Applied to judicial behavior, this means that judges are assumed to have preferences over outcomes that go beyond their own self-interest. In practice, it is about identifying what the judges want to achieve and why. Are their desired outcomes based on other-regard or not?

Previous research on other-regard expects that judges think about the de-sirability of different decisions for both individuals and institutions. The judges may care about how the outcome of their decisions affects individu-als, societal groups such as minorities, or the implications for society at large (Baum 1997: 58; Fowler and Kam 2007: 813). The judges’ identification with the parties to the case qualifies as being other-regarding (Burton 1992:45). Caring about one’s colleagues (Fehr and Fischbacher 2002: 2) in or outside the court also falls within the scope of other-regarding prefer-ences. The judges’ other-regard may also be directed towards institutions. The literature on national court behavior in the EU context has found that judges try to protect member state sovereignty and policies (Golub 1996: 381) and the coherence of the domestic legal system (Dehousse 1998: 173) from disruptive influence of EU law. It suggests that the judges’ reasons for making different decisions can be either legal (to care for the legal system) or political (to care for the member state’s sovereignty).

In a stylized example of a mode of reasoning based on other-regarding preferences, the judges motivate their decision to send cases to the CJEU with reference to how it benefits the parties to the cases.

The challenge from reasoning based on formal codified rules Theoretically it is possible to distinguish between two modes of reasoning within the overarching behavioral logic of appropriateness: one in which the judges’ considerations are shaped by formal codified rules and one in which they are shaped by informal professional practices. Within both these modes of reasoning the judges’ decisions are not the results of calculations about how to achieve specific outcomes. Instead, when faced with various alterna-tives, an actor’s behavior will be decided by the rules that govern the appro-priate course of action for a given role (March and Olsen 1989; March and Olsen 2009). When trying to find out what the appropriate course of action is, the judges think about what is expected of them, as judges, in a particular situation. For example, what constitutes their responsibilities and duties in

137

the preliminary ruling procedure? The judges find the answer to these ques-tions in the different types of institutional rules that they have internalized. Whatever those rules prescribe constitutes the appropriate decision.

In the case that the judges’ mode of reasoning is based on formal codified rules, the judges are expected to reflect upon what those rules prescribe as being the responsibilities that come with being a judge in a particular situa-tion. While there is an abundance of formal codified rules, the definition used here includes written, publicly known rules such as legislative acts and regulations. The most relevant ones include Treaty Article 267 (TFEU), the CILFIT-criteria, and the “Information note on the preliminary ruling proce-dure” that together regulate the preliminary ruling procedure. Other types of formal codified rules include domestic legislation, regulations of domestic legal proceedings, constitutional law, and different types of national legal rules that control the judges’ professional role.

In a stylized example of a mode of reasoning based on formal codified rules, the judges describe their decision to refer cases to the CJEU with ref-erence to the requirements in Article 267 TFEU. The consideration should be related to what the judges perceive as being the appropriate decision based on codified rules, not a calculation of which decision renders the best outcome for themselves or other actors. That is, the judges’ reasons are legal rather than personal or political.

The challenge from reasoning based on informal professional practices Apart from formal codified rules, judges can also base their considerations about what constitutes the appropriate decision on informal professional practices. This includes factors such as shared beliefs about what obligations actors have (March and Olsen 1989: 24), traditions of professional judicial conduct, as well as abstract legal principles, for example, the principle that judges are supposed to act in a way that is appropriate in relation to the in-tegrity and independence of the judiciary (Clayton 1999:15). Informal prac-tices can also include different types of shared understandings within a group of colleagues about what is the appropriate, standard operating procedure in different situations. For instance, in the US context one finds the norm of secrecy of deliberation prior to the announcement of decisions (O´Brien 1999: 91). The judges are expected to consider what the informal profes-sional practices prescribe as being the responsibilities that come with being a judge in a particular situation.

In a stylized example of the mode of reasoning based on informal profes-sional practices, the judges are expected to describe the decision to not refer a case to the CJEU with reference to this being inappropriate given their understanding of what the informal professional practices prescribe. For

138

example, an informal professional practice may be an agreement among the judges to show restraint in the number of cases referred to the CJEU. This may also be labeled a legal reason rather than a personal or political reason.

Exploring judicial reasoning in the preliminary ruling procedure The empirical analysis of how national judges behave in the preliminary ruling procedure is organized around what choices they make in the prelimi-nary ruling procedure. First, how they decide whether a case is to be referred to the CJEU or not, and second, how they decide whether or not to express opinions in the requests for preliminary rulings. To show which quote be-longs to what respondent, each respondent have been assigned a number between 1 and 20. The original quote in Swedish can be found in the foot-notes. The judges’ answers are analyzed with the help of the empirical indi-cators derived from the four different theoretical understandings of what drives judicial behavior, summarized in Table 6.1 below.

Table 6.1. Four modes of reasoning

The logic of consequentialism The logic of appropriateness Self-regarding prefe-rences

Other-regarding preferences

Formal codified rules

Norms and practices

Reputation, career, leisure, increase one’s powers (judi-cial empowerment), Promote one’s pre-ferred policies (sus-tained resistance)

To care for the well-being of other individuals, groups or institutions

Article 267 TFEU, Infor-mation note on the preliminary ruling procedure, various codified rules at the do-mestic level

Norms and practices about appro-priate con-duct and what it means to be a judge

The empirical analysis has been carried out by a careful reading of the tran-scripts from the interviews. Different questions have been posed to the mate-rial. For example, to find out whether the judges’ reasoning can be under-stood as self-regarding or other-regarding, the question asked was: Do the judges say that they want to achieve something in particular with their deci-sion? If they do, is it expected to benefit them personally (self-regarding) or someone else (other-regarding)? Conversely, if it is not possible to discern that the respondents make considerations about how to achieve a particular outcome, what do their considerations center upon instead? Are the judges thinking about how they should behave and is their decision-making based

139

on considerations about what is the appropriate behavior given their respon-sibilities? If so, is it formal codified rules that prescribe what is appropriate? Or, is it rather informal professional practices? This approach has also made it possible to identify what type of substantive reasons judges have for mak-ing different decisions in the preliminary ruling procedure.

The first decision: Whether or not to refer cases to the CJEU Why are national courts requesting preliminary rulings from the CJEU and why are they not? This first part of the analysis focuses on what the judges consider when deciding whether to either refer a case or to withhold it.101 The formal EU rules regulating the preliminary ruling procedure can be found in Article 267 TFEU, which states that:

Where such a question is raised before any court or tribunal of a Member State, that court or tribunal may, if it considers that a decision on the question is necessary to enable it to give judgment, request the Court to give a ruling thereon.

Where any such question is raised in a case pending before a court or tribunal of a Member State against whose decisions there is no judicial remedy under national law, that court or tribunal shall bring the matter before the Court.

The wording of the Article signifies that national courts of first instance and courts of appeal are not obliged to refer cases but can do so if they want to. Courts of final instance have less room to maneuver. According to the CJEU’s case law, a national court of final instance is obliged to refer cases to the CJEU unless:

it has established that the question raised is irrelevant or that the community provision in question has already been interpreted by the court or that the cor-rect application of community law is so obvious as to leave no scope for any reasonable doubt. The existence of such a possibility must be assessed in the light of the specific characteristics of community law, the particular difficul-ties to which its interpretation gives rise and the risk of divergences in judi-cial decisions within the community (Judgment of the Court of Justice 6 Oc-tober 1982, Srl CILFIT and Lanificio di Gavardo SpA v Ministry of Health, C-283/81, EU:C:1982:335, paragraph 21.)

101 The respondents were not asked about why they decided to either refer or withhold cases with a high degree of political sensitivity. Instead I asked them to describe why they decided to refer a case to the CJEU (all respondents had referred at least one case) and why in other cases they had decided against it. The reason for this choice was that I wanted them to freely elaborate on their decisions. Further information about the interview questions and the design of the interview guide can be found in chapter 3 and Appendix C.

140

As the CJEU writes in the above quote from the CILFIT-case, the courts of final instance can refrain from referring cases if the interpretation of EU law is clear from the CJEU’s previous judgments or that the EU provision in itself is unambiguous.

Reason 1: Referring cases because it is required Each of the interviewed judges has been involved in requesting at least one preliminary ruling from the CJEU. At the beginning of the interviews, all of the respondents described their decisions to refer cases as a simple matter of fact. They said that they had come to understand that the EU law on a partic-ular issue was unclear and that a referral to the CJEU was necessary for re-solving the case at hand. They also made explicit references to the formal EU rules found in Treaty Article 267 TFEU and the CILFIT-criteria.102 The following quote illustrates the respondents’ view:

If the legal case is about the interpretation of an EU legal rule and the inter-pretation is unclear then you just have to ask.103 (19)

This type of answer is not surprising given the formal EU rules; the judges are required to formulate the request to the CJEU by explicitly stating what they find unclear about the EU law. Thus far, it appeared as if the judges mainly consider what type of action the formal rules prescribe.

However, what is more interesting and requires thorough analysis is what the judges said next. When asked to elaborate on their decision to either refer or withhold cases, 13 out of the 20104 interviewees brought up considerations that cannot be traced to the codified rules. A total of four different reasons can be identified. One relates to why judges decide to send politically sensi-tive cases to the CJEU while the remaining three concern why the judges chose to withhold cases from the CJEU. The following sections analyze how the judges’ considerations can be understood with the help of the theoretical descriptions of different modes of reasoning.

Reason 2: A strategy of blame avoidance Two of the interviewees elaborate on why they decided to refer cases that they considered to be politically sensitive. They are not only motivating their decisions to request a preliminary ruling from the CJEU by referring to the

102 This was the first official question I asked the respondents and my impression is that the respondents, only knowing that I was a political scientist writing a dissertation on the topic of the preliminary ruling procedure, thought it important to inform me about the formal rules. After making this clear, most of them proceeded to elaborate on different aspects of their work and decisions. 103 ’Om målet handlar om tolkning av EU-rättslig regel och tolkningen är oklar, då är det bara att fråga.’ (19) 104 The remaining seven respondents did not express that they made any considerations apart from the ones based on the formal codified rules.

141

formal requirements in Article 267 TFEU, i.e., that the EU law was unclear. Instead, the respondents say that they expected negative consequences for not referring the case. In a case described by one of the respondents, the national court suspected that a domestic policy was incompatible with EU law. It also anticipated that the Swedish government would face high eco-nomic costs if the domestic policy were to be overruled in court:

The amount of money we're talking about, then one has to, so to speak, really know what one is talking about when deciding the case (...) yes, for example if we would have said that it is obvious that this is covered by EU law (...) then it would of course have led to a lot of turmoil in the system, then all X would have requested refunds retroactively and then the government would have had to yield and pay—I can’t even imagine how much money it would be, but it would be around 100 billion [SEK]. (…) We did not want to put ourselves in that position—and then the question is, should we, out of con-venience, because we are so afraid of the consequences of such an outcome, more or less let ourselves be steered into the opposite direction [i.e. as the tax authorities did and tax them] when we did not feel we were 100 percent sure it was the right decision and that was where we landed, and then we said that we would like to request a preliminary ruling.105 (5)

From this statement we learn that the anticipated consequences of setting aside a piece of domestic legislation are so severe, in terms of economic costs for the government, that the national court refrains from doing so with-out having a preliminary ruling from the CJEU to back it up. The national judges do not want it to appear as if they are causing the government such costs.

The other respondent is even more explicit about the intention. This judge describes it as being a sensitive situation when domestic law stands in con-flict with the supreme EU law. Requesting a preliminary ruling from the CJEU is then a way for the national court to avoid the blame for setting aside Swedish legislation:

105 ’Det var sådana pengar vi pratar om och då får man väl ändå så att säga se till att man har bra på fötterna när man dömer i målet. (…) ja, om vi t.ex. skulle ha dömt och sagt att det är helt självklart att det här omfattas av EU-rätten (…) då hade det ju blivit ett väldigt rabal-der i systemet naturligtvis då hade ju alla x kommit in med en begäran om återbetalningar retroaktivt och då hade ju regeringen haft att foga sig i det och varit tvungen att betala och det hade ju, jag kan inte ens föreställa mig, hur mycket pengar det skulle bli, men det skulle bli 100tals miljarder… ja, vi ville ju inte sätta oss i den sitsen och då är frågan, ska vi då av något sorts bekvämlighetsskäl, för att vi är så rädda för konsekvenserna i en sån här utgång, mer eller mindre styras in på den andra utgången [göra som skatteverket gjort och bara beskatta dem i största allmänhet] när vi ändå inte kände att vi var 100 hemma på det och det var väl ungefär där vi landade, och då sa vi att vi tar väl in ett förhandsbesked.’ (5)

142

It's later, when you get into more sensitive situations, where you maybe need that… that someone else says it. In our case, for example, that it was a scape-goat function really, very clear, one could say. (Q: How do you mean?) It's not actually we who are setting aside Swedish law, but it's actually the Euro-pean Court of Justice that says that we have to do it—“Look, our hands are tied” – it's a pretty good feature [of the EU legal system] and it's an important and deliberate role that the European Court of Justice has in the Union's ar-chitecture.106 (9)

In this quote the interviewee explicitly states that the CJEU is used as a scapegoat by the national judges. The CJEU takes the blame for unpopular decisions such as overturning national policies.

The phenomenon that actors try to blame others in order to escape re-sponsibility for decisions they make is well known in previous research (Hood 2011; Weaver 1986: 386-387). Such attempts at blame avoidance are considered to be based on the logic of consequentialism (Novak 2013: 1101). For instance, member state executives may try to shift the blame for unpopular decisions taken at the EU level from themselves and onto the Commission (Moravcsik 1993: 515-516; Schmidt 1998: 176). Similarly, it has been theorized that national courts have requested rulings from the CJEU to deflect criticism from themselves in controversial cases (Alter 2001: 50, note 21; Maher 1994: 229).

Turning to how the quotes can be understood theoretically, the most ap-parent feature of the judges’ considerations is that of consequentialism. The respondents describe how they chose to refer the politically sensitive cases to escape criticism for setting aside national legislation. It is important to note that the judges did nothing wrong when deciding to refer the cases. The cas-es did concern the interpretation of EU law and referring them to the CJEU falls well within Article 267. In other words, the judges did not disregard any formal rules. But the interesting finding is that the judges also considered which decision would lead to the best outcome for them.

The judges appear to have been convinced that the national policies in those particular cases were in fact incompatible with EU law. But overturn-ing national policies without having the support of a preliminary ruling from the CJEU leaves them vulnerable to critique. The judges’ preferred outcome is based on considerations for their own well-being. It is about avoiding be-ing blamed by the member state government for invalidating national poli-cies. This line of reasoning corroborates the expectations in the literature that the judges’ strive to safeguard their own position, a self-regarding prefer-

106 ’Det är sen när man börjar komma i känsligare lägen, där man behöver, kanske någon annan som säger det, i vårt mål t.ex. där var det ju scape-goat funktionen verkligen, väldigt tydlig får man säga, (F: Hur menar du?) det är faktiskt inte vi som åsidosätter svensk lag utan det är faktiskt EU-domstolen som säger att vi är tvungna att göra det ”look our hands are tied” – det är en ganska bra funktion och det är en viktigt och avsiktlig roll som ju EU-domstolen har i unionens arkitektur.’ (9)

143

ence. In this case national judges try to deflect potential criticism by refer-ring cases to the CJEU.

Reason 3: Uphold the functioning of the legal system The CJEU’s limited resources and the anticipated negative consequences for the parties are the reasons that make half of all respondents (10) say that they do not refer all cases in which the interpretation of EU law is unclear. First, five107 of the judges say that they do not strictly follow the formal codified rules out of concern for the capacity of the preliminary ruling procedure. Their reasoning can be illustrated by the following three quotes:

Indeed, the resources of the European Court of Justice are very limited and I think that one should probably think twice before using this opportunity to ask questions.108 (19)

I can say that in general, over the years, there have been so many times that we have interpreted the EU law in different ways and it is obvious that on many of those occasions we did not think it's all that clear, but if we should have sent all those cases to the European Court of Justice, we would be overwhelmed by requests for preliminary rulings, it is not possible. We inter-pret our domestic law, so to speak, without being completely sure in all situa-tions (…). To put it simply, one has to carefully consider when it really feels like it [a request for a preliminary ruling] is needed.109 (6)

But if all the courts of final instance were to ask questions, which they are obliged to do as soon as doubts arise, then the European Court of Justice would be drowning in requests for preliminary rulings so that it is perfectly clear that all courts of final instance do not ask all the questions that they should, (...) I would guess that… maybe they are thinking, “Well, that's not unclear enough”.110 (2)

The three quotes are illustrative of the five respondents’ view that national judges cannot request preliminary rulings in all cases in which the interpreta-tion of EU law is unclear due to the limited capacity of the CJEU. The inter-

107 Respondents: 2, 6, 14, 18, 19. 108 ‘EU-domstolens resurser är ju faktiskt väldigt begränsade och att jag tycker att, man ska nog tänka sig för innan man utnyttjar den här möjligheten att få ställa frågor’ (19) 109 ‘Ja, kan ju säga rent allmänt alltså att under årens lopp är det ju så många gånger som vi fått tolka EU-rätt på olika sätt och det är klart att det är många gånger vi tyckt att det inte är alldeles klart, men skulle man ha skickat allt det till EU-domstolen då skulle vi bli överlupna av begäran av förhandsavgöranden, det går inte, utan det är så också att vi tolkar ju inhemsk rätt utan att vara helt säkra i alla lägen (…) man måste ju noga överväga när det verkligen känns som att det [förhandsbesked] behövs helt enkelt.’ (6) 110’Men om alla slutinstanser skulle ställa frågor, vilket de är skyldiga att göra så fort det uppstår en tveksamhet, då skulle ju EU-domstolen drunkna i frågor så att det är väl rätt uppenbart i och för sig att alla slutinstanser inte ställer alla frågor som de borde, (…) jag skulle väl gissa att det är… möjligtvis resonerar man att ”nja, det är inte tillräckligt oklart”’ (2)

144

viewees are well aware of the formal rules. Yet, they express that it would be impossible to follow them. The respondents think that if they would strictly adhere to the wordings of Article 267 TFEU and the CILFIT- criteria the CJEU would be overburdened with insignificant cases. The judges express that national courts need to show restraint in what types of cases they refer in order for the CJEU to have the capacity to answer the most essential ques-tions concerning the application of EU law.

The other five111 respondents (out of 10) bring up a different aspect of the problems that may arise from a strict adherence to the formal rules. They think that requesting a preliminary ruling may negatively affect the parties to the case since it takes around two years for the CJEU to answer the ques-tions. The following quotes from two of the judges illustrate this view:

When the EU law is unclear, it has a tendency to happen in cases where things may be less important, and then you have to think about the costs and what it actually means for the parties if we put the case on hold for two to three years. So it's probably not customary to send questions on any trifles to the European Court of Justice, it must be something of importance.112 (5)

That has varied a lot, but I saw in one case, at least when [Court X] had de-cided to request a preliminary ruling, it took 2 years. Sometimes the pro-cessing time down in Luxembourg has been 2.5 years. [Q: That is quite a while?] Yes, it's not good at all; it's worsening the legal certainty and other things.113 (15)

The judges think that prolonging the legal procedure may result in economic costs and other negative consequences for parties. The respondents' concern for the parties’ situation relates to the principle that “justice delayed is jus-tice denied.” The time it takes to resolve a legal dispute is crucial for wheth-er or not citizens consider the justice system to be fair (Sourdin and Burstyner 2014). Waiting more than two years for an issue to be resolved is likely to contribute to a person’s dissatisfaction with the effectiveness of the legal system. If carried to an extreme, it may weaken the rule of law and the legal certainty.

How can the judges’ considerations be understood theoretically? The re-spondents all think that the action prescribed by the EU rules gives rise to

111 Respondent 1, 4, 5, 10, 15. 112 ’När det är oklart så har det en tendens att hända i mål där saker kan ha mindre betydelse, och då får du ju tänka på kostnaderna och var det rent faktiskt innebär för parter och omgiv-ning om vi stoppar målet i två till tre år och drar igång, alltså, det är nog inte kutym att man skickar frågor på vilka bagateller som helst till EU-domstolen, det ska ju ändå vara någon-ting att komma med.’(5) 113 ’Det där har väl varierat men jag såg i något av fallen så var det i alla fall från det att HD hade beslutat om att begära ett förhandsavgörande till dom så tog det väl 2 år, åtminstone tidvis har väl handläggningstiden där nere i Luxemburg varit 2,5 år. (F: Ett tag?) Ja det är inte alls bra, det… det försämrar rättssäkerheten och allt möjligt.’ (15)

145

negative consequences for the preliminary ruling system, the parties, and, by extension, the legal system as a whole. As a result, the judges are withhold-ing so-called less important cases from the CJEU. Such considerations corre-spond to a mode of reasoning based on other-regard. The judges are thinking about what consequences different choices have for others, including both actors and institutions. What can be concluded from the respondents’ an-swers is that national judges need to show restraint in what cases they refer to the CJEU in order to uphold the functioning of the legal system.

Reason 4: Protect one’s reputation This category of responses is centered on the judges’ fear of making an in-correct assessment of the legal issues in the cases. Four114 respondents elabo-rate on what they think are the risks associated with accidently asking the “wrong” questions or referring the “wrong” cases to the CJEU. One of them expresses that:

Embarrassing! It's an embarrassment for the court to ask a question and then get a response that lets you know that you have completely misunderstood [the EU law].115 (13)

The interviewee finds it embarrassing for the domestic court to receive an answer to a request for a preliminary ruling in which the CJEU makes a re-mark about how they have misunderstood EU law. The critique from the CJEU is likely to be known also by the judges’ colleagues since it is includ-ed in the final ruling, available for everyone to read.116

The fear among judges of having their decisions reversed or criticized by high courts has been identified in previous research. Such critique may un-dermine the judges’ status, hurt their reputation in the legal community, and jeopardize their future career (Heumann 1978: 144; Schneider 2005: 131; Tridimas and Tridimas 2004; Weaver 1986: 376).

Another one of the respondents is also anxious about disturbing the CJEU with “irrelevant” questions:

114 Respondent 1, 2, 9, 13. 115 ‘Pinsamt! Pinsamt för domstolen är det, att ställa en fråga där man får ett svar tillbaka där man kan förstå att det där hade man missuppfattat fullständigt.’ (13) 116 Another one of the interviewees describes how judges think about their reputation among their peers and that other colleagues, even in other countries, are likely to read their request for preliminary rulings and evaluate them. It is therefore important to perform well: They had a lot of colleagues from other courts in Europe, in the highest instances, I guess that (...) they realized that all of their colleagues will read the request for a preliminary ruling that they were about to formulate (...) but it turned out well because the question we formulated was present-ed as one of the “best question” that the European Court of Justice had received at one of those judicial conference… and the worst question was something from the highest court in Greece. (2)

146

The great horror, in my opinion, of asking the European Court of Justice a question is if the answer you receive is a reasoned order because it means that “you could have read about it in this and this judgment.” It's like “okay, do something sensible with your time instead of bothering us with this.”117 (9)

This quote illustrates that the respondent interprets the CJEU’s decision to issue a reasoned order as a critique against the national judges for wasting its time and resources. A reasoned order signifies that the CJEU considers that the legal questions asked by the national court have already been answered in a previous ruling and that no new preliminary ruling is necessary.118 In plain text, it means that the national judges could have found out how to apply the EU law by reading the CJEU’s case law instead of requesting a preliminary ruling.

The answers from the respondents show that they think about how the re-ferral of the “wrong” type of cases reflects badly on themselves as judges. It forms part of a self-regarding mode of reasoning. The judges want to avoid embarrassing themselves and receiving critique from the CJEU.

Reason 5: Limit the expansion of EU legal integration The potential risks associated with requesting a preliminary ruling from the CJEU also feature in the response given by one other judge. However, this respondent has a different understanding of what the risk entails. According to the judge, requesting a preliminary ruling from the CJEU in the “wrong” case threatens the constitutional division of power between the member state and the EU:

117 ‘Den stora skräcken tycker jag med att ställa en fråga till EU-domstolen är om man skulle få tillbaka ett motiverat yttrande, för det innebär att ”det här hade du kunnat läsa i den här och den här domen” det är liksom ”okej, gör något vettigt med din tid istället för att störa oss med det här’ (9) 118See Article 99 of the Rules of Procedure of the Court of Justice: “Where a question referred to the Court for a preliminary ruling is identical to a question on which the Court has already ruled, where the reply to such a question may be clearly deduced from existing case-law or where the answer to the question referred for a preliminary ruling admits of no reasonable doubt, the Court may at any time, on a proposal from the Judge-Rapporteur and after hearing the Advocate General, decide to rule by reasoned order.” European Union, EU (2012) Rules of Procedure of the Court of Justice. Official Journal of the European Union.

147

It is important that you always request a preliminary ruling when you are supposed to, but it is also important that you do not do it when it is unneces-sary. It is important not to do it in an area where there is no EU legal compe-tence (...) What I think may be problematic is if you ask the European Court of Justice a question about something that it does not have the competence to decide or about something outside the realm of EU law. If a high court judge does that, it has taken the standpoint that this issue is an international legal matter, and that it falls within the scope of EU law. (…) it is an important constitutional question, and even more so when a court of final instance takes such a stance, since it is an explicit statement.119 (16)

The respondent thinks that a request for a preliminary ruling signifies explic-it support for the view that EU law governs the legal area that is present in the case. The respondent finds it to be problematic if judges refer “unneces-sary cases” because it enables the CJEU to expand the reach of EU law. This is especially true if it is one of the judges in the highest courts who make the decision since their words and deeds carry a lot of weight when it comes to shaping public policy and the legal development. For instance, the highest courts set precedents that influence the decision-making in both lower courts and in government agencies. This standpoint, that the CJEU uses preliminary rulings to push for further EU legal integration, is well documented in previ-ous research (Alter 1998a; Alter 2001; Pollack 2013: 1262-1265) and in statements from former CJEU judges (Mancini 1989; Pescatore 2015).

It is evident that the judge is considering the expected consequences of sending a case to the CJEU. What the respondent wants to avoid is that the member state loses its control over domestic policy areas. The national judg-es should therefore avoid referring cases that may result in a de facto transfer of decision-making power from the member state to the EU. This type of consideration can be understood as being other-regarding since the decision is intended to benefit the member state and the national legal system.

119 ‘Det är ju viktigt att man alltid inhämtar förhandsavgöranden när man ska, men det är också viktigt att man inte gör det i onödan. Det är viktigt att man inte gör det inom ett område där det inte finns EU-rättslig kompetens (…) det som jag tänkt kan bli problematiskt är om man ställer en fråga till EU-domstolen, om någonting där domstolen inte har kompetens eller inte, som inte styrs av EU-rätten och man arbetar i högsta instans, då har man ju givit uttryck för att den här frågan är internationellt styrd, den styrs av EU-rätten. Det är ju det besked vi ger om vi frågar EU-domstolen om någonting, att ”i det här hänseendet så är det här direkti-vet som reglerar hanteringen av den här frågan” och de frågeställningarna kan ju ofta komma upp liksom, ”det här direktivet reglerar det här området” och ”hur långt stäcker sig den regleringen” och ”vilket handlingsutrymme finns på nationell nivå” och ”vilket har EU-domstolen kontroll över”, det är väl den viktigaste frågan (…) den är ju konstitutionellt bety-delsefull, och den får väl större betydelse om man i högsta instans gör det, därför att då har man givit uttryck för ett ställningstagande.’ (16)

148

The second decision: Whether or not to include opinions in the references to the CJEU Why are national courts expressing opinions in their requests for preliminary rulings? Previous research describes opinions as strategic “rhetorical weap-ons.” (Nyikos 2006: 530) that the national courts use to influence the CJEU’s understanding of the legal questions that have been raised in the case. The ultimate goal is to influence the outcome of the legal case (Nyikos 2006: 531-532). However, what reasons national judges have for including opinions have not yet been investigated empirically. Only three of the respondents said that they have included opinions in the request for a preliminary ruling. Due to the small share of respondents who have made considerations about what type of opinion to express, it is diffi-cult to investigate how the judges reason about the content of the opinions.120 The aim of this analysis is instead to uncover the judges’ mode of reasoning when deciding whether to express an opinion or not.

The practice of including opinions is described in the formal EU guide-lines from the CJEU, which state that:

If it considers itself able to do so, the referring court or tribunal may, finally, briefly state its view on the answer to be given to the questions referred for a preliminary ruling (European Court of Justice 2005: paragraph 23).121

As these guidelines make clear, the national judges have full discretion whether they want to express opinions or not. Opinions are welcome but it is not a requirement for any judge to express them.

Four of the judges stated that they had not thought about including opin-ions in the request at all.122 However, the other seventeen respondents have reflected upon it and they have, as we will see, rather strong and divergent views on the practice of including opinions.

120 This means that the judges’ reasoning regarding the second key choice, the content of the opinions, is difficult to examine properly when so few respondents have actually expressed opinions. Therefore, the analysis centers on the national judges’ choice of whether to express opinions or not. 121 This Information Note on References from National Courts for a Preliminary Ruling (2005) has been updated. The latest version is: Recommendations to national courts and tri-bunals in relation to the initiation of preliminary ruling proceedings: “In so far as it is able to do so, the referring court or tribunal should also briefly state its view on the answer to be given to the questions referred for a preliminary ruling” (Court of Justice of the European Union (2016) Recommendations to National Courts and Tribunals, in Relation to the Initiation of Preliminary Ruling Proceedings. Official Journal of the European Union, 59. paragraph 36) 122 Respondents 7, 8, 11, 12

149

Reason 1: Expressing opinions because it is required According to three123 of the respondents, national judges should express opinions. The following quotes capture their standpoint:

I got the impression that the court [the European Court of Justice] wanted it. One can’t just ask a question but you must also do some basic work yourself and then tell them about what you think.124 (2)

You should still state the court’s preliminary understanding of how this EU legal issue should be assessed. Yes, in principle, it is your responsibility to do so.125 (9)

As the quotes show, the respondents say that national judges are basically required to state their view of the legal case. It is what the CJEU expects from the national judges and it is simply something that national judges ought to do when engaging in the preliminary ruling procedure. For these respondents, the practice of including opinions in the request for a prelimi-nary ruling has become the natural and appropriate thing to do.

The judges’ considerations center on what responsibilities national judges have in the preliminary ruling procedure according to the formal codified rules. But what is interesting to note is that based on the judges’ answers one can be led to believe that opinions are somehow mandatory. However, ac-cording to the written guidelines judges may state their view but are not re-quired to do so. Still, the respondents describe it as being almost compulso-ry. It suggests that the judges’ reading of the guidelines and their view on opinions have been shaped by other factors as well. Such factors may in-clude signals coming from the CJEU or from actors within the Swedish legal system about how the guidelines should be interpreted, effectively encourag-ing the use of opinions. There will be reason to return to this matter in the concluding part of the chapter.

Reason 2: Influence the course of EU law Another respondent agrees with the view that including opinions in the re-quests for preliminary rulings is something that national judges should do. However, this interviewee differs from the aforementioned judges when it comes to the reason why. The respondent thinks that judges working in the courts of final instance should state their view in order to influence the de-velopment of the EU law:

123 Respondent: 1, 2, 9 124 ‘Jag fick uppfattningen då i alla fall, att domstolen [EU-domstolen] ville ha det. Alltså att man kan inte bara rafsa iväg en fråga utan man måste också göra grundarbete själv och då tala om vad man tror.’ (2) 125 ‘Man ska ändå ange domstolens preliminära uppfattning av hur den här EU-rättsliga frågan bör bedömas. Ja, det är man ju skyldig att göra i princip.’ (9)

150

It has been discussed at a general meeting in Court X, namely that Court X should provide some suggestions so to say or views that can steer the work of the EU court and I thought that could well be done (…) We are the bigger ones, we are more used than other courts to crack hard cases and to see mat-ters from a broad perspective, there would be nothing wrong with that.126 (18)

According to this judge, the highest national courts and their judges have the competence to formulate opinions while other lower courts do not. In the quote the respondent describes high court judges as being well accustomed to resolving difficult cases, meaning that they can provide the CJEU with valuable expertise by informing it about their understanding of EU law.

The quote offers a clearly articulated statement about what the respondent sees as the desired consequences of including opinions: to steer the course of EU law. This makes it distinct from the reasoning of the three respondents who said that national judges should state their view because it is the ex-pected behavior. In contrast, this judge’s consideration is based on how to achieve a preferred outcome, to influence the CJEU.

How should the judge’s preferences be understood? One interpretation is that the preference is self-regarding. By influencing the CJEU’s decision and by extension EU legal development, the judge becomes more influential and powerful. This, in turn, may boost one’s career. While this is a theoretically plausible suggestion, the quote does not provide strong support for it. In-stead, the preferences that predominantly appear in the quote are related to the well-being of others. The purpose of expressing opinions is not so much about benefitting oneself; it is about aiding the CJEU in resolving the case at hand. On a more general level, it is about contributing to, and caring for, the development of EU law. It is therefore more reasonable to interpret the judge’s consideration as being other-regarding.

Reason 3: Protect one’s reputation Two127 of the interviewees say that they are in favor of including opinions in the requests for preliminary rulings. Yet, they have not done so themselves because they consider it to be too risky. One of them states that:

126 ‘Det har diskuterats lite grann på ett sånt där allmänt möte i domstol X nämligen om domstol X skulle komma med någon, några förslag så att säga, eller styrande synpunkter och det tyckte jag att det kunde man gott och väl göra (…) vi är ju större, vi är ju mera vana än andra domstolar att knäcka svåra fall och se saker och ting i stort, det vore väl inget fel med det.’ (18) 127 Respondents 3 and 4.

151

(Q: What about expressing opinions?) A: I did not have the resources and I did not have the skills so I would have thought it to be incredibly risky to start predicting how it should go, so I didn’t. But one can always, if one wants to, hide behind the shield of impartiality.128 (4)

The same judge further elaborates on what the potential risk consists of:

An important part is that you're simply afraid to seem stupid. [Q: of making mistakes?] Yes, you're afraid to write something that will be in letters of fire, especially when the European Court of Justice says that, “but how can you not realize that this is a completely impossible way to see it.”129 (4)

As the quote shows, the perceived risk of expressing opinions is that the national judges might be publicly criticized by the CJEU for having misin-terpreted EU law. Essentially, the judges’ decision to not express opinions builds on considerations of how to avoid appearing stupid in front of col-leagues and peers. The same type of consideration was made by respondents who were afraid of referring the “wrong” type of cases to the CJEU. This consideration forms part of a self-regarding mode of reasoning in which the judges’ think about how their actions will reflect back on themselves. Previ-ous research also discusses how lower court judges think about their deci-sions in light of the expected reactions from high courts. The judges’ fear is that the critical remarks from higher courts will have a negative effect on their future career and on their reputation within the legal community (Heumann 1978; Tridimas and Tridimas 2004: 135).

Another interesting finding that can be discerned from the quotes is that the respondents think that many national judges share this fear and do not dare to include opinions in their requests for preliminary rulings. But instead of admitting it, the respondent says that those other judges motivate their decision by saying that expressing opinions jeopardizes their impartiality. It suggests that the judges may be unwilling to admit that their reasoning is based on self-regarding reasons and therefore they refer to the norm of im-partiality when describing their decision-making. There is reason to return to this matter in the final analysis.

Reason 4: Safeguard the courts’ impartiality So far, the analysis has been about the considerations made by the respond-ents who think that opinions should be included for different reasons. How-

128 [Har du bifogat åsikter?] ‘Jag hade inte resurser och jag hade inte kompetens så jag skulle tyckt att det vore otroligt riskfyllt att börja förutspå hur det borde gå så att då gör man ju inte det, men man kan ju alltid, om man vill, gömma sig bakom den här opartiskhetens sköld.’ (4) 129 ’En minst lika stor del av varför det är på det här sättet är att man är helt enkelt rädd för att verkar dum [F: tolka fel?] ja, man är rädd att skriva någonting som sedan står i eldskrift, när EU domstolen säger att ”men fattar ni inte från början att det här är ett helt omöjligt sätt att se det”.’ (4)

152

ever, the majority of the respondents hold the opposite view. When asked to elaborate as to why they believe that national judges should not state their opinions, ten respondents said that doing so is incompatible with the role of courts in the Swedish political system. This conclusion rests on conceptions of what responsibilities the judges have in general and in relation to the CJEU.

According to six of these respondents130, expressing opinions is not some-thing that the national judges should do. One of the respondents points out that:

It's not the national courts’ responsibility to defend the implementation that has been done, but rather to say that: "this is what Swedish law looks like and that's what we're dealing with," but that the national court would assess whether the Swedish law has been correct or not, that feels a bit strange.131 (19)

Another judge refers to the division of competences between the CJEU and the national courts when explaining why national judges should not include opinions in their requests for a preliminary ruling:

I can’t remember that we have ever expressed our own opinion regarding the questions we ask; it feels a bit strange (...). I think about it as being a question of constitutional division of skills and we do not have the competence to take a position on that question. There is someone else who is supposed to do that and then it is up to that court [The CJEU] to make the decision. Then it is not very relevant what we [as a national court] think and I'm not sure why you should express your point of view? It may possibly be linked to that, if you have a little more of a political approach, then you may try to represent the interests of the state and express the government's wishes for a certain deci-sion, but it would be strange for me to do so. My understanding is that we [as national courts] have certain competence and the judges of the European Court of Justice have their competence and we are supposed to ask them questions and then we should not say "but we think it should be interpreted like this" before they answer.132 (16)

130 Respondent 5, 14, 15, 16, 17 and 19. Some of the remaining respondents views: (not men-tioned in the text): No… I think… the question is whether it's really appropriate to do because we are not a party to the case (...) we are not supposed to express our own opinion. (14). If the [national] court would write ‘so we think that this matter should be interpreted in this way’ then the [national] court would be placing itself above the European Court of Justice.” (17) 131 ‘Det är ju inte den nationella domstolens sak att försvara gjord implementering, utan den ska väl snarare att säga såhär att ”såhär ser svensk rätt ut och det är den vi ska förhålla oss till” men att den nationella domstolen skulle sitta och värdera om den svenska lagen har blivit korrekt eller inte, det känns lite främmande’ (19) 132 ’Alltså jag kan inte komma ihåg att vi, vi nånsin har givit uttryck för vad vi själva har för åsikt i den fråga vi ställer, det känns också lite främmande (…) jag tänkte väl mer som att det är en fråga om konstitutionell kompetensfördelning och då har vi inte kompetens att ta ställ-ning till den frågan utan det är någon annan som ska göra det och då får den domstolen göra det ställningstagandet och då är det inte så väldigt relevant vad vi tycker och då kanske det inte, då förstår jag inte varför man ska ge uttryck för en sådan synpunkt, det kan möjligen

153

This respondent thinks that a judge who includes opinions in the request for a preliminary ruling exceeds his or her powers. It is the CJEU who has the responsibility to interpret EU law, not the national judges. The respondent concludes that national judges should not express opinions because doing so comes into conflict with the division of power between the national courts and the CJEU. Furthermore, the judge says that the habit of stating one’s view is something that judges who “have a political approach” might do, meaning judges who try to safeguard the interest of their own government. This is a behavior that the respondents view as strange.

The following quote from the same respondent provides valuable infor-mation as to why opinions are seen as such a strange practice:

I was thinking, we have something called the lift, that a first instance court can send a legal matter concerning precedent to the Supreme Court and say: “we think this is difficult, we think you should take a stand on this matter.” It would then be quite strange for the first instance court to say: "We think you should decide the case in this way," that’s about the same thing.133 (16)

In this quote the judge compares the preliminary ruling procedure with a similar domestic procedure and the relationship between the Swedish Su-preme Court and lower courts. In this domestic procedure, lower court judg-es should not state their view when asking higher court judges for guidance. According to the respondent, the same “rule” applies in the relationship be-tween the national judges and the CJEU. The national judges are subordinat-ed to the CJEU and should therefore not forestall its work by stating their own view. It appears as if the respondent’s standpoint on the matter is based on considerations about how judges are expected to behave that cannot be traced to the formal codified guidelines. Instead, the reasoning builds on the judge’s understanding of informal practices within the judiciary.

The other four134 respondents describe that the main problem with having judges expressing opinions is that it clashes with the court’s role as an im-partial arbitrator. One of them states that:

hänga samman med det där att om man har ett lite mer politiskt förhållningssätt så försöker man företräda statens intressen och ge uttryck för statens önskemål om ett visst avgörande men det, det skulle vara främmande för mig och jag ser det som att här har vi kompetens och där har de kompetens och vi ska fråga dem om de här frågorna och då är det inte meningen att vi ska säga ”men vi tycker så här” innan de har svarat,’ (16) 133 ‘Men jag tänker vi har ju något som heter hiss-dispens, att en tingsrätt kan skicka upp en prejudikatfråga till högsta domstolen och säga såhär: vi tycker att det här är svårt, vi tycker att ni ska ta ställning till den [frågan]. Det skulle ju vara helt främmande för tingsrätten att säga såhär: “vi tycker att ni ska döma såhär” det är väl ungefär samma sak.’ (16) 134 Respondent 6, 10, 13, 20. One of the remaining respondents view: (not mentioned in the text): The court is not allowed to take a position on the matter because the legal process con-tinues. (…) you have to be impartial until the final decision has made, it's a basic judicial norm (…). (20)

154

In my experience, although I haven’t read all requests for preliminary rulings, but I do not feel that it is that way, and it should not be that way either (…) I perceive it as a party writing. We are not supposed to sit around and argue how we think that the European Court of Justice should interpret EU law. It’s like if we are some sort of party to the case and we are not. We do not represent Sweden; we are a Swedish court that is supposed to request prelim-inary rulings. There are those who represent Sweden, but that is the govern-ment.135 (6)

According to this respondent, opinions are something that the parties to the case express, not the court itself. The judge also underscores that national judges do not represent Sweden during the proceeding in any way. Another respondent reflects upon the potential inconsistency between the practice of including opinions and the task of judging:

Yes, I just thought about this thing about expressing an opinion, I felt now when you said it, that it feels very odd in some way. If you were to decide that case later, will you then be disqualified from judging? It would be an un-familiar practice for me to express an opinion on how I think the case should be decided before I actually decide the case.136 (10)

As the judge makes clear in this quote, expressing opinions stands in conflict with the judges’ main task, namely, to deliver legitimate rulings. The re-spondent identifies that the big problem with including opinions is that it may disqualify the judges from judging.

How can the ten judges’ considerations be understood theoretically? The first thing to note is how these respondents describe the expression of opin-ions in the request for preliminary rulings. The reoccurring theme in the judges’ responses is that opinions are “strange,” “odd,” and “unfamiliar” and, expressing them exceeds the national judges’ authority in the Swedish legal system. These descriptive terms suggest that the respondents do not perceive the stating of opinions as being part of their responsibilities in the preliminary ruling procedure; it is even considered inappropriate. In other words, expressing opinions does not form part of the judges’ identity in the preliminary ruling procedure.

From the statements we can also learn about what the respondents think is the natural or appropriate behavior for a judge: to be an impartial arbitrator 135 ‘Jag upplever inte, jag har ju inte läst alla begäran om förhandsavgöranden, nämen alltså jag upplever inte att det är så och det ska inte heller vara så (…) uppfattar jag närmast som någon slags partsinlaga, vi ska ju inte sitta och argumentera hur vi tycker att EU-domstolen ska tolka EU-rätten, då är det ungefär som vi vore någon slags part i EU-domstolen och det är vi inte, vi företräder inte landet Sverige, vi är en svensk domstol som begära förhandsav-göranden och det finns de som företräder Sverige, men det är ju regeringen.’ (6) 136 ’Ja, jag bara tänkte på det här med att uttrycka en åsikt, jag kände nu när du sa det att det känns ju väldigt märkligt på något sätt, om man sen ska döma i målet, om man sen blir jävig då, för mig skulle det vara lite främmade att uttrycka en åsikt om hur jag tycker man ska döma innan jag dömer.’ (10)

155

who keeps within the perimeters of one’s competences. The respondents do not mention any preferred consequences when explaining their view on the practice of expressing opinions. Their reasoning appears to follow from their understanding of what obligations come with being a judge in the Swedish legal system. For these judges it means being impartial. They perceive that expressing one’s own opinions in the requests for preliminary rulings con-travenes the notion of impartiality.

It may appear to be a trivial finding that judges’ perceive impartiality as important. However, what makes it interesting is that the principle of impar-tiality is understood as incompatible with the inclusion of opinions in the request for a preliminary ruling. The judges do not base their standpoint regarding opinions on the prescription from any formal codified rules. As one might recall, opinions are allowed according to the EU guidelines. In-stead, the answers from the respondents suggest that the expression of opin-ions is not an accepted practice within some segments of the Swedish judici-ary. That is, the judges’ mode of reasoning is shaped by informal practices within the profession. What makes this finding even more fascinating is that other respondents view the inclusion of opinions as mandatory. The tentative conclusion that can be distilled from the findings is that different informal practices are at work in different spheres of the Swedish judiciary. These, in turn, lead to fundamentally different approaches to the preliminary ruling procedure. What the differences stem from will be further discussed in the upcoming section but it may be noted that almost all of the respondents who view the practice of expressing opinions as incompatible with the norm of impartiality belong to the courts of final instance.

Judicial reasoning beyond self-interest? Previous research has mainly portrayed judges as self-regarding actors. Whether and to what extent does this assumption hold up to empirical test-ing, and how are national judges actually reasoning in the preliminary ruling procedure? To answer these questions requires a summary and analyses of the respondents’ answers.

To begin with, what types of substantive reasons are individual national judges guided by when making choices in the preliminary ruling procedure? In total, seven different reasons can be identified in the judges’ responses. The frequency of these reasons is displayed in Table 6.2 below and divided into three different categories: personal, legal and political reasons.

156

Table 6.2. The frequency of substantive reasons

Logic of consequentialism

Logic of appropriateness

Substantive reasons Legal reasons

Formally required 10 Uphold the functioning of the legal system 10 Safeguard the courts’ impartiality 10 Personal reasons Protect one’s reputation 6 A strategy of blame avoidance 2 Political reasons Limit EU law expansion 1 Influence EU law 1 Sum of answers 20 20 Number of respondents 16137 20

Previous research on EU legal integration and judicial politics has mainly theorized that personal reasons matter for judicial decision-making (Baum 1997, Dunoff & Pollack 2017, Mattli & Slaughter 1998, Posner 1993). Here, the results show that a few judges do bring up personal considerations such as how to protect their reputation. But it is not the most common type of reason. Instead, and as can be seen in Table 6.2, the judges most commonly make what can be referred to as legal considerations in the preliminary rul-ing procedure. Legal considerations first and foremost include those re-sponses in which judges motivate their decisions by referring to formal rules. However, they also include those instances in which judges think about how different decisions matter for the legal system as such. For example, ten re-spondents say that they withhold cases from the CJEU in order to uphold the overall functioning of the legal system. This is a legal reason in the sense that it stems from a concern about the well-being of the national, and the European, legal order. Lastly, two judges put forward reasons that can be labeled political. For example, limit the expansion of EU law. This means that judges refrain from referring cases to the CJEU in order to avoid a de facto transfer of decision-making power from the member state to the EU. This is a political reason since it concerns national control over policy out-

137 Four respondents expressed two types of reasons. This explains the difference between number of answers (20) and number of respondents (16) giving those answers. Respondent 9 expressed two different types of self-regarding preferences. Respondent 1, 2 and 5 expressed both other-regarding preferences and self-regarding preferences.

157

comes (Golub 1996) and the delimitation of competences between the EU and the member state.

Turning to the analysis of what constitutes the driving forces of judicial behavior, the first important question is how frequent different modes of reasoning are when looking at the judges’ overall behavior in the preliminary ruling procedure. Table 6.3 below shows the number of individual respond-ents that reason according to each of the four modes of reasoning in at least one of the two choices they make in the preliminary ruling procedure.

Table 6.3. The frequency of different modes of reasoning Logics of action Mode of reasoning

(at least once) Number of

respondents

Consequentialism Self-regarding 7

Other-regarding 11

Appropriateness Formal codified rules 10

Informal prof. practices 10 Sum 38

As can be seen in Table 6.3, the judges are almost as likely to reason accord-ing to the logic of consequentialism (18) as they are to reason according to the logic of appropriateness (20). Table 6.3 also shows that the single most common mode of reasoning is that of other-regard. Eleven judges have based their decisions on other-regarding preferences at least once. The se-cond most common mode of reasoning encompasses both the reasoning based on formal codified rules (10) and the one based on informal profes-sional practices (10). Finally, the least frequent mode of reasoning is that of self-regard; seven of the respondents have made considerations based on self-regarding preferences at least once.

It is clear that the results from the present study stand in rather stark con-trast to the predominant assumption in much previous research. The judges do not mainly make considerations based on their self-interest as has been suggested by most theories on EU legal integration and by the judicial poli-tics literature. In fact, self-regard is the least common mode of reasoning. What we see instead is that other-regarding preferences as well as formal rules and informal professional practices mainly guide the judges’ behavior in the preliminary ruling procedure. The results highlight the diversity in the judges’ way of making decisions. There is no single way of reasoning that dominates the judges’ decision-making: all four modes of reasoning do oc-cur in the preliminary ruling procedure.

158

However, while the results clearly challenge the dominant assumption in the literature that judicial reasoning primarily is self-regarding, the findings can be interpreted in a different way when it comes to the case of Sweden. In chapter 3 the Swedish judiciary was portrayed as being least likely to exhibit a reasoning that builds on considerations related to self-interest. Yet, the statements from seven138 out of twenty judges are unambiguously based on self-regarding considerations. Given the expectation that a self-regarding mode of reasoning was unlikely to be present in Sweden, this is a significant share of cases. In particular, it shows that the assumption about self-interest being an important driving force of judicial behavior, originally formulated to understand US judges, also has relevance when trying to understand the reasoning of Swedish judges.

In addition, this result demonstrates the absence of any serious bias in the respondents’ answers. The methodological challenge that comes with doing interviews is that judges are likely to describe themselves in a way that fits with the normatively desirable ideal of how judges should behave (Baum 1997: 60). An indicator of a strong bias in the respondents’ answers would be to find that they predominantly refer to the formal EU rules when describ-ing their choices. However, this appears not to be the case here. First of all, the majority of the judges’ statements cannot be traced to formal codified rules, and second, several judges do express motivations related to self-regarding preferences.

The judges’ mode of reasoning can be further explored by investigating how consistent individual judges are in their reasoning across the two choic-es. The results are displayed in Table 6.4 below.

Table 6.4. The consistency in the judges’ mode of reasoning

Combined reasoning Number of respondents

Mixed 12

Consequentialism 2 Appropriateness 2 Sum 16139

As Table 6.4 shows, only four judges are consistent in their reasoning across the two choices. The majority of the judges instead display a mixed mode of reasoning; they use different considerations depending on the type of deci-sion they are faced with. For those twelve respondents it means expressing 138 Respondents 1, 2, 3, 4, 5, 9 and 13. 139 As previously discussed, four judges out of 20 did not take a stand on the second key choice which means that the number of respondents in the combined behavior column is 16.

159

consequentialist considerations in one of the choices while reasoning accord-ing to the logic of appropriateness in the other choice.

In order to disentangle how the judges mix different logics of actions it is necessary to disaggregate the data and look at the variation in the judges’ modes of reasoning in each of the two decisions. But what type of reasoning can be expected to dominate which decision? March and Olsen have sug-gested that, in a specific situation, a clear logic will dominate a less clear logic (March and Olsen 2009: 20). For example, if rules, here understood as the formal codified rules, are precise, consistent, and legally binding they provide the actors with clear behavioral guidance. All else being equal, such rules make it more likely that the logic of appropriateness would prevail over the logic of consequentialism (March and Olsen 2009: 21). This proposition of prescriptive clarity means that we should expect the logic of appropriate-ness to dominate the actors’ way of reasoning when the rules are clear and, conversely, that the logic of consequentialism dominates when the rules are unclear.

What about the prescriptive clarity of the rules regulating the preliminary ruling procedure? The formal codified rules regulating the decision to refer cases to the CJEU arguably have a relatively high degree of clarity. Article 267 TFEU and the CILFIT-criteria are prescribing a legally binding action, at least for the high court judges who shall refer all unclear cases to the CJEU. Lower courts may request references from the CJEU but they are not obliged to do so. In addition, judging from their statements during the inter-views, the respondents were all well aware of these rules. Based on the ar-gument the a clear logic will dominate, the judges are expected to be more likely to reason according to the logic of appropriateness than the logic of consequences when it comes to the decision of whether to send cases to the CJEU.

Turning to the practice of expressing opinion in the requests for prelimi-nary rulings the expectation is the opposite. The rules in the “Information note on the preliminary ruling procedure” are less precise in the sense that they do not prescribe obligatory action. They are more akin to guidelines than binding regulations. Moreover, the respondents seldom referred to them during the interviews. Here, the prescriptive clarity appears to be lower, meaning that we would expect the consequentialist logic to dominant the judges’ way of reasoning.

In order to assess this matter the national judges’ choices are examined separately. Table 6.2 showed that the respondents brought up at total of sev-en different types of substantive reasons for what choices they make in the preliminary ruling procedure. Tables 6.5 and 6.6 below show the frequency of the various reasons in each of the two choices and how those relate to the four different modes of reasoning.

160

Table 6.5. Whether or not to refer cases to the CJEU - the frequency of stated rea-sons140 Consequentialism Appropriateness Self-

regarding Other-

regarding

Formal codified

rules

Informal prof.

practices Reasons for whether to refer cases or not

Formally required

7

A strategy of blame avoidance

2

Uphold the functioning of the legal system

10

Protect one’s reputation

4

Limit EU law expansion

1

Sum of answers 6141 11142 7 0 Number of respondents 13 7

What are judges considering when deciding whether or not to refer a case to the CJEU? The first thing to note in Table 6.5 is the row “number of re-spondents.” It shows that, contrary to our expectations, the most common logic according to which the judges are reasoning is that of consequential-ism. A total of thirteen judges express that they based their decisions on ex-pected consequences while the remaining seven say that they made consid-eration based on the logic of appropriateness. Next, the row “sum of an-swers” displays how common each of the four modes of reasoning is. Within the logic of consequentialism, judges are most likely to exhibit a mode of reasoning based on other-regard. However, considerations that build on self-regarding preferences also occur. Three judges express both self-regarding and other-regarding preferences. This means that eleven respondents make considerations that can be understood as other-regarding while five express a type of reasoning that is based on self-regard. In contrast, within the logic of appropriateness, the answers from seven respondents are based on formal codified rules and none are based on informal professional practices.

140 Four of the respondents have stated two different reasons for their respective decisions “whether to refer cases or not.” Both of their answers have been included in the numbers reported in Table 6.2 which means that the total number of answers adds up to 24 instead of 20 responses; see the next two notes below. 141 Respondent 9 expressed two different types of self-regarding preferences. 142 Respondent 1, 2 and 5 expressed both other-regarding preferences and self-regarding preferences.

161

It is clearly the case that the logic of consequentialism dominates the judges’ reasoning in the first choice. However, the picture is reversed when turning to the overview of the judges’ second choice: whether or not to in-clude opinions in the requests for preliminary rulings. As Table 6.6 shows, a total of thirteen respondents now base their decisions on the logic of appro-priateness while only three judges reason according to the logic of conse-quentialism. Again, this is contrary to our theoretical expectations.

Table 6.6. Whether or not to express opinions in the request for a preliminary ruling - The frequency of stated reasons Consequentialism Appropriateness Self-

regar-ding

Other-regarding

Formal codified

rules

Informal prof.

practices Reasons for whether or not to express opinions

Formally required

3

Influence EU law

1

Protect one’s reputation

2

Safeguard the courts’ impartiality

10

Sum of answers 2 1 3 10 Number of respon-dents143 3 13

Moreover, the row “sum of answers” in Table 6.6 shows that the judges’ most common mode of reasoning when deciding whether to express opin-ions is that of informal professional practices. Ten of the judges’ answers are consistent with a mode of reasoning based on informal professional practices while three are based on formal codified rules. Conversely, only the response from one judge corresponds to considerations based on other-regard while two respondents based their decision on self-regarding preferences.

The conclusion that can be drawn from comparing the choice to refer cas-es to the CJEU with the choice to express opinions in those cases is that the judges’ way of reasoning differs depending on the type of choice they are making. However, this variation does not follow the theoretical expectations 143 Four of the respondents had not thought about expressing opinions at all and did not take a stand on whether they find it to be acceptable or not. Their answers are therefore not included in the table, meaning that the sum of respondents is 16 with regard to the second key choice (total number of respondents is 20).

162

about prescriptive clarity formulated by March & Olsen (1998, 2009). De-spite the clarity of the rules in Article 267 TFEU, the judges are most com-monly reasoning according to the logic of consequentialism instead of the logic of appropriateness when deciding whether or not to refer cases to the CJEU. To be sure, the prescriptive clarity is arguably higher for judges of final instance since they shall refer cases concerning the interpretation of EU law to the CJEU. Yet, the respondents from the high courts are not more likely than lower court respondents to apply a reasoning based on the logic of appropriateness.

What may be a possible explanation for the judges’ unexpected way of reasoning? A suggestion is to flip the proposed theoretical logic of prescrip-tive clarity. The judges are able to apply the consequentialist logic precisely because the formal legal framework is clear. The argument is that if the rules regulating the legally binding action are unambiguous this means the judges will be able to determine what aspects are not regulated. Based on this in-formation, they can easily identify what discretion they have within the framework of rules. In short, actors need to know the rules in order to cir-cumvent them. This is not to say that the judges break the formal rules, ra-ther, it means that they know what situations are not guided by the rules and, in those instances, they may apply any type of logic, even the logic of con-sequentialism. This means that the prescriptive clarity of the formal rules is not contributing to make the logic of appropriateness the dominant way of reasoning. Instead, it actually enables the actors to apply the consequentialist logic and think about how different decisions benefit themselves and others.

In the second choice, whether or not to include opinions in the cases sent to the CJEU, the prescriptive clarity of the formal rules is low yet the domi-nating logic is that of appropriateness and not consequentialism. Why is this? Here, it is important to note that the dominating logic we are talking about is the logic of appropriateness guided by informal professional prac-tices. It can be theorized that when there are no clear formal codified rules that prescribe what the appropriate action is in a given situation, the judges will resort to informal professional practices.144 The uncertainty of the situa-tion leads the judge to ask him or herself what is the fundamental basis of my professional identity and not what is the best decision given my prefer-ences. That is, the lack of legally binding regulations makes the judges un-sure about what discretion they have in a given situation. So when it comes to the matter of whether to include opinions in the requests for preliminary rulings, a fair share of the judges (ten) find this practice to be contrary to a fundamental professional principle of judicial impartiality. This result points 144 Alternatively, the judges do not see this as a decision in which any direct self-interest or other-regard is involved (March & Olsen 2009). This makes it difficult to act according to the logic of consequentialism and the judges will instead resort to the logic of appropriateness. One can also consider if the informal professional practice is so strong that weak formal guidelines cannot change/shape the judges’ behavior.

163

to the importance of analytically distinguishing between informal profes-sional practices and formal rules as well as their respective prescriptive clari-ty. It can be summarized as follows: if the prescriptive clarity of the formal codified rules is high, actors are more likely to consider expected conse-quences. Conversely, if the prescriptive clarity of the formal codified rules is low, actors are more likely to resort to a reasoning based on informal profes-sional practices.

Apart from the variation in the judges’ overall mode of reasoning, another finding from the analyses merits attention. It concerns the respondents’ dif-ferent views on the practices of expressing opinions in the requests for pre-liminary rulings. Previous research has hypothesized that judges make calcu-lations of whether it is worthwhile to spend time and resources on the formu-lation of opinions. The expected outcome is that the judges mainly include opinions in the types of cases in which they think that the CJEU will be sympathetic to their views (Nyikos 2006). This hypothesis appears to not be corroborated based on the findings from the present study. What the result in Table 6.6 instead shows is that the judges mainly consider the appropriate-ness of the practice. To be sure, one judge thinks that national courts should try to influence the course of EU law by including opinions and two judges bring up the potential negative effect of expressing opinions on one’s reputa-tion. But most of the respondents (10) find the practice of including opinions in the requests for preliminary rulings to be inappropriate. They think that the inclusion of opinions is alien to the Swedish legal system and that includ-ing opinions means exceeding one’s authority, possibly disqualifying the judge from being an impartial arbitrator.

However, what is interesting is that the formal EU regulation on the in-clusion of opinion does not provide any support for this view. Although the inclusion of opinions in the request for a preliminary ruling is by no means required, the national judges are allowed to express them. From the point of view of EU law there is nothing wrong with including opinions and at times it is even encouraged. What makes the matter even more interesting is that five other respondents say that they view the inclusion of opinions as being practically mandatory. How are we to understand these different approaches to the expression of opinions in the requests for preliminary rulings?

Two explanations can be discerned. First, it can be a case of obscuring the real motivation. The judges may refer to the need to safeguard the courts’ impartiality instead of admitting that they are not comfortable with the prac-tice of expressing opinions for other reasons. As suggested by another re-spondent, judges may use the term “impartiality” as an excuse for not includ-ing opinions in the request for a preliminary ruling:

164

[Q: What about expressing opinions?] I didn’t have resources and I didn’t have the skills so I would have found it to be incredibly risky to try to predict how the case should be resolved so I didn’t do it. But you can always if you want to use the notion of impartiality as a shield to hide behind. (4)

The quote would suggest that the judges do not want to take the risk of ex-pressing an opinion since they may be “wrong” in their understanding of EU law and its application to the case. This, in turn, might jeopardize one’s reputation in the legal community.

Although the fear of appearing incompetent may be the reason that some judges refrain from expressing opinions it does not, however, fit with the impression from the interviews. When asked about the practice of including opinions in the requests many of the respondents had not before considered the matter thoroughly. Their gut reaction was that opinions are inappropriate and a threat to the judges’ impartiality. Then, they started to elaborate on what reasons they had for this view in a rather probing manner, revealing various considerations connected to informal professional practices. For example:

Yes, I just thought about this thing about expressing an opinion, I felt now when you said it, that it feels very odd in some way. If you were to decide that case later, will you then be disqualified from judging? It would be an un-familiar practice for me to express an opinion on how I think the case should be decided before I actually decide the case. (10)

The lasting impression from the interviews is that the judges do not just make references to their “impartiality” as a pretext for not including opin-ions. Instead, it seems like the judges actually base their decisions on what they think the informal practices prescribe as being the appropriate course of action. This interpretation also corresponds with the argument outlined above about the relative prescriptive clarity of different rules; when the for-mal codified rules lack prescriptive clarity, for example because they are non-binding, judges resort to informal professional practices instead of thinking in term of expected consequences. We must therefore consider an-other explanation for the different attitudes that emerged.

The proposed explanation for the differences in how judges’ reason is that there is a divide within the Swedish judiciary when it comes to how judges at different levels of the judicial hierarchy interpret their role in the prelimi-nary ruling procedure and what practices to follow. All of the five judges who perceive the inclusion of opinions as being appropriate belong to lower courts. In contrast, almost all of the ten judges who are against the practice of expressing opinions work or have worked in the high courts.

What is particular about high courts as an institution is that they have a prominent role in the legal system. The responsibility of high courts, and by extension high court judges, is to protect the legal order and to steer overall

165

legal development (Alter 2001: 48). Lower court judges only need to focus on the case at hand. Given the differences in responsibilities it can be argued that the informal professional practices shaping the judges’ reasoning also differ between high and low courts. This results in varying conceptions of what is the appropriate behavior in the preliminary ruling procedure.

The high court judges are likely to place a greater emphasis on the profes-sional practices related to matters of overarching judicial principles since they are part of an institution that has the responsibility to protect the legal order. They are expected to be extra wary of potential threats against funda-mental principles such as the impartiality of the judiciary. For instance, if they perceive the expression of opinions as a threat against their impartiality it would be close to impossible for them to include opinions in the request for a preliminary ruling. Impartiality is embedded in the high court judges’ professional identity and expressing opinions is simply incompatible with what they represent. This is what the results show. When asked about their view on including opinions, high court judges find the principle of judicial impartiality to be paramount.

Indeed, lower court judges may also care about broad legal principles such as impartiality. However, considering overall legal development or how to protect the legitimacy of the judiciary is not part of their everyday job description. Therefore, they are not expected to experience a conflict be-tween impartiality and the inclusion of opinions in the requests. The argu-ment is that different professional contexts, such as what court the judge is working in, matter for the judges’ decision-making. The variation in infor-mal professional practices is thus suggested to be the reason why high court judges and not lower court judges view opinions as strange and inappropriate even though the formal codified rules allow them.

This brings us to the general question about how to understand the varia-tion in reasoning among those actors who all follow the logic of appropriate-ness. In particular, we may ask how the foregoing discussion can elucidate under what conditions informal professional practices, and not formal codi-fied rules, come to shape the judges’ mode of reasoning. Based on the find-ings presented in this chapter it can be hypothesized that the logic of appro-priateness will be dominated by formal codified rules if those rules prescribe a clear action. This can be exemplified by the judges’ reasoning in the deci-sion about whether to refer cases to the CJEU. Here, the formal rules are clearly defined and the judges who are reasoning according to the logic of appropriateness most commonly refer to those formal regulations (Table 6.5). In contrast, the logic of appropriateness will be dominated by informal professional practices if the formal codified rules are unclear and do not prescribe obligatory action. This is the mode of reasoning that the judges employ when deciding whether or not to include opinions in the request for a preliminary ruling (Table 6.6). The formal codified rules are less known among the judges and, importantly, they do not prescribe any obligatory

166

action. The rules merely suggest that opinions may be included. As a result, judges who are reasoning according to the logic of appropriateness are most commonly basing their considerations on informal professional practices.

The findings suggest that it is important to differentiate not only between the relative prescriptive clarity of the logic of consequentialism and the logic of appropriateness in a specific situation but also between the prescriptive clarity of formal codified rules and informal professional practices. This can be summarized in the following way: if the relative prescriptive clarity of the formal codified rules is low, actors are likely to resort to informal profes-sional practices when making decisions.

Summary of chapter This chapter set out to provide a first empirical account of how the national judges reason when making decisions in the preliminary ruling procedure. First, what are the judges’ substantive reasons for making different deci-sions? Contrary to the expectations in previous research, the results show that personal reasons are less common than legal ones. Furthermore, and in contrast to the prevalent view in the literature, consideration about how to maximize one’s self-regarding preferences is not the main driving force of the national judges’ behavior in the preliminary ruling procedure. Rather, other-regarding preferences, as well as formal rules and informal profession-al practices, mainly guide the judges in the preliminary ruling procedure. This demonstrates that there is not only one type of logic that guides the judges when they make decisions in the preliminary ruling procedure. In-stead of reasoning exclusively to the logic of consequentialism or the logic of appropriateness, the majority of the judges display a mix between the two logics of reasoning.

The chapter has also elaborated on under what conditions one type of log-ic will dominate over another. That is, under what circumstances actors re-sort to a way of reasoning based on expected consequences or one based on appropriateness. It theorizes that, contrary to the argument made by March and Olsen, a high prescriptive clarity of formal codified rules will not lead actors to reason according to the logic of appropriateness. A high prescrip-tive clarity allows for action based on consequentialist consideration since the actors are made aware of what room they have to maneuver in pursuit of their self- or other-regarding preferences. Conversely, the lack of clearly defined and legally binding codified rules is likely to lead judges to resort to informal professional practices based on their professional identity as judges. This also points to the importance of distinguishing between formal codified rules and informal professional practices when discussing what shapes ac-tors’ behavior within the logic of appropriateness instead of lumping them all under the same heading of “rules.”

167

Finally, the results indicate that we have reason to believe that the infor-mal professional practices vary depending on what institutional context ac-tors find themselves in. This is suggested to be the reason why high court judges and not lower court judges view the inclusion of opinions in the re-quests for preliminary rulings as strange and inappropriate even though the formal codified rules allow them.

One should be cautious about drawing general conclusions concerning the reasoning of judges in other EU member states on the basis of individual cases. Even though we have no reason to believe that Swedish judges sys-tematically deviate from other European judges, further studies on how judges make decisions in the preliminary ruling procedure are needed. How-ever, the critical, least likely, aspect of the Swedish judiciary is interesting in relation to the expectations formulated by the judicial politics literature. The result from the analysis shows that even Swedish judges employ a self-regarding mode of reasoning. This finding provides preliminary support for the conclusion that judges in other member states, in which the legal context bears closer resemblance to the US legal system, reason in a similar way in the preliminary ruling procedure. Future research should focus on whether the diversity in the Swedish judges’ mode of reasoning also appears in other member state contexts.

168

7. Conclusion: Findings and implications

This final chapter returns to the overall research aim and questions that were formulated in the introductory chapter to assess the collected empirical and theoretical findings. The national courts’ cooperation with the CJEU in the preliminary ruling procedure forms the crucial link between the member states and the EU legal system. The national courts’ acceptance of the CJEU’s authority in the early days of the European Community made further integration possible despite political disagreement among the member states’ governments (Alter 1998a). However, the role that the national judiciaries have come to play in European legal integration has been the subject of much debate. National courts have been portrayed either as supporters of EU legal integration or as defenders of national sovereignty. Therefore, the overarching aim of this thesis has been to shed light on the national courts’ and judges’ behavior as gatekeepers in the preliminary ruling procedure to improve our understanding of their role in EU legal integration.

The empirical focus of the thesis has been national court behavior in rela-tion to a supranational court in the multileveled political system of the EU. This thesis has analyzed the national courts’ aggregated behavioral patterns from a comparative perspective and the reasoning of individual judges in the Swedish context to answer questions regarding what choices national courts make in the preliminary ruling procedure and why. The upcoming sections discuss the comprehensive picture of national court behavior that this thesis has provided. These sections begin by summarizing the main findings from each analytical chapter before turning to a discussion of the study’s overall contributions and the implications of the findings for the research field. In particular, these sections will further elaborate on how the focus on the courts’ key choices in the preliminary ruling procedure, as well as the focus on the reasoning of individual judges, have contributed to our understanding of the role of national courts in EU legal integration.

Main findings To fulfill the overarching aim, this thesis set out to answer three intercon-nected research questions. Two of these questions sprung from the scholarly discussion concerning the type of aggregated behavior that national courts exhibit in the preliminary ruling procedure. The first research question is

169

directly concerned with the main theoretical controversy: whether national courts should be characterized as either supporters of further integration or as defenders of national sovereignty. The second research question also fol-lows from this debate and concerns the factors that can explain the variations in the courts’ behavior and why some national courts support EU legal inte-gration, while other courts instead protect national sovereignty. In this thesis, these questions have been empirically examined by a systematic comparison of the types of choices that national courts make in the preliminary ruling procedure and the factors that can explain the variations in these choices.

What types of choices are national courts making in the preliminary rul-ing procedure? As shown in chapter 4, in just over half of all the references that were examined, national courts allow the CJEU to decide legal issues with a high degree of political sensitivity. This is understood as being a con-siderable share of cases in which the CJEU, with its track record of extend-ing the scope of EU legal integration through its rulings, can specify how EU law should be interpreted in cases that concern the validity of national poli-cies. In relation to previous research, this result is consistent with the find-ings from Sweden (Leijon and Karlsson 2013).

Turning to the national courts’ second key choice, namely, the types of opinions that they express, the results accord with previous empirical find-ings from Sweden (Leijon and Karlsson, 2013), and the most commonly expressed opinion is support for EU law. This finding suggests that national courts are more likely to invite the CJEU to expand the scope of EU legal integration than they are to defend national law and challenge EU law. We also learned that national courts expressed opinions in nearly half of all the requests for preliminary rulings, which was a slightly higher share than that found in a previous study (Nyikos 2006). Considering previous research, this result suggests that national courts express such opinions, although they are not formally required. This result further suggests that an important share of the so-called judicial dialogue between national courts and the CJEU (Jacobs 2003; Rosas 2007) takes the shape of explicit opinions that are included in the requests for preliminary rulings.

Returning to the main theoretical debate about how we can understand the national courts’ behavior in EU legal integration, do the courts’ decisions lead to a behavioral pattern that supports EU legal integration or a behavioral pattern that defends national sovereignty? The findings clearly show that national courts most commonly make choices that support EU legal integra-tion. Accordingly, they refer cases with a high degree of political sensitivity and express opinions in support of EU law. This behavior, which has been predicted by the judicial empowerment approach, is almost three times more common than the behavior that is envisaged by the sustained resistance ap-proach. Concerning the theoretical controversy between the two approaches, this thesis finds that the judicial empowerment approach provides a more

170

accurate depiction of national court behavior in EU legal integration than does the sustained resistance approach.

The other main finding, however, is that the national courts’ behavior regularly deviates from the predictions that are derived from both of the main theoretical approaches, judicial empowerment and sustained resistance. These deviations are emphasized when examining how national courts com-bine the two key choices. The behavioral pattern that is proposed by the ju-dicial empowerment approach is still the single most common pattern when investigating the types of opinions that national courts express in different types of cases. However, in slightly less than half of all observed behaviors, national courts exhibit two previously untheorized behavioral patterns: ac-cess contained and compatibility defended. This finding means that a sub-stantial share of the national courts’ combined behavior has not been proper-ly defined in previous research. This thesis interprets these two behavioral patterns as attempts by national courts to balance the demands from the EU level on the one hand and the member state level on the other hand.

This inquiry brings us to the second research question, namely, what fac-tors can explain the variations in the types of choices that national courts make in the preliminary ruling procedure. Chapter 5 showed that when it comes to the variations in the types of cases that national courts refer to the CJEU, the particular member state’s tradition of majoritarian democracy was found to play a role. Specifically, courts in the member states of Denmark, Finland, Sweden and the UK, which all have a strong tradition of parliamen-tary sovereignty, were found to be less likely than other courts to provide the CJEU with access to cases that have a high degree of political sensitivity. Moreover, courts of first instance were more likely than all higher courts to refer cases with a high degree of political sensitivity to the CJEU. Turning to the second choice, the only factor that could account for the variations in the types of opinions that national courts express was the national courts’ placement in the domestic legal hierarchy. Courts of final instance were found to be more likely than courts of first instance to express support for national law. Conversely, courts of first instance were found to be more like-ly to express support for EU law than courts of final instance.

These findings further inform the overarching theoretical debate on un-derstanding the behavior of national courts in EU legal integration. Overall, courts from the Nordic countries and the UK are more likely than courts from other member states to defend national sovereignty. The same is true for higher courts; in particular, courts of final instance are making choices that defend national law. In contrast, courts of first instance can be under-stood as supporters of EU legal integration. These results show that the theo-ries that are derived from both the sustained resistance approach and from the judicial empowerment approach can help us to understand the choices that national courts make in the preliminary ruling procedure. In relation to previous research, this result further means that variations in not only refer-

171

ral rates (Mayoral 2012; Sweet and Brunell 1998b; Vink et al. 2009) but also the types of cases and opinions that national courts send to the CJEU can be attributed to the courts’ placement in the domestic legal hierarchy.

By answering these two research questions, this thesis provides the first comprehensive study of the choices that national courts make in the prelimi-nary ruling procedure. In addition, the findings shed new light on the theo-retical controversy. However, to fully understand the national courts’ behav-ior as gatekeepers in EU legal integration, it is necessary to explore how individual judges experience the preliminary ruling procedure. The final analytical chapter therefore shifts the empirical and theoretical focuses from the national courts’ overall behavioral pattern to the decision-making of in-dividual judges. Specifically, this thesis examines how judges reason when deciding whether to refer a case to the CJEU and when deciding whether to include opinions in the requests for preliminary rulings.

What types of reasons guide individual national judges when they make decisions in the preliminary ruling procedure? The findings reported in chap-ter 6 show that these reasons fall into three categories, i.e., legal, personal and political reasons. Previous research has primarily suggested that person-al reasons matter for judicial decision-making. Indeed, the findings in this thesis show that several judges mention personal considerations when they describe the choices they make. Here, one finds considerations regarding how to protect one’s reputation and how to avoid being blamed for making unpopular decisions. However, different types of legal reasons are more common and include such legal considerations as references to formal rules and how different decisions matter to the legal system. For example, some judges say that they withhold cases from the CJEU to uphold the functioning of the preliminary ruling procedure because making too many referrals would overburden the system. A few judges also advanced reasons that can be labeled political. For example, to refrain from referring cases to the CJEU to avoid a de facto transfer of decision-making power from the member state to the EU. This finding corroborates a mechanism that has been proposed by previous research, namely, that national judges are reluctant to engage in the preliminary ruling procedure because they are concerned that doing so un-dermines the coherence of the national legal order (Alter 2001: 48; Dehousse 1998: 173; Maher 1998).

Apart from unveiling the substantive reasons that guide individual judges, the findings also contribute to our understanding regarding what drives judi-cial decision-making. Contrary to the predominant assumption in previous research (Alter 2001; Burley and Mattli 1993; Tridimas and Tridimas 2004; Weiler 1994), judges are not primarily guided by calculations of expected consequences based on self-regarding preferences. Instead, this turns out to be the least common mode of reasoning. Rather, other-regarding prefer-ences, formal rules and informal professional practices mainly guide the judges in the preliminary ruling procedure. Regarding the debate in the judi-

172

cial politics literature about what overarching logic that drives judicial deci-sion-making, this thesis finds that judges are almost as likely to employ a mode of reasoning that follows from the logic of consequentialism as they are to employ a mode of reasoning that builds on the logic of appropriate-ness.

Theoretical contributions Returning to the overarching aim of this thesis, it is time to discuss how we should understand the results from the analytical chapters in light of previous research. How has this thesis’ analytical focus on the national courts’ two choices in the preliminary ruling procedure and the reasoning of individual judges contributed to our knowledge on the national courts’ role as gate-keepers in EU legal integration?

National courts and national judges: The implications of incorporating the micro perspective One of the two claims that were made in the introductory chapter was that it is insufficient to focus solely on the national courts’ aggregated behavioral patterns; assessing what drives the actions of individual judges is just as important if we want to understand the role of national courts in EU legal integration. Judicial decision-making is subject to extensive research in the US, but few empirical studies have explored the actions of individual judges in the European context. Consequently, little is known regarding how na-tional judges experience the preliminary ruling procedure.

In chapter 4, judicial empowerment was identified as the single most common behavioral pattern among national courts in the preliminary ruling procedure. This finding means that based on their aggregated behavioral patterns, national courts are being supportive of EU legal integration. In addition, it is first and foremost the choices made by first-instance courts that enable the CJEU to foster EU legal integration, just as predicted by the inter-court competition hypothesis. However, these descriptions of the national courts’ role in EU legal integration changes when we take into account how individual judges experience the preliminary ruling procedure.

To understand how the micro perspective informs our understanding of the national courts’ behavior as gatekeepers, it is important to recall how previous research expects individual judges to act. Scholars such as Alter (Alter 1998b; Alter 2001), Mattli and Slaughter (Burley and Mattli 1993; Mattli and Slaughter 1998) and Weiler (Weiler 1994) have elaborated on how general theories concerning European integration can be used to under-stand the role of national courts and judges in the EU legal context. For in-

173

stance, the theoretical underpinnings of the judicial empowerment approach is the neofunctionalist claim that one of the most important driving forces of European integration is the self-interest of various actors (Stone Sweet and Sandholtz 1997). In turn, the self-interest of national courts and judges has been conceptualized as their wish to increase their powers vis-à-vis other courts and political actors. This assumption, that actors are driven by consid-erations concerning expected consequences, also mirrors the dominant per-spective in the judicial politics literature (Dunoff and Pollack 2017; Larsson and Naurin 2016; Rehder 2007; Spiller and Gely 2008: 34).

Whether and to what extent previous research is correct in portraying na-tional judges as acting on the basis of the logic of consequentialism and self-regarding preferences is the key question. The findings in this thesis show that national judges in fact consider the expected outcomes of different choices when making decisions in the preliminary ruling procedure, which is in line with the consequentialist logic of reasoning. Strikingly, however, in most of these instances, judges are not considering how their choices relate to their self-regarding preferences. Instead, the judges consider how the choices they make might affect the well-being of the parties to the case or the legitimacy of the legal system. That is, national judges exhibit other re-gard.

These findings are highly important since they create space for a revision regarding how we should understand judicial behavior in the preliminary ruling procedure. In particular, the national judges’ modes of reasoning have implications for how we should interpret the national courts’ aggregated choices in relation to EU legal integration. As an example, previous research has interpreted the practice among courts of withholding cases from the CJEU as an indication of resistance to further integration and as a means for national courts to maintain control over policy outcomes (Golub 1996). But the findings that are presented here show that national judges may have other reasons not to request preliminary rulings. For instance, judges withhold cases with the intention of safeguarding the functioning of the preliminary ruling procedure and the legitimacy of the judiciary, not to defend national sovereignty. This particular finding shows that what might be interpreted as resistance toward EU legal integration on the aggregated level may be a by-product of other-regarding preferences rather than an attempt by national courts to deliberately hinder further EU legal integration. If we ignore how judges reason, we run the risk of misinterpreting the driving forces of judi-cial decision-making by drawing faulty conclusions about micro behavior from patterns at the macro level.

Exploring the national courts’ behavior from a micro perspective sheds new light on previously unquestioned assumptions in the literature about judicial decision-making. What on an aggregated level either looks similar to a defense of national sovereignty or support for further integration may in fact be driven by considerations that have little to do with the national judg-

174

es’ attitudes toward the EU. Adding the micro perspective thus enriches our understanding of what drives EU legal integration.

Moreover, the results presented in chapter 6 demonstrate that judges are just as likely to reason according to the logic of appropriateness as they are to reason according to the logic of consequentialism. This finding clearly shows that by assuming that the logic of consequentialism is the dominating mode of reasoning among judicial actors, we run the risk of overlooking how other types of considerations shape judges’ behavior. This includes, for ex-ample, considerations regarding what the judges conceive as being the ap-propriate course of action given their professional identities. In fact, failure to consider how formal codified rules and informal professional practices shape judicial behavior would mean disregarding many of the respondents’ reasoning in the preliminary ruling procedure.

The analysis of how national judges experience the preliminary ruling procedure also sheds light on the relationship between low and high courts and their varying dispositions toward EU legal integration. Previous research has suggested that inter-court competition is the key mechanism for under-standing the observed variations in national court behavior. Alter has theo-rized that lower courts, primarily courts of first instance, realize that they can become judicially empowered by participating in the preliminary ruling pro-cedure. Higher courts, especially courts of final instance, instead avoid the procedure because they want to keep their control over national legal devel-opments and not give this power to the CJEU (Alter 1998b; Alter 2001).

Indeed, the inter-court competition hypothesis finds support in the analy-sis of which factors can account for variations in the national courts’ aggre-gated behavioral patterns (chapter 5). Accordingly, the choices that national courts make vary with court level, and courts of first instance are more likely than high courts to make choices that allow the CJEU to expand the scope of EU legal integration. However, a macro relationship is insufficient to estab-lish that inter-court competition is the mechanism underlying the courts’ different choices. The question concerning what leads to the observed differ-ence in behavior among courts at different levels of the judicial hierarchy must be examined from the perspective of individual judges. Contrasting the results from the analysis of the courts’ behavioral pattern with the results from the analysis of individual judges’ experiences of the preliminary ruling procedure yields important new insights. The judges’ modes of reasoning reveal that internal competition between high and low courts is not the only possible explanation regarding why judges’ placement in the judicial hierar-chy matters to the choices that they make. Instead, another mechanism emerges.

The mechanism uncovered by this study is that differences in behavior between high and low courts originate from variations in the types of infor-mal professional norms. This argument builds on the insights from chapter 6 regarding how national judges view the practice of including opinions in the

175

requests for preliminary rulings. According to the formal EU rules, expres-sion of opinions is allowed but not obligatory. However, it turns out that judges who are placed at different levels of the judicial hierarchy have dia-metrically opposed beliefs concerning the appropriateness of this practice. Most of the lower-court judges view the inclusion of opinions in the requests for preliminary rulings as practically mandatory. In contrast, the high court judges instead find the practice of expressing opinions to be irreconcilable with the overarching principle of judicial impartiality.

This finding suggests that depending on their position in the judicial hier-archy, judges face different informal professional norms regarding what constitutes appropriate behavior. This variation in norms is expected to lead judges to make different choices in the preliminary ruling procedure, which in turn results in varying behavioral patterns between low and high courts on the macro level of analysis.

The role of national courts: Gatekeepers who balance conflicting demands The second claim that was made in the introductory chapter was that any attempt to understand the national courts’ behavior as gatekeepers in EU legal integration requires the investigation of the key choices that they make in the preliminary ruling procedure. What types of cases national courts refer to the CJEU (Alter 2001; Golub 1996; Wind 2010) and what types of opin-ions they express in these requests for preliminary rulings (Conant 2013; Nyikos 2006; Rosas 2007) have been identified as important for understand-ing EU legal integration. Thus far, however, the two choices have only gen-erated a handful of empirical case studies. In addition, theoretical elabora-tions on how the two choices should be understood together have to date been absent.

This thesis’ focus on the national courts’ two key choices in the prelimi-nary ruling procedure has yielded one particularly important discovery, namely, that to understand the role of national courts, we must consider the relationship between these choices. By viewing the two choices together, this thesis finds that nearly half of all the courts’ observable behavior deviat-ed from the expectations that were formulated by the two main theoretical approaches of judicial empowerment and sustained resistance (see chapter 4).

This thesis argues that the previously untheorized behavioral patterns that emerged when combining the choices are the result of the national courts’ attempts to balance conflicting demands. The role of national courts as gate-keepers in the preliminary ruling procedure places them at the intersection between two political systems. This position exposes them to demands from both the member state and the EU. The EU rules that regulate the prelimi-

176

nary ruling procedure are clear: national courts are requested to send EU legal cases to the CJEU. However, from a member state perspective, it is controversial to allow a supranational court such as the CJEU to decide po-litically sensitive cases. The member state governments and the CJEU are likely to disagree about where to draw the line between EU competences and national competences. In general, the CJEU is believed to favor extensions of EU legal competences, whereas the member states want to maintain their control over national policies (Alter 2001; Blauberger 2014).

This thesis shows that national courts respond to these conflicting de-mands in two ways. First, they can make choices that lead to the behavioral pattern compatibility defended. This response means that national courts refer cases with a high degree of political sensitivity to the CJEU, as re-quired by EU law, while simultaneously including opinions that defend the contested national law. By proclaiming their views that a national law is compatible with EU law, national courts have considered the concerns of the member state government when requesting preliminary rulings from the CJEU. In other words, national courts actively defend the compatibility be-tween national law and EU law. The second behavioral pattern shows that national courts make choices that lead to access contained. Here, courts ac-commodate the demands of the EU and the CJEU by indicating that an ex-pansion of the scope of EU law is indeed acceptable in the cases that they refer. This response is not expected to upset the member state governments because the cases in which the opinions are expressed have a low degree of political sensitivity. It is therefore unlikely that the national courts’ opinions in support of EU law will result in the overturning of any sensitive member state policy. The CJEU’s access to the domestic legal sphere is thereby con-tained to cases with a low degree of political sensitivity.

The investigation into the types of choices that national courts make in the preliminary ruling procedure has thus resulted in a revised and deepened understanding of national court behavior in EU legal integration. While pre-vious research has characterized national courts as either supporters of inte-gration or defenders of national sovereignty, the findings presented in this thesis suggest a different role for the courts. Their job as gatekeepers is not only to choose between opening or closing the gate to the prospect of further EU legal integration. Instead, they may also try to strike a balance between the interests of the member state and the interests of the EU.

What are the wider implications of finding that national courts regularly take on the role of balancing these different interests? In particular, what are the implications for EU legal integration and the legitimacy of the EU politi-cal project? To answer these questions, we must broaden the discussion to the overall process of European integration and elaborate on how the nation-al courts’ behavior relates to one of the key challenges to the EU project, namely, how to ensure that citizens perceive the EU as legitimate.

177

Why have countries chosen to become members of the EU? The long-standing answer to this question is that the EU contributes to peace, prosperi-ty and democracy in Europe (Scharpf 2009: 181). The member states trust that the EU is better equipped than them to address the societal challenges that go beyond national borders (Tallberg 2002). Countries have therefore voluntarily given up part of their sovereignty to EU institutions. This transfer of decision-making power is a prerequisite for the functioning of the EU political system and its core, the internal market. It is also vital that all mem-ber states follow EU law for the internal market to be efficient. To ensure that this is the case, the Commission oversees the member states’ observance of EU rules and may initiate infringement proceedings against the member states that fail to comply with EU law. Ultimately, it is the CJEU that has the final say in regard to interpreting EU law, and it may strike down national policies that conflict with the supreme EU law.

The CJEU is providing authoritative interpretations of EU law with the purpose to ensure a uniform application of EU law in all member states, something that is necessary for the EU political system to function properly. However, not everyone favors this institutional setup in which the CJEU has de facto influence over national legislation. Scholars have noted that the CJEU has a pro-EU integration bias, meaning that the CJEU’s decisions often leads to a transfer of power from the national to the EU level (Pollack 2013: 1264-1265) and that this threatens the maintenance of the politically acceptable balance between European and national competences in the EU (Grimm 2015: 471; Kelemen and Schmidt 2012). Upholding the balance is regarded as a pillar stone for the EU’s legitimacy. Scharpf goes as far as to identify the role that the CJEU has taken on to foster European integration as a major threat to the EU’s survival (Scharpf 2009). In the words of Scharpf, “the politically unsupported extension of judge-made European law in areas of high political salience within member state polities is undermining the legitimacy bases of the multi-level European polity” (Scharpf 2009: 198-199).

To understand how the CJEU is undermining the EU’s legitimacy, it is necessary to review the compliance relationship among the citizens, the member state governments and the EU. What is particular about the multi-leveled EU political system is that it is not the EU institutions that enforce EU law and demand compliance from the citizens; these tasks are instead performed by national agencies. The citizens are therefore expected to hold their member state government, not the Commission, the European parlia-ment or the CJEU, accountable for the decisions that are made at the EU level (Scharpf 2009: 180). To uphold the legitimacy of the EU, the member state governments must therefore be able to justify all the rules and policies that emanate from EU institutions and explain to their citizens why these decisions are necessary. This becomes especially important when decisions are unpopular among the majority of citizens and when they concern highly

178

salient interests. Otherwise, there is a risk that the citizens perceive the deci-sions as illegitimate (Scharpf 2009: 189; Schmidt 2004: 993).

In the case of the decisions that are made by the CJEU, the EU’s legiti-macy problem comes to a fore. First, the CJEU is known for making politi-cally controversial decisions that are unpopular among citizens. For instance, it is well-known that the CJEU’s interpretations of EU law have had disrup-tive impacts on the member states’ legal and political traditions (Alter 2001: 48; Dehousse 1998: 173; Maher 1998). There are several examples of CJEU decisions that have undermined the member states’ financial and social poli-cies and resulted in negative reactions from the public (Scharpf 2009: 191-192). Citizens may therefore perceive that the CJEU is failing to maintain the balance between EU legal integration and national competences. The citizens that disapprove of the CJEU’s decisions and want to see them re-voked will turn to the member state government and demand legislative changes. This, however, leads to a second problem; Scharpf claims that it is actually very difficult for the member state governments to contest the CJEU’s rulings by drafting new legislation. Treaty revisions require unanim-ity among the member states, and changing parts of the secondary legislation also has high consensus requirements (Scharpf 2009: 182).

In sum, the member state governments have little possibility to change the CJEU’s decisions due to the high consensus requirements in EU policy mak-ing. The CJEU is thus almost immune to political corrections. According to Scharpf (2009) and Grimm (2015), this undermines the EU’s legitimacy. The decisions made by the unelected CJEU have great political impact. This results in “a state of integration that the citizens were never asked to agree to, but cannot change either, even if they do not support it” (Grimm 2015: 471). From the point of view of these authors, the key problem is that the CJEU is not able or willing to uphold the balance between member state autonomy and European legal integration that is necessary for ensuring the legitimacy of the EU in the eyes of the citizens.

If we accept Scharpf’s and Grimm’s framing of the problem, then the EU political project is faced with an important challenge: how to resolve the tension between the politically legitimate national concerns and the equally legitimate constraints that countries must accept as members of the EU (Scharpf 2009: 199)? On the one hand, Scharpf warns that “ the capacity of member states to comply with EU law reaches its limits when doing so would undermine their own legitimacy in relation to their national constitu-encies” (Scharpf 2009: 188). There is a risk that the member states openly declare noncompliance when the CJEU decides to overturn salient national policies because these policies conflict with EU law (Scharpf 2009: 189, 200). On the other hand, there is no avoiding the fact that the functioning of the EU is dependent on having an institution similar to the CJEU that up-holds EU law. Without the CJEU’s supranational review over the national

179

policies’ compatibility with EU law, the EU would not have an efficient internal market.

What is needed to remedy this problem is an actor that can strike a fair balance between EU legal integration and member state autonomy. Accord-ing to Scharpf, the CJEU is not capable of maintaining such a balance be-cause the CJEU is not embedded in the member states’ political and legal traditions (Scharpf 2009: 186). The CJEU cannot keep track of what are the politically acceptable concerns in each of the 28 different member states. This lack of insight distinguishes the CJEU from national courts. National courts are expected to have an in-depth understanding of the shared values in the national political community. Since the CJEU lacks this type of under-standing of national political and legal practices, it cannot be sensitive to the member states’ concerns (Scharpf 2009: 187). This, in combination with its pro-EU stance, means that the CJEU is not able to balance the interests of the member states with those of the EU.

How should we understand the implications of the national courts’ behav-ior in light of this discussion? Scharpf claims that the EU lacks an actor who can strike a just balance between member state concerns and EU legal obli-gations. However, is this truly the case? This thesis suggests that national courts have in part taken on the responsibility of balancing these different interests. To be sure, Scharpf would be correct in his misgivings if national courts were only to engage in the pro-integration behavior that has been en-visaged by the judicial empowerment approach. In this scenario, the CJEU would be given excessive possibilities to widen the scope of EU law and accelerate the pace of integration at the expense of core member state poli-cies. The member states would have a difficult time justifying the CJEU’s decisions to their citizens, and in the worst-case scenario, it would lead to a severe legitimacy crisis for the EU and the national governments. A similar problem would arise if the national courts only exhibited behavior character-ized by sustained resistance. In this situation, the CJEU would only have restricted access to the national legal orders, and its ability to create and up-hold common legal standards would be limited. The member states, in con-trast, would maintain a high level of autonomy. In the worst-case scenario, the result of this unrestricted member state autonomy is a return to protec-tionist policies, which would make it impossible for the EU to uphold its internal market. Eventually, this would lead to the disintegration of the EU.

This thesis finds that neither of these two worst-case scenarios is an accu-rate portrayal of the national courts’ behavior in EU legal integration. Ad-mittedly, the single most common behavioral pattern of the national courts is to provide the CJEU with opportunities to extend EU law. However, when considering the national courts’ choices in conjunction, we find that courts frequently engage in a type of behavior that can be interpreted as an attempt to strike a fair balance between national concerns and EU legal integration. Instead of exacerbating the tensions between the member states and the EU

180

by bringing conflicts to a head, the findings show that the national courts can also try to minimize these tensions. The national courts may thereby contrib-ute to creating equilibrium between member state autonomy and EU legal integration. Such equilibrium is deemed necessary for upholding the EU’s legitimacy among the public.

For EU legal integration, the choices that national courts make in the pre-liminary ruling procedure imply that the gatekeeping role of national courts has more dimensions than that suggested by previous research. The national courts are not keeping the gate to the domestic legal sphere firmly shut to effectively shield domestic policies from EU legal intrusions and to obstruct legal harmonization within the internal market. Conversely, they are not providing the CJEU with an open invitation to accelerate the pace and ex-pand the scope of European integration. Rather, we find that national courts can also carefully balance the two dominant and often conflicting interests in EU legal integration, namely, the concerns of the member states and the uniform application of EU rules that countries have to accept as members of the EU.

Avenues for future research and concluding remarks European integration is one of the most significant political developments in modern-day Europe. Starting out as a coal and steel community, the EU has evolved into a full-scale political system. This thesis has focused on one of the actors that have been identified as important in this process of European integration: the national courts.

Previous research has done a remarkable job of probing how and why na-tional courts once became the CJEU’s main interlocutors in the early days of the EU. Much of the EU’s successful development has been attributed to its effective legal system, and the keystone of this system is the preliminary ruling procedure. The CJEU used this procedure to ensure that member states actually fulfilled their EU legal obligations. Individual actors were allowed to challenge national policies that they believed were contrary to EU law in national courts. National courts, in turn, referred these challenges to the CJEU and requested preliminary rulings. It was then an easy task for the CJEU to exercise de facto judicial review of national legislation and to in-form the referring courts whether or not the national policies violated EU law. Subsequently, national courts applied these rulings to the cases at hand, which forced the member state to comply with EU law. We know from these previous studies that national courts have, by and large, accepted the prelim-inary ruling procedure and the CJEU’s position as the ultimate authority on EU law. This thesis has examined how this acceptance finds expression in the national courts’ and judges’ ongoing contact with the CJEU.

181

The empirical and theoretical findings of this thesis have important impli-cations for future research regarding European legal integration. First, more attention should be paid to the micro foundations of the national courts’ ob-served macro behavior. The study undertaken in this thesis has demonstrated that Swedish judges are not reasoning as expected by the dominant theoreti-cal outlook in the literature on judicial politics and European legal integra-tion. That is, Swedish judges are not mainly guided by the logic of conse-quentialism and self-regarding preferences when making decisions in the preliminary ruling procedure. While they do think about expected conse-quences, their considerations are mainly other-regarding. Swedish judges show consideration for other actors and for the preliminary ruling procedure as such. Moreover, Swedish judges frequently reason according to the logic of appropriateness. However, to what extent national judges in other member states employ similar modes of reasoning is yet to be determined. There is thus a clear need for comparative empirical studies on judicial reasoning and decision-making in different European contexts. In addition, the thesis shows that judges working in lower courts have a completely different view of the expression of opinions in the requests for preliminary rulings than judges working in final instance courts. Lower-court judges are positive about the inclusion of opinions, while judges in the final-instance courts find the prac-tice of expressing opinions to be irreconcilable with being an impartial judge. The thesis finds that the explanation for these opposing standpoints lies in the varying informal professional practices that Swedish judges at different levels of the judicial hierarchy face. Future research should test whether this hypothesis, that informal professional practices shapes judicial behavior, holds in other EU member states.

Second, the thesis contributes to redefine the main theoretical controversy regarding what role national courts play in EU legal integration by identify-ing previously untheorized behavioral patterns. Importantly, this discovery has implications for the broader debate about European integration. These behavioral patterns demonstrate that the theoretical debate between judicial empowerment and sustained resistance does not capture all the relevant as-pects of national court behavior. The judicial empowerment approach ex-pects national courts to make choices that contribute to expanding the scope of EU legal integration. In contrast, the sustained resistance approach be-lieves that national courts defend national law and member state sovereignty. As was discussed in the preceding sections, if national courts were to exhibit only one of these two behavioral patterns, it would likely exacerbate the tension between EU legal integration and member state autonomy. In a worst-case scenario, it would lead to the breakdown of EU legal integration. However, this thesis has revealed that national court behavior is not limited to either supporting or resisting further EU legal integration. The findings show that national courts, in fact, frequently make choices that may alleviate parts of the inherent tension between, on the one hand, legitimate national

182

concerns and, on the other hand, the EU legal obligations that member states must accept in order for the EU to function efficiently. In other words, the thesis demonstrates that national courts may contribute to strike a politically acceptable balance between member state autonomy and EU legal integra-tion. Balancing these different interests is arguably necessary for the EU political project to maintain its legitimacy in the long run.

These findings raise important questions regarding the theoretical and normative implications of national court behavior. Are national courts able to perform the balancing act between national concerns and EU legal obliga-tions, and if so, is it desirable? Rytter and Wind (2011) think that it is both possible and necessary that national judiciaries take an active role in shaping the application of EU law. That is, national courts and judges should be ac-tive co-producers and co-interpreters of EU law, not passive consumers (Rytter and Wind 2011: 488). Rytter and Wind emphasize that if national courts do not engage in shaping the development of EU law, this means that they leave the stage to the CJEU to apply the EU law as it see fits (Rytter and Wind 2011: 500). Moreover, safeguarding member state autonomy by employing a reluctant approach to EU law is not an option. Courts from oth-er member states may seize the opportunity to impose their values on the interpretation of EU law by taking an active part in the preliminary ruling procedure. For an individual member state whose judiciary is reluctant to engage in the EU legal system, the results of the national courts behavior might be that salient national values and policies are overruled by the CJEU.

The key point is that the existing balance between member state autono-my and EU legal integration, the status quo, cannot be preserved unless na-tional courts are proactive in their contacts with the CJEU (Rytter and Wind 2011: 499-500). If national courts are in fact capable of balancing national concerns and EU legal harmonization, they may contribute to resolving the legitimacy problem that the EU faces. National courts, embedded in the do-mestic legal and political traditions, can focus the CJEU’s attention on na-tional concerns that otherwise would pass the EU court by. The national courts’ role is thus to inform the CJEU about what constitute the acceptable legal interpretation of EU law in the member state and, when necessary, argue in favor of what they, as judges embedded in a national context, find to be the politically acceptable balance between national political concerns and EU legal integration.

It is clearly the case that the type of engagement that Rytter and Wind call for places high demands on national courts. National judges should actively engage in the preliminary ruling procedure and understand the wider impli-cations of their choices, in particular in relation to EU legal integration. Fur-thermore, national judiciaries should be aware of fact that the EU faces a legitimacy deficit, and they should understand how their actions can contrib-ute to resolving or exacerbating this type of problem. Ultimately, the key issue is understanding that EU law is not something that is fixed but rather

183

an evolving set of norms that can be shaped by the courts that participate in the EU legal system.

The question is, however, whether national judges are aware of this ten-sion between national autonomy and EU legal integration and whether they think about the wider implications of their choices. Based on the findings in this thesis, the tentative answer is no. The study of national judicial reason-ing that has been undertaken in this thesis shows that few Swedish judges reflect upon how their choices may influence the course of EU legal integra-tion. Only one of the judges mention that it is important that national courts do not transfer national competence to the EU level by referring to the CJEU cases that do not fall under EU law. In contrast, what most judges consider are the formal and informal rules and how their decisions may influence the functioning of the EU legal system. In particular, judges think about how to avoid overburdening the CJEU. Whether these standpoints are unique to the Swedish judiciary or whether they mirror how most European judges under-stand their role in the preliminary ruling procedure remains to be seen. The national judges’ awareness of the tension between member state autonomy and EU legal harmonization, and how they respond to it, are important em-pirical question that deserves to be further examined.

From a normative perspective, the question is whether national courts and judges should be entrusted to strike a balance between EU legal integration and member state autonomy. Grimm (2015) and Scharpf (2009) would not agree with Rytter and Wind that national courts ought to engage in shaping the future of EU law. Leaving national courts in charge of balancing the level of EU legal integration may in fact result in new legitimacy problems. Grimm claims that many of the legal issues that end up before national courts and the CJEU are inherently political in character, and he fears that decision-making power of politically salient issues are being transferred from the legislative and executive institutions to the unelected courts (Grimm 2015: 470). Even if national courts are embedded in the member state context, they are still non-representative institutions, just like the CJEU. There is no means to hold national judges accountable if citizens would find that the courts have developed EU legal integration in a politically unac-ceptable direction. The EU does not need more judicial influence over policy making. Instead, Scharpf and Grimm believe that the solution to the EU’s legitimacy deficit is re-politicization.

Scharpf (2009) suggests that the institution that is best suited to handle the tension between member state autonomy and EU legal harmonization is the European Council. The proposed re-politicization means that the member states’ Heads of State or Government in the European Council should have the possibility to reverse the CJEU’s rulings. The argument is that the mem-ber states’ own experience of how the CJEU’s interpretations of EU law can distort national law makes them well equipped to determine the politically acceptable balance between legitimate national concerns and EU legal inte-

184

gration (Scharpf 2009: 199). However, whether giving the European Council this power will actually resolve the legitimacy deficit can be questioned. Perhaps the most obvious objection to this reform is that it will lead to an increased level of member state autonomy at the expense of a functioning EU legal system. The member states’ political leaders are unlikely to force any of their peers to abide by EU law. The political leaders know that if they would deny one of the member states to avoid an unpopular EU legal obliga-tion, they will sooner or later be forced to give up one of their own national policies. To avoid this outcome, national political leaders that want to remain in power are likely to engage in political horse-trading regarding what EU legal obligations the member states can accept. Another important objection to re-politicization is that it might increase the political conflicts between the member states, thereby negatively affecting the European Council’s ability to reach agreements. Both these scenarios are likely to undermine the EU legal system. European integration will be put to a halt, and in a worst-case scenario, disintegration will ensue.

Whether the solution to the EU’s legitimacy deficit is to be found in en-trusting national courts to be co-producers of EU law or in a re-politicization of certain EU legal issues is a question linked to the perennial debate in so-cial science research regarding how courts fit into democratic government and the benefits and perils of judicial review. Proponents of judicial review stress the importance of allowing impartial courts to check the actions of the legislative and executive branches to avoid that these institutions misuse their powers. The courts’ review of legislation is believed to protect liberal democratic principles, such as minority rights, from majority overreach (Dworkin 1978; Ely 1980: 135-179; Guarnieri et al. 2002). In contrast, crit-ics of judicial review find this institution to be ‘counter-majoritarian’ and irreconcilable with democratic government since it allows judges to overturn the decisions made by popularly elected politicians (Bellamy 2007; Waldron 2006).

In research regarding EU legal integration, this debate about the role of courts in democratic government has first and foremost focused on the CJEU’s behavior, including its alleged judicial activism and pro-integration preferences (Grimm 2015; Hirschl 2008; Scharpf 2009). However, the study undertaken in the present thesis has shown that the national courts’ behavior in the preliminary ruling procedure must be taken into account in this discus-sion. An important avenue for future research is thus to revisit the overarch-ing debate about how the relationship between law and politics in the EU should be organized in light of the findings presented in this thesis in order to carefully examine the normative implications of the national courts’ role as gatekeepers in EU legal integration. If striking a fair balance between member state autonomy and EU legal integration is desirable, we would like to know what conditions must be fulfilled for national courts to contribute to this outcome and, by extension, be instrumental in upholding the legitimacy

185

of the EU legal system. Conversely, we would like to understand under what conditions national courts would be unable to successfully perform this task. A particularly interesting discussion concerns to what extent national judges are willing to take on the responsibility of balancing different demands if it means that they become de facto political actors. Rytter & Wind’s (2011) proposal, that national courts ought to be active co-producers of EU legal norms, would place the judiciary in the political spotlight. The courts’ at-tempts to strike a balance between national concerns and EU harmonization are likely to lead to difficult trade-offs, and both citizens and EU institutions might criticize the courts’ decisions. Whether national courts are prepared to assume this political role and whether they should in fact be entrusted with such a task are key questions to explore if we are to fully understand the normative implications of the national courts’ role as gatekeepers in the EU legal system.

In conclusion, member state governments, EU institutions, sub-national actors and individual citizens are likely to have different opinions regarding what constitutes an acceptable level of European integration and what the future of the EU political project should be. This thesis shows that national courts and judges play an important role as gatekeepers in a deepened and widened EU in which the question of how to strike a proper balance between member state autonomy and European integration is widely debated. In light of the importance of this role, it is necessary to pay close attention to the courts’ and judges’ participation in the process of EU legal integration. Compared to the judicial politics literature regarding courts in the US, re-search on European courts and judges is still rather limited. Future research should therefore deploy more resources to explore the extent to which na-tional courts and judges contribute to promote, curtail, or reverse European integration through the choices they make in the preliminary ruling proce-dure.

186

8. References

Alter, Karen J (1996) The European Court's Political Power. West European Politics, 19, 458-487.

Alter, Karen J (1998a) Who Are the "Masters of the Treaty"?: European Governments and the European Court of Justice. International Organization, 52, 121-148.

Alter, Karen J (2000) The European Union's Legal System and Domestic Policy: Spillover or Backlash? International Organization, 54, 489-518.

Alter, Karen J (2009) The European Court's Political Power: Selected Essays, Oxford: Oxford University Press.

Alter, Karen J. (1998b) 'National Courts: The Critical Intermediaries in Legal Integration Alternative Explanations of National Judicial Behaviour in Legal Integration'. In: Slaughter, Anne-Marie, Stone Sweet, Alec & Weiler, Joseph H H (eds.) The European Courts and National Courts - Doctrine and Jurisprudence: Legal Change in Its Social Context. Oxford: Hart Publishing.

Alter, Karen J. (2001) Establishing the Supremacy of European Law : The Making of an International Rule of Law in Europe, Oxford: Oxford University Press.

Anton, Thomas J. (1980) Administered Politics: Elite Political Culture in Sweden, Boston, Mass.: Martinus Nijhoff publ.

Arrow, Kenneth J (1994) Methodological Individualism and Social Knowledge. The American Economic Review, 84, 1-9.

Bailey, Michael A. & Maltzman, Forrest (2011) The Constrained Court : Law, Politics, and the Decisions Justices Make, Princeton, N.J: Princeton University Press.

Bainbridge, Stephen M & Gulati, G Mitu (2002) How Do Judges Maximize - the Same Way Everybody Else Does - Boundedly: Rules of Thumb in Securities Fraud Opinions. Emory Law Journal, 51, 83.

Baum, Lawrence (1997) The Puzzle of Judicial Behavior, Ann Arbor: University of Michigan Press.

Baum, Lawrence (2006) Judges and Their Audiences: A Perspective on Judicial Behavior, Princeton: Princeton University Press.

Bebr, Gerhard (1983) The Rambling Ghost of “Cohn-Bendit”: Acte Clair and the Court of Justice. Common Market Law Review, 20, 439–472.

Bell, John (2006) Judiciaries within Europe: A Comparative Review, Cambridge: Cambridge Univ. Press.

Bellamy, Richard (2007) Political Constitutionalism : A Republican Defence of the Constitutionality of Democracy, Cambridge: Cambridge University Press.

Bergvall, Daniel (2007) Politik eller juridik? : Regeringens Agerande I Lavalmålet, Stockholm: Svenska institutet för europapolitiska studier (SIEPS).

Bernitz, Ulf (2010) Förhandsavgöranden av EU-domstolen : svenska domstolars hållning och praxis, Stockholm: SIEPS.

Bernitz, Ulf (2016) Förhandsavgöranden av EU-domstolen : Utvecklingen av svenska domstolars hållning och praxis 2010-2015, Stockholm: SIEPS.

187

Black, Ryan C & Owens, Ryan J (2016) Courting the President: How Circuit Court Judges Alter Their Behavior for Promotion to the Supreme Court. American Journal of Political Science, 60, 30-43.

Blauberger, Michael (2012) With Luxembourg in Mind… the Remaking of National Policies in the Face of ECJ Jurisprudence. Journal of European Public Policy, 19, 109-126.

Blauberger, Michael (2014) National Responses to European Court Jurisprudence. West European Politics, 37, 457-474.

Brace, Paul & Boyea, Brent D (2008) State Public Opinion, the Death Penalty, and the Practice of Electing Judges. American Journal of Political Science, 52, 360-372.

Brigham, John (1999) 'The Constitution of the Supreme Court'. In: Gillman, Howard & Clayton, Cornell (eds.) The Supreme Court in American Politics: New Institutionalist Interpretations. US: University Press of Kansas.

Broberg, Morten & Fenger, Niels (2013) Variations in Member States’ Preliminary References to the Court of Justice—Are Structural Factors (Part of) the Explanation? European Law Journal, 19, 488-501.

Burley, Anne-Marie & Mattli, Walter (1993) Europe before the Court: A Political Theory of Legal Integration. International organization, 47, 41-41.

Burnham, Peter (2008) Research Methods in Politics, Basingstoke: Palgrave Macmillan.

Burton, Steven J. (1992) Judging in Good Faith, Cambridge: Cambridge University Press.

Börzel, Tanja A (2002) Member State Responses to Europeanization. JCMS: Journal of Common Market Studies, 40, 193-214.

Börzel, Tanja a & Risse, Thomas (2003) 'Conceptualizing the Domestic Impact of Europe '. In: Featherstone, Kevin & Radaelli, Claudio M. (eds.) The Politics of Europeanization. Oxford Scholarship Online: Oxford Scholarship Online.

Caldeira, Gregory A & Gibson, James L (1992) The Etiology of Public Support for the Supreme Court. American journal of political science, 635-664.

Cameron, A Colin & Miller, Douglas L (2015) A Practitioner’s Guide to Cluster-Robust Inference. Journal of Human Resources, 50, 317-372.

Cappelletti, Mauro, Seccombe, Monica & Weiler, Joseph (1986) Integration through Law : Europe and the American Federal Experience. Vol. 1, Methods, Tools and Institutions, Book 1, a Political, Legal and Economic Overview, Berlin ;: de Gruyter.

Carrubba, Clifford J, Gabel, Matthew & Hankla, Charles (2008) Judicial Behavior under Political Constraints: Evidence from the European Court of Justice. American Political Science Review, 102, 435-452.

Carrubba, Clifford J & Murrah, Lacey (2005) Legal Integration and Use of the Preliminary Ruling Process in the European Union. International organization, 59, 399-418.

Carrubba, Clifford J. & Gabel, Matthew J. (2015) International Courts and the Performance of International Agreements: A General Theory with Evidence from the European Union, New York, NY: Cambridge University Press.

Casillas, Christopher J., Enns, Peter K. & Wohlfarth, Patrick C. (2011) How Public Opinion Constrains the U.S. Supreme Court. American Journal of Political Science, 55, 74-88.

Clayton, Cornell W & Gillman, Howard (1999) Supreme Court Decision-Making: New Institutionalist Approaches, Chicago: University of Chicago Press.

188

Conant, Lisa (2001) 'Europeanization and the Courts: Variable Patterns of Adaptation among National Judiciaries'. In: Cowles Green, Maria, Caporaso, James & Risse, Thomas (eds.) Transforming Europe: Europeanization and Domestic Change. Ithaca, N.Y: Cornell University Press.

Conant, Lisa (2002) Justice Contained: Law and Politics in the European Union, Ithaca, N.Y.: Cornell University Press.

Conant, Lisa (2007) Review Article: The Politics of Legal Integration*. JCMS: Journal of Common Market Studies, 45, 45-66.

Conant, Lisa (2013) 'Whose Agents? The Interpretation of International Law in National Courts'. In: Dunoff, Jeffrey L. & Pollack, Mark A. (eds.) Interdisciplinary Perspectives on International Law and International Relations: The State of the Art. Cambridge: Cambridge University Press.

Cook, Karen S. & Levi, Margaret (1990) The Limits of Rationality, Chicago: University of Chicago Press.

Coppedge, Michael (1999) Thickening Thin Concepts and Theories: Combining Large N and Small in Comparative Politics. Comparative Politics, 465-476.

Court of Justice of the European Union (2016) Recommendations to National Courts and Tribunals, in Relation to the Initiation of Preliminary Ruling Proceedings. Official Journal of the European Union, 59.

Court of Justice of the European Union (2017) Annual Report 2016 Judicial Activity. Publications Office of the European Union Communication Directorate, Publications and Electronic Media Unit.

Craig, Paul P (1998) 'Report on the United Kingdom'. In: Slaughter, Anne-Marie, Stone Sweet Alec & Weiler, Joseph H H (eds.) The European Courts and National Courts - Doctrine and Jurisprudence: Legal Change in Its Social Context. Oxford: Hart Publishing.

Crisp, Brian F, Escobar-Lemmon, Maria C, Jones, Bradford S, Jones, Mark P & Taylor-Robinson, Michelle M (2004) Vote‐Seeking Incentives and Legislative Representation in Six Presidential Democracies. Journal of Politics, 66, 823-846.

Dahl, Robert A (1957) Decision-Making in a Democracy: The Supreme Court as a National Policy-Maker. J. Pub. L., 6, 279.

Davies, Gareth (2012) Activism Relocated. The Self-Restraint of the European Court of Justice in Its National Context. Journal of European Public Policy, 19, 76-91.

Davis, Sue (1999) 'The Chief Justice and Judicial Decision-Making: The Institutional Basis for Leadership on the Supreme Court'. In: Clayton, Cornell & Gillman, Howard (eds.) Supreme Court Decision-Making: New Institutionalist Approaches. Chicago: University of Chicago Press.

Dehousse, Renaud (1998) The European Court of Justice: The Politics of Judicial Integration, New York: Macmillan.

Downs, Anthony (1957) An Economic Theory of Political Action in a Democracy. Journal of Political Economy, 65, 135-150.

Dunoff, Jeffrey L. & Pollack, Mark A. (2017) The Judicial Trilemma. American Journal of International Law, 111, 225-276.

Dworkin, Ronald (1978) Taking Rights Seriously, Cambridge: Harvard University Press.

Dyevre, Arthur (2010) Unifying the Field of Comparative Judicial Politics: Towards a General Theory of Judicial Behaviour. European Political Science Review: EPSR, 2, 297.

Elster, Jon (1982) The Case for Methodological Individualism. Theory and Society, 11, 453-482.

189

Ely, John Hart (1980) Democracy and Distrust : A Theory of Judicial Review, Cambridge, Mass.: Harvard University Press.

Epstein, Lee & Knight, Jack (1997) The Choices Justices Make, Washington: Sage. Epstein, Lee & Knight, Jack (2000) Toward a Strategic Revolution in Judicial

Politics: A Look Back, a Look Ahead. Political Research Quarterly, 53, 625-661.

Epstein, Lee & Kobylka, Joseph Fiske (1992) The Supreme Court and Legal Change : Abortion and the Death Penalty, Chapel Hill: University of North Carolina Press.

Epstein, Lee, Landes, William M & Posner, Richard A (2013) The Behavior of Federal Judges: A Theoretical and Empirical Study of Rational Choice: Harvard University Press.

Epstein, Lee & Lindquist, Stefanie A. (2017) The Oxford Handbook of U.S. Judicial Behavior, Oxford: Oxford University Press.

Epstein, Lee & Segal, Jeffrey A (2005) Advice and Consent: The Politics of Judicial Appointments, New York: Oxford University Press.

Epstein, Richard A (1990) The Independence of Judges: The Uses and Limitations of Public Choice Theory. BYU L. Rev., 827.

European Court of Justice (2005) Information Note on References from National Courts for a Preliminary Ruling Official Journal of the European Union, 48, 1-5.

European Union, EU (2012) Rules of Procedure of the Court of Justice. Official Journal of the European Union.

Falkner, Gerda & Treib, Oliver (2008) Three Worlds of Compliance or Four? The EU‐15 Compared to New Member States. JCMS: Journal of Common Market Studies, 46, 293-313.

Fehr, Ernst & Fischbacher, Urs (2002) Why Social Preferences Matter - the Impact of Non-Selfish Motives on Competition, Cooperation and Incentives. The economic journal, 112.

Fowler, James H & Kam, Cindy D (2007) Beyond the Self: Social Identity, Altruism, and Political Participation. Journal of Politics, 69, 813-827.

Friedman, Barry (2006) Taking Law Seriously. Perspectives on Politics, 4, 261-276. Fujii, Lee Ann (2013) 'Working with Interpreters'. In: Mosley, Layna (ed.)

Interview Research in Political Science. Ithaca: Cornell University Press. Garrett, Geoffrey (1995) The Politics of Legal Integration in the European Union.

International organization, 49, 171-181. Garrett, Geoffrey, Kelemen, R Daniel & Schulz, Heiner (1998) The European Court

of Justice, National Governments, and Legal Integration in the European Union. International organization, 52, 149-176.

Garrett, Geoffrey & Tsebelis, George (1996) An Institutional Critique of Intergovernmentalism. International organization, 50, 269-299.

Garrett R. And Barry, R. Weingast (1993) 'Ideas, Interests and Institutions: Constructing the European Community’s Internal Market'. In: Goldstein, Judith & Keohane, Robert (eds.) Ideas and Foreign Policy. Cornell University Press.

Gerring, John (2007) Case Study Research : Principles and Practices, New York: Cambridge University Press.

Gibson, James L (2012) Electing Judges: The Surprising Effects of Campaigning on Judicial Legitimacy, Chicago: University of Chicago Press.

Giles, Micheal W., Blackstone, Bethany & Viningjr, Richard L. (2008) The Supreme Court in American Democracy: Unraveling the Linkages between Public Opinion and Judicial Decision Making. The Journal of Politics, 70, 293-306.

190

Gillman, Howard (2001) What's Law Got to Do with It? Judicial Behavioralists Test the “Legal Model” of Judicial Decision Making. Law & Social Inquiry, 26, 465-504.

Goldstein, Kenneth (2002) Getting in the Door: Sampling and Completing Elite Interviews. PS: Political Science & Politics, 35, 669-672.

Golub, Jonathan (1996) The Politics of Judicial Discretion: Rethinking the Interaction between National Courts and the European Court of Justice. West European Politics, 19, 360-385.

Granger, Marie-Pierre (2006) States as Successful Litigants before the European Court of Justice: Lessons from the ‘Repeat Players’of European Litigation. Croatian Yearbook of European Law and Policy, 2, 27-49.

Green Cowles, Maria, Caporaso, James A. & Risse, Thomas (2001) Transforming Europe: Europeanization and Domestic Change, Ithaca, N.Y: Cornell University Press.

Grimm, Dieter (2015) The Democratic Costs of Constitutionalisation: The European Case. European Law Journal, 21, 460-473.

Guarnieri, Carlo, Pederzoli, Patrizia & Thomas, C. A. (2002) The Power of Judges : A Comparative Study of Courts and Democracy, Oxford: Oxford University Press.

Haas, Ernst B. (1958) The Uniting of Europe : Political, Social and Economical Forces 1950-1957, London: Stevens.

Hall, Melinda Gann (1995) Justices as Representatives:Elections and Judicial Politics in the American States. American Politics Quarterly, 23, 485-503.

Haverland, Markus (2000) National Adaptation to European Integration: The Importance of Institutional Veto Points. Journal of public policy, 20, 83-103.

Heath, Anthony (1976) Rational Choice and Social Exchange: A Critique of Exchange Theory, Cambridge: Cambridge University Press.

Helmke, Gretchen & Sanders, Mitchell S (2006) Modeling Motivations: A Method for Inferring Judicial Goals from Behavior. Journal of Politics, 68, 867-878.

Hessel, Josefina (2017). To Ask, or Not to Ask - on the Role of National Courts in the European Union Legal System Master's, Uppsala University.

Heumann, Milton (1978) Plea Bargaining : The Experience of Prosecutors, Judges, and Defense Attorneys, Chicago: Univ. of Chicago P.

Hirschl, Ran (2008) The Judicialization of Mega-Politics and the Rise of Political Courts. Annu. Rev. Polit. Sci., 11, 93-118.

Hix, Simon & Høyland, Bjørn (2011) The Political System of the European Union, third ed, Basingstoke: Palgrave Macmillan.

Holley, David (1999) Self-Interest and Beyond, New York: Paragon House. Hood, Christopher (2011) The Blame Game : Spin, Bureaucracy, and Self-

Preservation in Government, Princeton, N.J: Princeton University Press. Hornuf, Lars & Voigt, Stefan (2015) Analyzing Preliminary References as the

Powerbase of the European Court of Justice. European Journal of Law and Economics, 39, 287-311.

Hübner, Denise Carolin (2016) The ‘National Decisions’ Database (Dec. Nat): Introducing a Database on National Courts’ Interactions with European Law. European Union Politics, 17, 324-339.

Hönnige, Christoph (2009) The Electoral Connection: How the Pivotal Judge Affects Oppositional Success at European Constitutional Courts. West European Politics, 32, 963-984.

Hönnige, Christoph (2011) Beyond Judicialization: Why We Need More Comparative Research About Constitutional Courts. European Political Science, 10, 346-358.

191

Höpner, Martin & Schäfer, Armin (2012) Embeddedness and Regional Integration: Waiting for Polanyi in a Hayekian Setting. International Organization, 66, 429-455.

Jacobs, Francis G (2003) Judicial Dialogue and the Cross-Fertilization of Legal Systems: The European Court of Justice. Tex. Int'l LJ, 38, 547.

Jaremba, Urszula (2013) National Judges as EU Law Judges: The Polish Civil Law System: Martinus Nijhoff Publishers.

Kahn, Ronald (1999) 'Institutional Norms and Supreme Court Decision-Making: The Rehnquist Court on Privacy and Religion'. In: Clayton, Cornell W & Gillman, Howard (eds.) Supreme Court Decision-Making: New Institutionalist Approaches. Chicago: Chicago University Press.

Kelemen, R. Daniel & Schmidt, Susanne K. (2012) Introduction – the European Court of Justice and Legal Integration: Perpetual Momentum? Journal of European Public Policy, 19, 1-7.

King, Nigel & Horrocks, Christine (2010) Interviews in Qualitative Research, Los Angeles: SAGE.

Knight, Jack & Epstein, Lee (1998) The Choices Justices Make. Washington, DC: CQ Press.

Kohler-Koch, Beate & Eising, Rainer (1999) The Transformation of Governance in the European Union, London: Routledge.

Kurz, Mordecai (1978) Altruism as an Outcome of Social Interaction. The American Economic Review, 68, 216-222.

Kvale, Steinar & Brinkmann, Svend (2009) Interviews : Learning the Craft of Qualitative Research Interviewing, Los Angeles: Sage Publications.

Ladrech, Robert (2010) Europeanization and National Politics: Palgrave Macmillan. Landes, William M & Posner, Richard A (1975) The Independent Judiciary in an

Interest-Group Perspective. The Journal of Law and Economics, 18, 875-901. Landes, William M & Posner, Richard A (1976) Legal Precedent: A Theoretical and

Empirical Analysis. The Journal of Law and Economics, 19, 249-307. Larsson, Olof & Naurin, Daniel (2016) Judicial Independence and Political

Uncertainty: How the Risk of Override Affects the Court of Justice of the EU. International Organization, 70, 377-408.

Leech, Beth L (2002a) Asking Questions: Techniques for Semistructured Interviews. PS: Political Science & Politics, 35, 665-668.

Leech, Beth L (2002b) Interview Methods in Political Science. PS: political science & politics, 35, 663-664.

Leijon, Karin & Karlsson, Christer (2013) Nationella Domstolar Som Politiska Aktörer:-Främjare av Rättslig Integration eller Försvarare av Nationella Intressen? Statsvetenskaplig Tidskrift, 115, 5-34.

Lijphart, Arend (1999) Patterns of Democracy : Government Forms and Performance in Thirty-Six Countries, New Haven, Conn.: Yale University Press.

Long, J. Scott (1997) Regression Models for Categorical and Limited Dependant Variables, Thousand Oaks, Calif: SAGE.

Macintyre, Alasdair C. (1988) Whose Justice? Which Rationality?, London: Duckworth.

Maddex, Robert L (2007) Constitutions of the World: Cq Press. Maher, Imelda (1994) National Courts as European Community Courts. Legal

Studies, 14, 226-243. Maher, Imelda (1998) Community Law in the National Legal Order: A Systems

Analysis. JCMS: Journal of Common Market Studies, 36, 237-254.

192

Maltzman, Forrest, Spriggs, James F & Wahlbeck, Paul J (1999) 'Strategy and Judicial Choice: New Institutionalist Approaches to Supreme Court Decision-Making'. In: Clayton, Cornell W & Gillman, Howard (eds.) Supreme Court Decision-Making: New Institutionalist Approaches. University of Chicago Press.

Mancini, G Federico (1989) Making of a Constitution for Europe, The. Common Market L. Rev., 26, 595.

Mansbridge, Jane J (1990) Beyond Self-Interest: University of Chicago Press. March, James G & Olsen, Johan P (1995) Democratic Governance: Free Press. March, James G & Olsen, Johan P (2009) The Logic of Appropriateness, Oslo:

Arena. The Center for European Studies. March, James G. & Olsen, Johan P. (1984) The New Institutionalism:

Organizational Factors in Political Life. The American Political Science Review, 78, 734-749.

March, James G. & Olsen, Johan P. (1989) Rediscovering Institutions : The Organizational Basis of Politics, New York: Free Press.

Marsh, David & Stoker, Gerry (2010) Theory and Methods in Political Science: Palgrave Macmillan.

Martinsen, Dorte Sindbjerg (2011) Judicial Policy-Making and Europeanization: The Proportionality of National Control and Administrative Discretion. Journal of European Public Policy, 18, 944-961.

Marvell, Thomas B. (1978) Appellate Courts and Lawyers : Information Gathering in the Adversary System, Westport, Conn.: Greenwood P.

Mattli, Walter & Slaughter, Anne-Marie (1998) 'The Role of National Courts in the Process of European Integration: Accounting for Judicial Preferences and Constraints'. In: Slaughter, Anne-Marie; Stone Sweet, Alec & Weiler, Joseph H H (Eds.) (ed.) The European Courts and National Courts - Doctrine and Jurisprudence: Legal Change in Its Social Context. Oxford: Hart Publishing.

Mattli, Walter & Slaughter, Anne-Marie (1995) Law and Politics in the European Union: A Reply to Garrett. International organization, 49, 183-183.

Mattli, Walter & Slaughter, Anne-Marie (1998) Revisiting the European Court of Justice. International organization, 52, 177-209.

Maveety, Nancy L (2009) The Pioneers of Judicial Behavior: University of Michigan Press.

Mayoral, Juan A (2012) Overcoming the Legal Iron Curtain: Similarities and Differences in the Use of Preliminary References between New and Old Member States. Perspectives on Federalism, 4.

Mayoral, Juan A (2017) In the CJEU Judges Trust: A New Approach in the Judicial Construction of Europe. JCMS: Journal of Common Market Studies, 55, 551-568.

Mcguire, Kevin T & Stimson, James A (2004) The Least Dangerous Branch Revisited: New Evidence on Supreme Court Responsiveness to Public Preferences. Journal of Politics, 66, 1018-1035.

Miceli, Thomas J & Coşgel, Metin M (1994) Reputation and Judicial Decision-Making. Journal of Economic Behavior & Organization, 23, 31-51.

Mishler, William & Sheehan, Reginald S. (1996) Public Opinion, the Attitudinal Model, and Supreme Court Decision Making: A Micro-Analytic Perspective. The Journal of Politics, 58, 169-200.

Moravcsik, Andrew (1993) Preferences and Power in the European Community: A Liberal Intergovernmentalist Approach. JCMS: Journal of Common Market Studies, 31, 473-524.

193

Murphy, Walter F. (1964) Elements of Judicial Strategy, Chicago, Ill.: Univ. of Chicago Press.

Nergelius., Joakim (2001) 'North and South: Can the Nordic States and the European Continent Find Each Other in the Constitutional Area - or Are They Too Different'. In: Scheinin, Martin (ed.) Welfare State and Constitutionalism : Nordic Perspectives. Nord (København), 99-0634809-7 ; 2001:5: Nordic Council of Ministers.

Novak, Stéphanie (2013) The Silence of Ministers: Consensus and Blame Avoidance in the Council of the European Union. JCMS: Journal of Common Market Studies, 51, 1091-1107.

Nyikos, Stacy A. (2000). The European Court of Justice and National Courts: Strategic Interaction within the EU Judicial Process. University of Virginia.

Nyikos, Stacy A (2006) Strategic Interaction among Courts within the Preliminary Reference Process–Stage 1: National Court Preemptive Opinions. European Journal of Political Research, 45, 527-550.

O´Brien, David M (1999) 'Institutional Norms and Supreme Court Opinions: On Reconsidering the Rise of Individual Opinions'. In: Clayton, Cornell W & Gillman, Howard (eds.) Supreme Court Decision-Making: New Institutionalist Approaches. University of Chicago Press.

Obermaier, Andreas J (2008) The National Judiciary—Sword of European Court of Justice Rulings: The Example of the Kohll/Decker Jurisprudence. European Law Journal, 14, 735-752.

Orren, Karen & Skowronek, Stephen (1994) Beyond the Iconography of Order: Notes for a New Institutionalism. The dynamics of American politics, 311-330.

Ostrom, Elinor (1991) Rational Choice Theory and Institutional Analysis: Toward Complementarity. American political science review, 85, 237-243.

Pampel, Fred C. (2000) Logistic Regression : A Primer, Thousand Oaks, Calif.: SAGE.

Pellegrina, Lucia Dalla & Garoupa, Nuno (2013) Choosing between the Government and the Regions: An Empirical Analysis of the Italian Constitutional Court Decisions. European Journal of Political Research, 52, 558-580.

Pepper, John V (2002) Robust Inferences from Random Clustered Samples: An Application Using Data from the Panel Study of Income Dynamics. Economics Letters, 75, 341-345.

Perry, H. W. (1991) Deciding to Decide : Agenda Setting in the United States Supreme Court, Cambridge, Mass.: Harvard Univ. Press.

Pescatore, Pierre (1974) The Law of Integration : Emergence of a New Phenomenon in International Relations, Based on the Experience of the European Communities, Leiden.

Pescatore, Pierre (2015) The Doctrine of" Direct Effect": An Infant Disease of Community Law. European Law Review, 40, 135-153.

Pollack, Mark A (1997) Delegation, Agency, and Agenda Setting in the European Community. International organization, 51, 99-134.

Pollack, Mark A. (2013) The New EU Legal History: What's New, What's Missing? American University International Law Review, 28, 1257.

Posner, Richard A (1995) Overcoming Law: Harvard University Press. Posner, Richard A. (1993) What Do Judges and Justices Maximize? (the Same

Thing Everybody Else Does). Supreme Court Economic Review, 3, 1-41. Raffaelli, Rosa (2012) Dissenting Opinions in the Supreme Courts of the Member

States. European Parliament: European Parliament. Rasmussen, Hjalte (1986) On Law and Policy in the European Court of Justice : A

Comparative Study in Judicial Policymaking, Dordrecht: Nijhoff.

194

Rehder, Britta (2007) What Is Political About Jurisprudence?: Courts, Politics and Political Science in Europe and the United States. MPIfG - Max-Planck-Institut für Gesellschaftsforschung.

Rehder, Britta (2014) 'Judicial Politics in Europe'. In: Magone, José María (ed.) Routledge Handbook of European Politics. Milton Park, Abingdon, Oxon ;: Routledge.

Rosas, Allan (2007) The European Court of Justice in Context: Forms and Patterns of Judicial Dialogue. Eur. J. Legal Stud., 1, 121.

Rytter, Jens Elo & Wind, Marlene (2011) In Need of Juristocracy? The Silence of Denmark in the Development of European Legal Norms. International Journal of Constitutional Law, 9, 470-504.

Scharpf, Fritz W (2009) Legitimacy in the Multilevel European Polity. European Political Science Review, 1, 173-204.

Schauer, Frederick (1999) Incentives, Reputation, and the Inglorious Determinants of Judicial Behavior. U. Cin. L. Rev., 68, 615.

Schauer, Frederick (2000) Incentives, Reputation, and the Inglorious Determinants of Judicial Behavior. University of Cincinnati Law Review, 68, 615-636.

Schmidt, Susanne K (1998) Commission Activism: Subsuming Telecommunications and Electricity under European Competition Law. Journal of European Public Policy, 5, 169-184.

Schmidt, Vivien A (2004) The European Union: Democratic Legitimacy in a Regional State? JCMS: Journal of Common Market Studies, 42, 975-997.

Schneider, Martin R. (2005) Judicial Career Incentives and Court Performance: An Empirical Study of the German Labour Courts of Appeal. European Journal of Law and Economics, 20, 127-144.

Schwartz, Edward P, Spiller, Pablo T & Urbiztondo, Santiago (1994) A Positive Theory of Legislative Intent. Law and Contemporary Problems, 57, 51-74.

Segal, Jeffrey A (1997) Separation-of-Powers Games in the Positive Theory of Congress and Courts. American Political Science Review, 91, 28-44.

Segal, Jeffrey A (1999) 'The Supreme Court Deference to Congress: An Examination of the Marksist Model'. In: Clayton, Cornell W & Gillman, Howard (eds.) Supreme Court Decision-Making: New Institutionalist Approaches. University of Chicago Press.

Segal, Jeffrey A & Cover, Albert D (1989) Ideological Values and the Votes of US Supreme Court Justices. American Political Science Review, 83, 557-565.

Segal, Jeffrey A & Spaeth, Harold J (2002) The Supreme Court and the Attitudinal Model Revisited: Cambridge University Press.

Segal, Jeffrey A. & Spaeth, Harold J. (1993) The Supreme Court and the Attitudinal Model, Cambridge: Cambridge Univ. Press.

Shapiro, Martin (1992) 'The European Court of Justice'. In: Sbragia, Alberta M. (ed.) Euro-Politics : Institutions and Policymaking in the New European Community. Washington, D.C.: Brookings Institution.

Shapiro, Martin & Stone Sweet, Alec (2002) On Law, Politics, & Judicialization, Oxford: Oxford University Press.

Skowronek, Stephen (1995) Order and Change. Polity, 28, 91-96. Slaughter, Anne Marie, Stone Sweet, Alec & Weiler, Joseph (1998) The European

Court and National Courts - Doctrine and Jurisprudence : Legal Change in Its Social Context, Oxford: Hart.

195

Slot, Piet Jan (1998) Cases C-157/94, Commission V. Netherlands; C-158/94, Commission V. Italy; C-159/94, Commission V. France; C-160/94, Commission V. Spain; C-189/95, Harry Franzén; Judgments of 23 October 1997, Full Court,[1997] Ecr I-5699, I-5789, I-5815, I-5851, I-5909. Common Market Law Review, 35, 1183-1203.

Smith, Rogers M (1988) Political Jurisprudence, the “New Institutionalism,” and the Future of Public Law. American Political Science Review, 82, 89-108.

Sourdin, Tania & Burstyner, Naomi (2014) Justice Delayed Is Justice Denied. Victoria UL & Just. J., 4, 46.

Spiller, Pablo T & Gely, Rafael (1992) Congressional Control or Judicial Independence: The Determinants of US Supreme Court Labor-Relations Decisions, 1949-1988. The RAND journal of Economics, 463-492.

Spiller, Pablo T. & Gely, Rafael (2008) 'Strategic Judicial Decision-Making'. In: Whittington, Keith E. & Kelemen, R. Daniel (eds.) The Oxford Handbook of Law and Politics. Oxford: Oxford University Press.

Staton, Jeffrey K (2006) Constitutional Review and the Selective Promotion of Case Results. American Journal of Political Science, 50, 98-112.

Stein, Eric (1981) Lawyers, Judges, and the Making of a Transnational Constitution. American journal of international law, 75, 1-27.

Steinmo, Sven, Thelen, Kathleen Ann & Longstreth, Frank (1992) Structuring Politics : Historical Institutionalism in Comparative Analysis, Cambridge ;: Cambridge University Press.

Stone Sweet, Alec & Sandholtz, Wayne (1997) European Integration and Supranational Governance. Journal of European Public Policy, 4, 297-317.

Stout, Lynn A (2001) Judges as Altruistic Hierarchs. Wm. & Mary L. Rev., 43, 1605. Sweet, Alec Stone & Brunell, Thomas L (1998a) Constructing a Supranational

Constitution: Dispute Resolution and Governance in the European Community. American Political Science Review, 92, 63-81.

Sweet, Alec Stone & Brunell, Thomas L. (1998b) The European Court and the National Courts: A Statistical Analysis of Preliminary References, 1961–95. Journal of European Public Policy, 5, 66-97.

Tallberg, Jonas (2002) Delegation to Supranational Institutions: Why, How, and with What Consequences? West European Politics, 25, 23-46.

Tamanaha, Brian Z (2009) Beyond the Formalist-Realist Divide: The Role of Politics in Judging, Princeton: Princeton University Press.

Tocqueville, Alexis De (2010) Democracy in America: Historical-Critical Edition of De La Démocratie En Amérique. Translated from the French by James T. Schleifer a bilingual french-english edition. Indianapolis: Liberty Fund.

Tridimas, George & Tridimas, Takis (2004) National Courts and the European Court of Justice: A Public Choice Analysis of the Preliminary Reference Procedure. International review of Law and Economics, 24, 125-145.

Tsebelis, George (1995) Decision Making in Political Systems: Veto Players in Presidentialism, Parliamentarism, Multicameralism and Multipartyism. British journal of political science, 25, 289-325.

Ugland, Trygve (2000) Impact of Europeanization on Nordic Alcohol Control Policies: A Discussion of Processes and National Differences. Journal of European Social Policy, 10, 59-67.

Waldron, Jeremy (2006) The Core of the Case against Judicial Review. The Yale Law Journal, 115, 1346-1406.

196

Wallerman, Anna (2018) 'Referring Court Influence in the Preliminary Ruling Procedure: The Swedish Example'. In: Derlén, Mattias & Lindholm, Johan (eds.) The Court of Justice of the European Union : Multidisciplinary Perspectives. Oxford, UK Hart Publishing.

Vallinder, Torbjörn (1994) The Judicialization of Politics—a World-Wide Phenomenon: Introduction. International Political Science Review, 15, 91-99.

Wasserfallen, Fabio (2010) The Judiciary as Legislator? How the European Court of Justice Shapes Policy-Making in the European Union. Journal of European Public Policy, 17, 1128-1146.

Weaver, R Kent (1986) The Politics of Blame Avoidance. Journal of public policy, 6, 371-398.

Weiler, J.H.H (1994) A Quiet Revolution : The European Court of Justice and Its Interlocutors. Comparative Political Studies, 26.

Weiler, Joseph Hh (1991) The Transformation of Europe. Yale Law Journal, 2403-2483.

Whittington, Keith E (2000) Once More Unto the Breach: Postbehavioralist Approaches to Judicial Politics. Law & Social Inquiry, 25, 601-634.

Whittington, Keith E. & Kelemen, R. Daniel (2008) The Oxford Handbook of Law and Politics, Oxford: Oxford University Press.

Wind, Marlene (2010) The Nordics, the EU and the Reluctance Towards Supranational Judicial Review. Journal of Common Market Studies, 48, 1039-1063.

Wind, Marlene, Martinsen, Dorte Sindbjerg & Rotger, Gabriel Pons (2009) The Uneven Legal Push for Europe Questioning Variation When National Courts Go to Europe. European Union Politics, 10, 63-88.

Vink, Maarten, Claes, Monica & Arnold, Christine (2009) Explaining the Use of Preliminary References by Domestic Courts in EU Member States: A Mixed-Method Comparative Analysis. 11th Biennial Conference of the European Union Studies Association. Marina del Rey.

Witte, Bruno De, Wind, Marlene, Mayoral, Juan A., Jaremba, Urszula & Podstawa, Karolina (2016) National Courts and EU Law : New Issues, Theories and Methods, Cheltenham, UK: Edward Elgar Publishing.

Volcansek, Mary L (1986) Judicial Politics in Europe: P. Lang.

197

Legal decisions cited

Court of Justice of the European Union (CJEU) • Judgment of the Court of 5 February 1963. NV Algemene Transport- en Expe-

ditie Onderneming van Gend & Loos v Netherlands Inland Revenue Admin-istration. Tariefcommissie - Netherlands. C-26/62. ECLI:EU:C:1963:1

• Judgment of the Court of 15 July 1964. Flaminio Costa v E.N.E.L. Giudice conciliatore di Milano - Italy. C-6/64. ECLI:EU:C:1964:6

• Judgment of the Court of Justice 6 October 1982, Srl CILFIT and Lanificio di Gavardo SpA v Ministry of Health, C-283/81, EU:C:1982:335.

• Judgment of the Court of 10 April 1984. Sabine von Colson and Elisabeth Kamann v Land Nordrhein-Westfalen. C-14/83. ECLI:EU:C:1984:153

• Judgment of the Court of 4 October 1991. The Society for the Protection of Unborn Children Ireland Ltd v Stephen Grogan and others. C-159/90 ECLI:EU:C:1991:378.

• Judgment of the Court of 19 October 1995. The Queen v Secretary of State for Health, ex parte Cyril Richardson. C-137/94. EU:C:1995:342

• Judgment of the Court of Justice 14 December 1995, Criminal proceedings against Giorgio Domingo Banchero, C-387/93, EU:C:1995:439.

• Judgment of the Court of 23 October 1997. Criminal proceedings against Harry Franzén. Landskrona tingsrätt - Sweden. C-189/95. ECLI:EU:C:1997:504

• Judgment of the Court of 19 March 1998, The Queen v Minister of Agriculture, Fisheries and Food, ex parte Compassion in World Farming Ltd. C-1/96. EU:C:1998:113.

• Judgment of the Court of 30 April 1998 , August De Vriendt v Rijksdienst voor Pensioenen, C-377/96, EU:C:1998:183

• Opinion of Mr. Advocate General Léger 20 September 2001, DaimlerChrysler AG v Land Baden-Württemberg, C-324/99, EU:C:2001:459.

• Judgment of the Court of Justice 25 October 2001, Urlaubs- und Lohnaus-gleichskasse der Bauwirtschaft v.Turiprata Construções Civil Ldª, C-54/98, EU:C:2001:564.

• Judgment of the Court of 21 November 2002. Robelco NV v Robeco Groep NV. C-23/01. EU:C:2002:706

• Judgment of the Court of Justice 6 October 2005, Medion AG v Thomson mul-timedia Sales Germany & Austria GmbH., C-120/04, EU:C:2005:594.

• Judgment of the Court of 22 May 2008. Skatteministeriet v Ecco Sko A/S. C-165/07. EU:C:2008:302

198

Appendices

Appendix A – Units of analysis This appendix includes the following additional information mainly related to chapters 3 and 4: • Frequency of cases (data set and population) by member state • Frequency of cases (data set and population) by year • The distribution of cases with regard to their degree of political sensitivi-

ty • The share of cases with opinions by member state

Table A.1. The frequency and percentage of cases in the data set and in the popula-tion by member state

Member state Frequency in sample

Percent in sample

Frequency in population

Percent in population

Austria 42 8.9 410 7.4 Belgium 47 10.0 448 8.1 Bulgaria 4 0.9 55 1.0 Czech Republic 1 0.2 27 0.5 Denmark 14 3.0 127 2.3 Estonia 2 0.4 12 0.2 Finland 6 1.3 79 1.4 France 30 6.4 434 7.9 Germany 101 21.5 1215 22.1 Greece 12 2.6 135 2.5 Hungary 1 0.2 64 1.2 Ireland 2 0.4 45 0.8 Italy 59 12.6 893 16.2 Latvia 2 0.4 25 0.4 Lithuania 1 0.2 13 0.2 Luxembourg 7 1.5 53 1.0 Netherlands 49 10.4 488 8.9 Poland 8 1.7 49 0.9 Portugal 4 0.9 96 1.7 Romania 1 0.2 46 0.8 Spain 23 4.9 271 4.9 Sweden 11 2.3 99 1.8 UK 43 9.2 415 7.5 Total 470 100 5509 100

199

Table A.2. The frequency and percentage of cases in the data set and in the popula-tion by year Years Frequency

in sample Percent in

sample Frequency in popu-

lation

Percent in population

1992 14 2.9 162 2.9 1993 13 2.8 204 3.7 1994 24 5.1 203 3.7 1995 34 7.2 251 4.6 1996 13 2.8 256 4.6 1997 18 3.8 239 4.3 1998 18 3.8 264 4.8 1999 19 4.0 255 4.6 2000 19 4.0 224 4.1 2001 23 4.9 237 4.3 2002 17 3.6 216 3.9 2003 21 4.5 210 3.8 2004 23 4.9 249 4.5 2005 17 3.6 221 4.0 2006 21 4.5 251 4.6 2007 21 4.5 265 4.8 2008 29 6.2 288 5.2 2009 26 5.5 302 5.5 2010 30 6.4 385 7.1 2011 27 5.7 423 7.7 2012 43 9.2 404 7.3

Total 470 100 5509 100

200

Table A.3. The distribution of cases with regard to their degree of political sensitivi-ty (1992-2012)

Countries

Politically non-sensitive

Somewhat politically sensitive

Politically sensitive

Highly politically sensitivee

Total

Austria 6 (14 %) 11 (26 %) 24 (57 %) 1 (2 %) 42 (100 %) Belgium 8 (17 %) 13 (28 %) 25 (53 %) 1 (2 %) 37 (100 %) Bulgaria 0 (0 %) 2 (50 %) 2 (50 %) 0 (0 %) 4 (100 %) Czech Republic 0 (0 %) 1 (100 %) 0 (0 %) 0 (0 %) 1 (100 %) Denmark 5 (36 %) 6 (43 %) 3 (21 %) 0 (0 %) 14 (100 %) Estonia 0 (0 %) 2 (100 %) 0 (0 %) 0 (0 %) 2 (100 %) Finland 2 (33 %) 3 (50 %) 1 (17 %) 0 (0 %) 6 (100 %) France 3 (10 %) 9 (30 %) 17 (57 %) 1 (3 %) 30 (100 %) Germany 11 (11 %) 37 (37 %) 46 (46 %) 7 (7 %) 101 (100 %) Greece 0 (0 %) 5 (42 %) 5 (42 %) 2 (17 %) 12 (100 %) Hungary 0 (0 %) 1 (100 %) 0 (0 %) 0 (0 %) 1 (100 %) Ireland 1 (50 %) 0 (0 %) 1 (50 %) 0 (0 %) 2 (100 %) Italy 1 (2 %) 21 (36 %) 37 (63 %) 0 (0 %) 59 (100 %) Latvia 0 (0 %) 1 (50 %) 1 (50 %) 0 (0 %) 2 (100 %) Lithuania 0 (0 %) 1 (100 %) 0 (0 %) 0 (0 %) 1 (100 %) Luxembourg 1 (14 %) 1 (14 %) 4 (57 %) 1 (14 %) 7 (100 %) Netherlands 5 (10 %) 16 (33 %) 24 (49 %) 4 (8 %) 49 (100 %) Poland 0 (0 %) 3 (38 %) 5 (63 %) 0 (0 %) 8 (100 %) Portugal 0 (0 %) 3 (75 %) 1 (25 %) 0 (0 %) 4 (100 %) Romania 0 (0 %) 0 (0 %) 1 (100 %) 0 (0 %) 1 (100 %) Spain 2 (9 %) 6 (26 %) 14 (61 %) 1 (4 %) 23 (100 %) Sweden 2 (18 %) 4 (36 %) 5 (45 %) 0 (0 %) 11 (100 %) UK 2 (5 %) 19 (44 %) 20 (47 %) 2 (5 %) 43 (100 %) All 49 (10 %) 165 (35 %) 236 (50 %) 20 (4 %) 470 (100 %) EU 6 29 (10 %) 97 (33 %) 153 (52 %) 14 (5 %) 293 (100 %)

EU 9 20 (13 %) 57 (36 %) 74 (47 %) 6 (4 %) 157 (100 %)

201

Table A.4. The share of cases with opinions by member state 1992-2012 Member State

Number of cases

with opinions Total number of

cases Austria 20 (48 %) 42 Belgium 22 (47 %) 47 Bulgaria 2 (50 %) 4 Czech Republic 1 (100 %) 1 Denmark 4 (29 %) 14 Estonia 1 (50 %) 2 Finland 3 (50 %) 6 France 8 (27 %) 30 Germany 61 (60 %) 101 Greece 5 (42 %) 12 Hungary 0 (0 %) 1 Ireland 2 (100 %) 2 Italy 29 (49 %) 59 Latvia 1 (50 %) 2 Lithuania 0 (0 %) 1 Luxembourg 2 (29 %) 7 Netherlands 22 (45 %) 49 Poland 4 (50 %) 8 Portugal 0 (0 %) 4 Romania 0 (0 %) 1 Spain 16 (70 %) 23 Sweden 3 (27 %) 11 UK 20 (47 %) 43 All 226 (48 %) 470

EU 6 144 49 % 293 EU 9 73 47 % 157

202

Appendix B – Regression tables and statistics This appendix includes the following additional information mainly related to the analysis in chapter 5. • Variable description • The average share of the population (1992-2012) that consider the EU to

be “a bad thing” • Regressing the variable Euroscepticism on the variable Majoritarian

democracy • Additional models on the referral of politically sensitive cases. These

models include: Logistic regression models without clustered standard errors Logistic regression models (Restricted sample/Including dummy variables for certain member states) OLS models Ordered logit model • Additional models on the expression of opinions supporting national

law: Logistic regression models without clustered standard errors Logistic regression models (Restricted sample/Including dummy variables for certain member states) OLS models Examining the behavior of courts of appeal

Table B.1: Variable description N mean sd min max Dependent variables Type of case 470 0.545 0.499 0 1 Type of opinion 182 0.352 0.479 0 1 Independent variables Inter-court comp: Final instance 470 0.360 0.480 0 1 Inter-court comp: Appeal and final instance 470 0.606 0.489 0 1 Majoritarian democracy 470 0.157 0.365 0 1 Euroscepticism (proportions) 470 0.145 0.0730 0.0300 0.380 Experience 470 593.3 502.2 0 1.885 Concrete judicial review 470 0.728 0.446 0 1

203

Table B.2. The average share of the population (1992-2012) that consider the EU to be “a bad thing”

Member state Percentage Member state Percentage

Ireland 6 Slovenia 10 Luxembourg 6 Portugal 11 Slovakia 6 Germany 12 The Netherlands 7 Czech Republic 12 Poland 7 France 15 Romania 7 Latvia 17 Estonia 8 Hungary 17 Bulgaria 8 Denmark 19 Belgium 10 Finland 21 Greece 10 Austria 22 Spain 10 UK 25 Italy 10 Sweden 30 Lithuania 10

Source: The Eurobarometer

Table B.3. Regressing the variable Euroscepticism on the variable Majoritarian democracy (OLS)

(1) Majoritarian democracy 0.12*** (0.02) Constant 0.13*** (0.01) Observations 470 R-squared 0.37

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

204

Table B.4. The referral of cases with a high degree of political sensitivity (logistic regression, odds ratio) (1) (2) (3) (4) (5) (6) (7) Final instance 0.86 (0.17) First instance 1.45* 1.50** (0.28) (0.29) Euroscepticism 0.14 1.18 (0.18) (1.94) Majoritarian democracy 0.55** 0.44** (0.14) (0.16) Experience 1.00 1.00 (0.00) (0.00) Concrete judicial review 1.03 0.81 (0.21) (0.19) Constant 1.26** 1.04 1.60** 1.32*** 1.18 1.17 1.49 (0.15) (0.12) (0.33) (0.13) (0.17) (0.21) (0.47) Prob > chi2 0.4345 0.0519 0.1190 0.0181 0.9120 0.8811 0.0483 Pseudo R2 0.00094 0.0058 0.0038 0.0086 1.89e-

05 3.45e-05

0.0172

Observations 470 470 470 470 470 470 470 Standard errors in parentheses

*** p<0.01, ** p<0.05, * p<0.1 Note: These are the same models as was reported in Table 5.2. but without clustered standard errors. No substantial change in results.

205

Table B.5. The referral of cases with a high degree of political sensitivity (logistic regression, odds ratio)

(1) (2) (3) (4) First instance 1.51*** 1.50** 1.47*** 1.48*** (0.20) (0.26) (0.21) (0.21) Euroscepticism 1.10 3.46 1.02 1.35 (1.78) (4.41) (1.66) (2.27) Majoritarian democracy 0.46** 0.43*** 0.43** 0.42** (0.16) (0.13) (0.15) (0.14) Concrete judicial review 0.85 0.87 0.84 0.83 (0.12) (0.12) (0.12) (0.12) Experience 1.00 1.00 1.00*** 1.00*** (0.00) (0.00) (0.00) (0.00) Germany 0.82 (0.19) MS after 2004 0.59 (0.24) Constant 1.39 1.09 1.60** 1.54* (0.40) (0.30) (0.37) (0.37) Prob > chi2 0.0775 0.0119 0.0006 0.0002 Pseudo R2 0.0176 0.0209 0.0190 0.0187 Countries 23 22 23 15 Observations 470 369 470 450

*** p<0.01, ** p<0.05, * p<0.1 Note: Models 1-4 further explores the relationship between the independent variables and the referral of cases with a high degree of political sensitivity that was presented in Table 5.2. Model (7). Model (1) adds a dummy variable for German courts – “Germany” (non-clustered standard errors in parentheses). Model (2) excludes all observations from German courts and thereby reduces the number of observations (country clustered robust standard errors in paren-theses). Model (3) adds a dummy variable for the member state courts that joined the EU after 2004 – “MS after 2004” (country clustered robust standard errors in parentheses). Model (4) excludes the observations from the member states that joined the EU after 2004 and thereby reduces the number of observations. As can be seen, there are no substantial changes in result.

206

Table B.6. The referral of cases with a high degree of political sensitivity (OLS) (1) (2) (3) (4) (5) (6) (7) Final instance -0.04 (0.03) First instance 0.09*** 0.10*** (0.03) (0.03) Euroscepticism -0.49* 0.04 (0.28) (0.39) Majoritarian democracy -0.15* -0.20** (0.07) (0.08) Experience 0.00 -0.00* (0.00) (0.00) Concrete judicial review 0.01 -0.05 (0.04) (0.03) Constant 0.56*** 0.51*** 0.62*** 0.57*** 0.54*** 0.54*** 0.60*** (0.02) (0.02) (0.03) (0.02) (0.04) (0.03) (0.06) R2 0.00 0.01 0.01 0.01 0.00 0.00 0.02 Countries 23 23 23 23 23 23 23 Observations 470 470 470 470 470 470 470

Robust standard errors in parentheses (clustered at member state level) *** p<0.01, ** p<0.05, * p<0.1

Note: Models 1-7 are Ordinary Least Square (OLS) models that explore the relationship between the independent variables and the referral of cases with a high degree of political sensitivity that was presented in Table 5.2. When the dependent variable is binary, the OLS regression coefficient shows the decrease or increase in the predicted probability of having a characteristic because of a one-unit change in the independent variable. As can be seen, the estimations of the hypothesized relationships with the OLS models yield only minor differ-ences in the results (compared to the Logistic regression models) and these differences were not found to be of substantial importance.

207

Table B.7. The referral of cases with a high degree of political sensitivity (Ordered logit)

(1) First instance 1.50*** (0.19) Euroscepticism 1.83 (3.23) Majoritarian democracy 0.42** (0.17) Experience 0.74* (0.13) Concrete judicial review 1.00 (0.00) Constant cut1 0.09*** (0.02) Constant cut2 0.69* (0.15) Constant cut3 19.39*** (7.77) Prob > chi2 Pseudo R2 0.0124 Countries 23 Observations 470

Robust standard errors in parentheses (clustered at member state level) *** p<0.01, ** p<0.05, * p<0.1

Estimating an ordered logit model yields no substantive difference in result. Assumption underlying ordered logistic regression: the relationship between each pair of outcome groups is the same. Since it is difficult to make this assumption in relation to the four categories of cases (politically non-sensitive, somewhat politically sensitive, politically sensitive and highly politically sensitive) an ordered logit model has not been used in the main analysis.

208

Table B.8. Support for national legislation (logistic regression, odds ratios) (1) (2) (3) (4) (5) (6) (7) (8) (9)

Odds ratio

Odds ratio

Odds ratio

Odds ratio

Odds ratio

Odds ratio

Odds ratio

Odds ratio

Odds ratio

Final instance 2.58*** 2.58*** (0.83) (0.88) First instance 0.49** 0.48** (0.16) (0.16) Euroscepticism 1.21 0.28 0.25 (2.60) (0.79) (0.72) Majoritarian democracy

1.12 1.40 1.03

(0.51) (0.94) (0.66) High degree of pol. sensitivity

0.83 0.85 0.81

(0.27) (0.28) (0.27) Experience 1.00 1.00 1.00 (0.00) (0.00) (0.00) Concrete judicial review

0.82 1.05 0.97

(0.29) (0.44) (0.39) Constant 0.38*** 0.72* 0.53* 0.53**

* 0.61* 0.67 0.63 0.51 1.27

(0.08) (0.14) (0.19) (0.09) (0.15) (0.17) (0.19) (0.31) (0.75) Prob > chi2 0.0033 0.0248 0.9296 0.7979 0.5639 0.2899 0.5815 0.1448 0.3660 Pseudo R2 0.0367 0.0213 3.30e-05 0.0002

78 0.00141

0.00475 0.00129

0.0405 0.0277

Observations 182 182 182 182 182 182 182 182 182 Standard errors in parentheses

*** p<0.01, ** p<0.05, * p<0.1 Note: These are the same models as was reported in Table 5.3. but without clustered standard errors. No substantial change in results.

209

Table B.9. Support for national legislation (logistic regression, odds ratio) (1) (2) (3) (4) (5) (6) (7) (8) Final instance 2.55**

* 2.50** 2.52**

* 2.42**

(0.88) (1.11) (0.86) (0.84) First instance 0.49* 0.38* 0.49* 0.51* (0.20) (0.19) (0.19) (0.20) Euroscepticism 0.31 0.40 0.30 0.21 0.34 0.69 0.32 0.23 (0.65) (1.03) (0.68) (0.51) (0.78) (1.93) (0.82) (0.60) Majoritarian democracy

0.92 1.18 1.05 1.10 1.27 1.49 1.42 1.46

(0.41) (0.57) (0.44) (0.48) (0.71) (0.99) (0.77) (0.83) Concrete judicial review

0.86 1.03 0.93 0.94 0.96 1.07 1.02 1.02

(0.19) (0.20) (0.17) (0.18) (0.22) (0.30) (0.23) (0.23) Experience 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) High degree of pol. sensitivity

0.79 0.88 0.82 0.79 0.83 0.91 0.86 0.83

(0.32) (0.46) (0.32) (0.32) (0.32) (0.44) (0.32) (0.31) Germany 1.59 1.48 (1.03) (0.87) MS after 2004 2.10 1.83 (1.67) (1.48) Constant 1.52 0.88 1.17 1.23 0.60 0.33* 0.49 0.54 (1.03) (0.42) (0.59) (0.63) (0.39) (0.19) (0.24) (0.26) Prob > chi2 - 0.4688 0.0782 0.1330 - 0.1967 0.0177 0.0355 Pseudo R2 0.0297 0.0360 0.0301 0.0235 0.0419 0.0342 0.0421 0.0348 Countries 18 17 18 14 18 17 18 14 Observations 182 135 182 177 182 135 182 177

*** p<0.01, ** p<0.05, * p<0.1 Note: Models 1-8 further explores the relationship between the independent variables and expressing support for national law that was presented in Table 5.3. Models (8) and (9). Models (1) and (5) add a dummy variable for German courts – “Germany” (clustered standard errors in parentheses). Models (2) and (6) exclude all observations from German courts and the number of observations is reduced (country clustered robust standard errors in parenthe-ses). Model (3) and (7) add a dummy variable for the member state courts that joined the EU after 2004 – “MS after 2004” (country clustered robust standard errors in parentheses). Mod-els (4) and (8) exclude the observations from the member states that joined the EU after 2004 and the number of observations is reduced. As can be seen, there are no substantial changes in result.

210

Table B.10. Support for national legislation (OLS) (1) (2) (3) (4) (5) (6) (7) (8) (9) Final in-stance

0.22***

0.22**

(0.07) (0.08) First in-stance

-0.16* -0.16*

(0.08) (0.08) Eurosceptic-ism

0.04 -0.28 -0.30

(0.38) (0.54) (0.49) Majoritarian democracy

0.03 0.08 0.01

(0.07) (0.12) (0.09) High degree of pol. sensitivity

-0.04 -0.04 -0.05

(0.08) (0.08) (0.09) Experience -0.00 -0.00 -0.00 (0.00) (0.00) (0.00) Concrete judicial review

-0.05 0.01 -0.01

(0.05) (0.05) (0.04) Constant 0.27*

** 0.42***

0.35***

0.35***

0.38***

0.40***

0.39***

0.34***

0.54***

(0.02) (0.04) (0.07) (0.02) (0.06) (0.05) (0.05) (0.11) (0.11) R-squared 0.05 0.03 0.00 0.00 0.00 0.01 0.00 0.05 0.04 Countries 18 18 18 18 18 18 18 18 18 Observat-ions

182 182 182 182 182 182 182 182 182

Robust standard errors in parentheses (clustered at member state level) *** p<0.01, ** p<0.05, * p<0.1

Note: Models 1-7 are Ordinary Least Square (OLS) models that explore the relationship between the independent variables and the expression of opinions supporting national law that was presented in Table 5.3. When the dependent variable is binary, the OLS regression coef-ficient shows the decrease or increase in the predicted probability of having a characteristic because of a one-unit change in the independent variable. As can be seen, the estimations of the hypothesized relationships with the OLS models yield only minor differences in the re-sults (compared to the logistic regression models).

Examining the behavior of courts of appeal

To further investigate the behavior of the courts of appeal in relation to courts of final instance and courts of first instance two logistic regression models were estimated. The results are reported in Table B.10 below. In Model (1) the baseline category is courts of final instance. In Model (2) the baseline category is courts of first instance. As can be seen, the behavior of courts of appeal (the variable “Appeal”) does not differ significantly from

211

either courts of final instance or courts of first instance. That is, courts of appeal are not less likely than courts of final instance to express support for national law. Similarly, courts of appeal are not more likely than first in-stance courts to express support for national law.

Table B.11. Support for national legislation (logistic regression, odds ratios) (1) (2) Independent variables Baseline

Final Baseline

First First instance (base: Final) 0.36*** (0.14) Appeal (base in Model (1): Final / base in Model (2): First) 0.45 1.25 (0.23) (0.70) Final instance (base: First ) 2.79*** (1.08) Euroscepticism 0.23 0.23 (0.53) (0.53) Majoritarian democracy 1.37 1.37 (0.74) (0.74) Politically sensitivity 0.84 0.84 (0.32) (0.32) Concrete judicial review 1.06 1.06 (0.21) (0.21) Experience 1.00 1.00 (0.00) (0.00) Constant 1.38 0.50 (0.70) (0.24) Prob > chi2 0.0113 0.0113 Pseudo R2 0.0416 0.0416 Countries 18 18 Observations 182 182

Robust standard errors in parentheses (clustered at member state level) *** p<0.01, ** p<0.05, * p<0.1

212

Appendix C – Information to the respondents and the interview questionnaire This appendix contains the following: • Information to the respondents (English translation) • The interview guide (English translation) • List of respondents

Information to the respondents:

My name is Karin Leijon and I am a PhD student at the Department of Gov-ernment (Uppsala University). I am writing a thesis dealing with the role of national courts in the preliminary ruling procedure and I would like to ask you about your perspective on the preliminary ruling procedure. For exam-ple, why did you decide to refer cases to the CJEU? There is some previous research on the subject, both from a legal and a political science perspective, but little is known about how the procedure functions from the perspective of national judges. The reason that you are among the selected interviewees is that you have request a preliminary ruling from the Court of Justice of the European Union in case xx and therefore have direct experience of the pro-cedure. Overall, I will talk to about 20 judges that have been involved in requesting preliminary rulings from the CJEU.

If you have the opportunity to take part in the study, I am flexible both regarding the place and time of the interview. I have the possibility to visit your city during week xx. The interview is expected to last about an hour. If you have any questions, please feel free to contact me by sending an e-mail: karin.leijon@statsvet.uu.se or by calling me: 070-XX XX XXX. If you want to know more about my research my supervisor, Professor Christer Karlsson, can be reached at: christer.karlsson@statsvet.uu.se or by phone 018-471 XX XX. Below, I describe my research project in detail.

Information about the research project: "The choice courts make: the role of national courts in the preliminary ruling procedure". The overall pur-pose of my dissertation is to describe and explain the role of national courts in EU legal integration. The first part of the dissertation is a statistical survey of approximately 470 cases referred from national courts around the EU member states to the European Court of Justice in the context of the prelimi-nary ruling procedure. Based on this material, I study the types of referral (policy area, parties, degree of political sensitivity) and whether the national courts choose to include their own views (opinions) on how EU law should be understood in relation to national law in the specific objective.

The interview guide

213

Can you describe the circumstance of the case/cases that you have referred to the CJEU? Specific question about a specific case: How come you decided to request a preliminary ruling? How would you describe the case? Could it be considered sensitive in any way? How was the request for a preliminary ruling formulated? How did you de-cide what questions to include? Did you discuss the questions with your colleagues? How long time did it take to formulate the questions? In some cases that have been referred to the CJEU I have noticed that judges express opinions regarding how they think that the case should be resolved, it that something you did in this case? Why/why not? How long time did it take for you to get the answer form the CJEU? Was it useful? Have there been other EU legal cases in which you have contemplated ask-ing the CJEU for a preliminary ruling but decided against it? Why? Is there anything you would like to add?

Table C.1. The respondents and their mode of reasoning Respondspond-

ent Gender Court level Choice 1 Choice 2 Combination

1 Female First (admin) Self/Other Formal Mixed 2 Male First (admin) Self/Other Formal Mixed 3 Male First (civil) Formal Self Mixed 4 Male First (civil) Other Self Consequentialism 5 Male Appeal (admin) Self/Other Informal Mixed 6 Male Appeal (admin) Other Informal Mixed 7 Female Appeal (admin) Formal - - 8 Male Appeal (civil) Formal - - 9 Male Appeal (civil) Self/Self Formal Mixed 10 Female Appeal (civil) Other Informal Mixed 11 Female Final (admin) Formal - - 12 Male Final (admin) Formal - - 13 Female Final (admin) Self Informal Mixed 14 Female Final (admin) Other Informal Mixed 15 Male Final (civil) Other Informal Mixed 16 Female Final (civil) Other Informal Mixed 17 Female Final (civil) Formal Informal Appropriateness 18 Male Final (civil) Other Other Consequentialism 19 Female Final (civil) Other Informal Mixed 20 Male Final (civil) Formal Informal Appropriateness

Statsvetenskapliga föreningen i Uppsala

Skrifter utgivna genom Axel Brusewitz

1. Axel Brusewitz, Studier över riksdagen och utrikespolitiken. I. Hemliga utskottet i 1809 års författning. 144 sid. 1933.

2. Axel Brusewitz, Studier över riksdagen och utrikespolitiken. II. Nordiska utrikesnämnder i komparativ belysning. 266 sid. 1933.

3. Gunnar Heckscher, Konung och statsråd i 1809 års författning. Regeringsmaktens inre gestaltning under det nya statsskickets första halvsekel. 484 sid. 1933.

4. De nordiska ländernas statsråd. Studier över deras sociala sammansättning. Utarbetade inom seminariet för statskunskap i Uppsala. 234 sid. 1935.

5. Sven Lindman, Parlamentarismens införande i Finlands statsförfattning. 85 sid. 1935. 6. Elis Håstad, Regeringssättet i den schweiziska demokratien. 735 sid. 1936. 7. Studier över den svenska riksdagens sociala sammansättning. Utarbetade inom Semi-

nariet för statskunskap i Uppsala. 233 sid. 1936. 8. Ragnar Simonsson, Millerands presidentur. En studie över presidentmakt och parla-

mentarism i Frankrike. 319 sid. 1938. 9. Gunnar Heckscher, Svensk konservatism före representationsreformen. I. Den historiska

skolans genombrott i Sverige. 257 sid. 1939. 10. Arne Björnberg, Parlamentarismens utveckling i Norge efter 1905. 400 sid. 1939. 11. Arne Wåhlstrand, 1905 års ministärkriser. 434 sid. 1941. 12. Festskrift till professor skytteanus Axel Brusewitz. Utgiven till 60-årsdagen den 9 juni

1941 av vänner och lärjungar. 506 sid. 1941. 13. Axel Brusewitz, Studier över riksdagen och utrikespolitiken. III. Utrikesfrågors behand-

ling i den svenska riksdagen. 155 sid. 1941. 14. Sixten Björkblom, Södermanlands läns landsting. Sammansättning, organisation och verk-

samhet. 467 sid., 62 sid. ill. 1942. 15. Edward Thermaenius, Adressdebatten i underhuset. En studie i parlamentarisk teknik

och tradition. 236 sid. 1942. 16. Lennart Linnarsson, Riksrådens licentiering. En studie i frihetstidens parlamentarism.

360 sid. 1943. 17. Gunnar Heckscher, Svensk konservatism före representationsreformen. II. Doktrin och

politik 1840–1865. 271 sid. 1943. 18. Torsten Petré, Svenska regeringen och de norska frågorna 1884–1886. 132 sid. 1944. 19. Olle Nyman, Krisuppgörelsen mellan socialdemokraterna och bondeförbundet 1933. 114

sid. 1944. 20. Statsvetenskapliga studier. Till Statsvetenskapliga föreningens i Uppsala tjugofemårsdag

7 november 1944. Av medlemmar. 701 sid. 1944. 21. Torsten Petré, Ministären Themptander. 305 sid., I sid. ill. 1945. 22. Carl-Erik Höjer, Le régime parlamentaire belge à 1918–1940. 373 sid. 1946. 23. Ragnar Edenman, Socialdemokratiska riksdagsgruppen 1903–1920. En studie i den

svenska riksdagens partiväsen. 298 sid., I sid. ill. 1946. 24. Arne Wåhlstrand, Allmänna valmansförbundets tillkomst. 104 sid. 1946. 25. Gunnar Gerdner, Det svenska regeringsproblemet 1917–1920. Från majoritetskoalition

till minoritetsparlamentarism. 396 sid., 10 sid. ill. 1946. 26. Nils Andrén, Den klassiska parlamentarismens genombrott i England. 361 sid., 9 sid. ill.

1947. 27. Olle Nyman, Svensk parlamentarism 1932–36. Från minoritetsparlamentarism till majori-

tetskoalition. 570 sid., 14 sid. ill. 1947. 28. Arne Wåhlstrand, Regeringsskiftena 1900 och 1902. 238 sid., 10 ill. 1947.

Skrifter utgivna genom C. A. Hessler

30. Gert Hornwall, Regeringskris och riksdagspolitik 1840–1841. 465 sid., 6 sid. ill. 1951. 31. Eyvind Bratt, Småstaterna i idéhistorien. En studie i äldre statsdoktriner. 231 sid. 1951. 32. Ragnar Andersson, Svenska Dagbladet och det politiska livet 1897–1918. 460 sid., 4 sid.

ill. 1952. 33. Arne Wåhlstrand, Karlstadskonferensen 1905. Protokoll och aktstycken. 474 sid., 4 sid.

ill. 1953. 34. Sten-Sture Landström, Svenska ämbetsmäns sociala ursprung. 179 sid. 1954. 35. Gunnar Gerdner, Parlamentarismens kris i Sverige vid 1920-talets början. 568 sid., ill.

1954. 36. Carl Arvid Hessler, Stat och religion i upplysningstidens Sverige. 242 sid. 1956. 37. Olle Nyman, Högern och kungamakten 1911–1914. Ur borggårdskrisens förhistoria. 408

sid. 1957. 38. Elias Berg, The Historical Thinking of Charles A. Beard. 83 sid. 1957. 39. Olle Gellerman, Staten och jordbruket 1867–1918. 391 sid. 1958. 40. Olle Nyman, Der Westdeutsche Föderalismus. Studien zum Bonner Grundgesetz. 266

sid. 1960. 41. Mats Kihlberg och Donald Söderlind, Två studier i svensk konservatism 1916–1922.

268 sid. 1961. 42. Nils Elvander, Harald Hjärne och konservatismen. Konservativ idédebatt i Sverige

1865–1922. 530 sid. 1961. 43. Clarence Nilsson, Sam Stadener som kyrkopolitiker. 392 sid. 1964. 44. Carl Arvid Hessler, Statskyrkodebatten. 462 sid. 1964. 45. Olof Wennås, Striden om latinväldet. Idéer och intressen i svensk skolpolitik under

1800-talet. 412 sid. 1966. 46. Leif Lewin, Planhushållningsdebatten. 564 sid. 1967. 47. Artur Mägi, Das Staatslebens Estlands während seiner Selbständigkeit. 1. Das Regie-

rungssystem. 327 sid. 1967. 48. Lars Rudebeck, Party and People. A Study of Political Change in Tunisia. 275 sid. 1967. 49. Bo Jonsson, Staten och malmfälten. En studie i svensk malmfältspolitik omkring sekel-

skiftet. 388 sid. 1969. 50. Idéer och ideologier. Sex studier utgivna till Statsvetenskapliga föreningens i Uppsala

femtioårsdag den 7 november 1969. 375 sid. 1969. 51. Terry Carlbom, Högskolelokaliseringen i Sverige 1950–1965. En studie i politiskt infly-

tande. 309 sid. 1970. 52. Leif Lewin, Folket och eliterna. En studie i modern demokratisk teori. 260 sid. 1970. 53. Reidar Larsson, Theories of Revolution. From Marx to the First Russian Revolution.

381 sid. 1970. 54. Sverker Gustavsson, Debatten om forskningen och samhället. En studie i några teore-

tiska inlägg under 1900-talet. 359 sid. 1971. 55. Ragnar Simonsson, Frankrikes författningar. 617 sid. 1971. 56. Leontios Ierodiakonou, The Cyprus Question. 311 sid. 1971. 57. Evert Vedung, Unionsdebatten 1905. En jämförelse mellan argumenteringen i Sverige

och Norge. 475 sid. 1971. 58. Hans Sjöberg, De Gaulles Europatankar. 148 sid. 1972. 59. Georg Andrén, Den unge Churchill. 100 sid. 1972. 60. Leif Lewin, Bo Jansson, Dag Sörbom, The Swedish Electorate 1887–1968. 293 sid.

1972. 61. Barry Holmström, Koreakriget i Svensk debatt. 353 sid. 1972.

Skrifter utgivna genom Leif Lewin

62. Leif Lewin, Statskunskapen, ideologierna och den politiska verkligheten. 28 sid. 1972. 63. Jan Åke Dellenbrant. Reformists and Traditionalists. A Study of Soviet Discussions

about Economic reform, 1960–1965. 163 sid. 1972. 64. Peter Wallensteen, Structure and War. On International Relations 1920–1968. 269 sid.

1973. 65. Georg Landberg, Hans Järta som landshövding i Kopparbergs län 1812–1822. 295 sid.

1973. 66. Bengt Bogärde, Svenska folkhögskolans lärarförening och staten 1902–1970. En studie i

organisationsinflytande. 187 sid. 1974. 67. Michael Ståhl, Ethiopia. Political Contradictions in Agricultural Development. 186 sid.

1974. 68. Roger Henning, Staten som företagare. En studie av Statsföretag AB:s mål, organisation

och effektivitet. 183 sid. 1974. 69. Lars Nord, Nonalignment and Socialism. Yugoslav Foreign Policy in Theory and Prac-

tice. 306 sid. 1974. 70. Karl-Göran Algotsson, Från katekestvång till religionsfrihet. Debatten om religionsun-

dervisningen i skolan under 1900-talet. 538 sid. 1975. 71. Onesimo Silveira, Africa South of the Sahara: Party Systems and Ideologies of Social-

ism. 212 sid. 1976. 72. Carl-Johan Westholm, Ratio och universalitet. John Stuart Mill och dagens demokrati-

debatt. 266 sid. 1976. 73. Leif Lewin, Hur styrs facket? Om demokratin inom fackföreningsrörelsen. 245 sid. 1977. 74. Kodbok till fackföreningsundersökningen. 128 sid. 1977. 75. Axel Hadenius, Facklig organisationsutveckling. En studie av Landsorganisationen i

Sverige. 214 sid. 1976. 76. Leif Lewin, Åsiktsjournalistiken och den fackliga demokratin. 94 sid. 1977. 77. Michael Sahlin, Neo-Authoritarianism and the Problem of Legitimacy. A General Study

and a Nigerian Example. 240 sid. 1977. 78. Lars-lnge Ström, Idrott och politik. Idrottsrörelsens politiska roll i Tredje Riket, DDR

och Förbundsrepubliken. 201 sid. 1977. 79. Gabriele Winai Ström. Development and Dependence in Lesotho. The Enclave of South

Africa. 192 sid. 1978. 80. Erik Åsard, LO och löntagarfondsfrågan. En studie i facklig politik och strategi. 210 sid.

1978. 81. Maurice Georgi. La Politique de l’URSS au Proche-Orient 1954–1964, “attitude et ana-

lyse”. 178 sid. 1978. 82. Lars Ohlsson, Lokal förvaltning i Sovjetunionen. Sovjeterna: Vision och verklighet. 328

sid. 1979. 83. Leif Lewin, Det politiska spelet. 155 sid. 1979. 84. Olof Petersson, Regeringsbildningen 1978. 177 sid., 8 sid. ill. 1978. 85. Evert Vedung, Kärnkraften och regeringen Fälldins fall. 233 sid., 8 sid. ill. 1979. 86. Abu Baker El Obeid, The Political Consequences of the Addis Ababa Agreement. 167

sid. 1980. 87. Roger Henning, Partierna och Stålverk 80. En studie av industripolitiskt beslutsfattande.

183 sid. 1980. 88. Axel Hadenius, Spelet om skatten. Rationalistisk analys av politiskt beslutsfattande. 301

sid. 1981. 89. Erik Amnå, Planhushållning i den offentliga sektorn? Budgetdialogen mellan regering

och förvaltning under efterkrigstiden. 235 sid. 1981. 90. Olof Petersson, Väljarna och världspolitiken. 133 sid. 1982.

91. Sven-Erik Svärd, Icke-beslut – maktens andra ansikte. En analys av begrepp, metod och empiriska tillämpningar. 145 sid. 1982.

92. Sven Eliaeson, Bilden av Max Weber. En studie i samhällsvetenskapens sekularisering. 336 sid. 1982.

93. Jörgen Wedin, Spelet om trafikpolitiken. 344 sid. 1982. 94. Barry Holmström, Äganderätt och brukningsansvar. Idéer och intressen i svensk jord-

politik. 271 sid. 1983. 95. Axel Hadenius, Medbestämmandereformen. 208 sid. 1983. 96. Joachim Schäfer, Från tillväxt till trygghet. Tio års debatt om strukturpolitikens mål och

medel. 292 sid. 1984. 97. Helle Kanger, Human Rights in the U.N. Declaration. 208 sid. 1984. 98. Axel Hadenius, Roger Henning, Barry Holmström, Tre studier i politiskt beslutsfattande.

188 sid. 1984. 99. Jörgen Hermansson, Kommunism på svenska? SKP/VPK:s idéutveckling efter Komin-

tern. 388 sid. 1984. 100. Barbro Lewin, Johan Skytte och de skytteanska professorerna. 232 sid., ill. 1985. 101. Agnete Nissborg, Danmark mellan Norden och Väst. 151 sid. 1985. 102. Stig Arne Nohrstedt, Tredje världen i nyheterna. 250 sid. 1986. 103. Festskrift till professor skytteanus Carl Arvid Hessler. Utgiven till 80-årsdagen den 10

februari 1987 av hans lärjungar genom Leif Lewin. 342 sid. 1987. 104. Anders Weidung, Frihandelns dilemma. 186 sid. 1987. 105. Gunvall Grip, Vill du frihet eller tvång? Svensk försäkringspolitik 1935–1945. 268 sid.

1987. 106. Adolphe Lawson, Les Accords de Lancaster House en 1979: l’aboutissement de deux

décennies de débats sur les conditions de l’indépendence du Zimbabwe. 209 sid. 1988. 107. Michael Nydén, Kris i massupplaga. En studie i hållbarhet, relevans och neutral presen-

tation i fyra svenska dagstidningars rapportering om de internationella oljebolagens skuld till oljekrisen 1973–1974. 269 sid. 1988.

108. Alexander Davidson, Two Models of Welfare: the Origins and Development of the Wel-fare State in Sweden and New Zealand 1888–1988. 432 sid. 1989.

109. Jörgen Hermansson, Spelteorins nytta. Om rationalitet i vetenskap och politik. 385 sid. 1990.

110. Karin Hadenius, Jämlikhet och frihet: Politiska mål för den svenska grundskolan. 342 sid. 1990.

111. Ove Nordenmark, Aktiv utrikespolitik. Sverige – södra Afrika 1969–1987. 187 sid. 1991.

112. Stefan Björklund, Forskningsanknytning genom disputation. 161 sid. 1991. 113. Terry Carlbom, Lycka och nytta. Jeremy Benthams idévärld. 107 sid., 2 ill. 1991. 114. Fredrika Björklund, Samförstånd under oenighet. Svensk säkerhetspolitisk debatt under

det kalla kriget. 225 sid. 1992. 115. Lars Niklasson, Bör man lyda lagen? En undersökning av den offentliga maktens legiti-

mitet. 214 sid. 1992. 116. Mats Lundström, Politikens moraliska rum. En studie i F. A. Hayeks politiska filosofi.

248 sid. 1993. 117. Nils Karlson, The State of State. An Inquiry Concerning the Role of Invisible Hands in

Politics and Civil Society. 234 sid. 1993. 118. Nils Öberg, Gränslös rättvisa eller rättvisa inom gränser? Om moraliska dilemman i väl-

färdsstaters invandrings- och invandrarpolitik. 190 sid. 1994. 119. Leif Lewin, Politiskt ledarskap. Sex forskningsprojekt om konsten att leda ett land. 217

sid. 1994.

120. Torsten Svensson, Socialdemokratins dominans. En studie av den svenska socialdemo-kratins partistrategi. 332 sid. 1994.

121. Christina Bergquist, Mäns makt och kvinnors intressen. 226 sid. 1994. 122. PerOla Öberg, Särintresse och allmänintresse: Korporatismens ansikten. 304 sid. 1994. 123. Anders Lindbom, Medborgarskapet i välfärdsstaten. Föräldrainflytande i skandinavisk

grundskola. 248 sid. 1995. 124. Stefan Björklund, En författning för disputationen. 234 sid. 1996. 125. Mikael Axberg, Rättvisa och demokrati. Om prioriteringar i sjukvården. 240 sid. 1997. 126. Carina Gunnarson, Den förförande makten. Frankrikes Afrikapolitik under Mitterrand.

233 sid. 1997. 127. Åsa Lundgren, Europeisk identitetspolitik. EUs demokratibistånd till Polen och Turkiet.

220 sid. 1998. 128. Per Löwdin, Det dukade bordet. Om partierna och de ekonomiska kriserna. 496 sid.

1998. 129. Jan Teorell, Demokrati eller fåtalsvälde? Om beslutsfattande i partiorganisationer. 391

sid. 1998. 130. Barry Holmström, Domstolar och demokrati. Den tredje statsmaktens politiska roll i

England, Frankrike och Tyskland. 495 sid. 1998. 131. Shirin Ahlbäck, Att kontrollera staten. Den statliga revisionens roll i den parlamentariska

demokratin. 335 sid. 1999. 132. Torkel Nyman, Kommittépolitik och parlamentarism. Statsminister Boström och rikspoli-

tiken 1891–1905. En studie av den svenska parlamentarismens framväxt. 247 sid., 7 sid. ill. 1999.

133. PerOla Öberg, Jörgen Hermansson, Per-Åke Berg. I väntan på solnedgången. Arbets-livsfondens tillkomst och verksamhet, en studie av makt och rationalitet inom förvalt-ningspolitiken. 342 sid. 1999.

134. Anders Berg, Staten som kapitalist. Marknadsanpassningen av de affärsdrivande verken 1976–1994. 278 sid. 1999.

135. Sverker Härd, Den godtyckliga demokratin. En studie av olika metoder att tillgodose kravet på proportionell rättvisa. 216 sid. 1999.

136. Stefan Olsson, Den svenska högerns anpassning till demokratin. 328 sid. 2000. 137. Anna-Carin Svensson, In the Service of the European Union. The Role of the presidency

in Negotiating the Amsterdam Treaty 1995–97. 233 sid. 2000. 138. Ann-Cathrine Jungar, Surplus Majority Government. A Comparative Study of Italy and

Finland. 384 sid. 2000. 139. Leif Lewin, Svenskt kynne. En konferens anordnad av Humanistisk-samhällsvenskapliga

vetenskapsområdet vid Uppsala universitet. 124 sid. 2000. 140. Ingrid Widlund, Paths to Power and Patterns of Influence. The Dravidian Parties in

South Indian Politics. 404 sid. 2000. 141. Carl Melin, Makten över trafikpolitiken. Korporatism, lobbying och opinionsbildning

inför 1998 års transportpolitiska beslut. 301 sid. 2000. 142. Joakim Johansson, SAF och den svenska modellen. En studie av uppbrottet från förvalt-

ningskorporatismen 1982–91. 248 sid. 2000. 143. Johan Matz, Constructing a Post-Soviet International Political Reality. Russian Foreign

Policy towards the Newly Independent States 1990–95. 288 sid. 2001. 144. Torsten Svensson, Marknadsanpassningens politik. Den Svenska Modellens förändring

1980–2000. 311 sid. 2001. 145. Robin Travis, Air Transport Liberalisation in the European Community 1987–1992: A

Case of Integration. 181 sid. 2001. 146. Christer Karlsson, Democracy, Legitimacy and the European Union. 314 sid. 2001.

147. Charles F. Parker, Controlling Weapons of Mass Destruction: An Evaluation of Interna-tional Security Regime Significance. 294 sid. 2001.

148. Cecilia Garme, Newcomers to Power. How to sit on someone else’s throne. Socialists conquer France in 1981; non-socialists conquer Sweden in 1976. 226 sid. 2001.

149. Tobias Krantz, Makten över regionen. En idékritisk studie av svensk regiondebatt 1963–1996. 278 sid. 2002.

150. Leif Lewin, Svenskt okynne. Att underminera det auktoritära samhället. En konferens anordnad av Humanistisk-samhällsvetenskapliga vetenskapsområdet vid Uppsala univer-sitet. 132 sid. 2002.

151. Karin Borevi, Välfärdsstaten i det mångkulturella samhället. 343 sid. 2002. 152. Per Strömblad, Politik på stadens skuggsida. 226 sid. 2003. 153. Fredrik Bynander, The Rise and Fall of the Submarine Threat: Threat Politics and Sub-

marine Intrusions in Sweden 1980–2002. 270 sid. 2003. 154. Thomas Persson, Normer eller nytta? Om de politiska drivkrafterna bakom Regerings-

kansliets departementsindelning. 293 sid. 2003. 155. Helena Wockelberg, Den svenska förvaltningsmodellen. Parlamentarisk debatt om för-

valtningens roll i styrelseskicket. 379 sid. 2003. 156. Hanna Bäck, Explaining Coalitions. Evidence and Lessons From Studying Coalition

Formation in Swedish Local Government. 203 sid. 2003. 157. Peder Nielsen, Kommunindelning och demokrati. Om sammanläggning och delning av

kommuner i Sverige. 367 sid. 2003. 158. Per Adman, Arbetslöshet, arbetsplatsdemokrati och politiskt deltagande. 171 sid. 2004. 159. Katarina Barrling Hermansson, Partikulturer. Kollektiva självbilder och normer i Sveri-

ges riksdag. 330 sid. 2004. 160. Karolina Hulterström, In Pursuit of Ethnic Politics. Voters, Parties and Policies in Ken-

ya and Zambia. 278 sid. 2004. 161. Magnus Öhman, The Heart and Soul of the Party. Candidate Selection in Ghana and Af-

rica. 313 sid. 2004. 162. Jan Andersson, Yttrandefrihetens dilemman. Debatten om hetspropaganda, mediekon-

centration samt personlig integritet mellan 1940– och 2000-tal. 188 sid. 2004. 163. Leif Lewin, Osvenskt kynne: forskare med utländsk bakgrund om sin syn på upsaliensk

humaniora och samhällsvetenskap; en konferens anordnad av Humanistisk-samhälls-vetenskapliga vetenskapsområdet vid Uppsala universitet den 15 februari 2005. 99 sid. 2005.

164. Ellen Almgren, Att fostra demokrater. Om skolan i demokratin och demokratin i skolan. 223 sid. 2006.

165. Daniel Wohlgemuth, Den responsiva demokratin: Effekter av medborgarnas delaktighet i den lokala demokratin. 238 sid. 2006.

166. Andreaz Strömgren, Samordning, hyfs och reda. Stabilitet och förändring i svensk plan-politik 1945–2005. 275 sid. 2007.

167. Gunnar Myrberg, Medlemmar och medborgare. Föreningsdeltagande och politiskt engage-mang i det etnifierade samhället. 198 sid. 2007

168. Sara Monaco, Neighbourhood Politics in Transition. Residents’ Associations and Local Government in Post-Apartheid Cape Town. 223 sid. 2008.

169. Jörgen Ödalen, Rolling Out the Map of Justice. 154 sid. 2008. 170. Statsvetare ifrågasätter. Uppsalamiljön vid tiden för professorsskiftet den 31 mars 2008.

538 sid. 2008. 171. Björn Lindberg, Fit for European Democracy? Party Discipline in the European Parlia-

ment. 201 sid. 2008. 172. Aaron Maltais, Global Warming and Our Natural Duties of Justice. A cosmopolitan po-

litical conception of justice. 234 sid. 2008.

173. Mose Auyeh, Comparing No-Party Participatory Regimes. Why Uganda Succeeded and Others Failed. 282 sid. 2008.

174. Karl-Göran Algotsson, Grundläggande rättigheter eller handelshinder? Den svenska de-batten om EU, föreningsfriheten och rätten till fackliga stridsåtgärder. 349 sid. 2008.

Skrifter utgivna genom Li Bennich-Björkman och Jörgen Hermansson

175. Stefan Björklund, Reformismens anatomi. Erasmus versus Luther. 119 sid. 2009. 176. Marianne Danielsson. Fängslande idéer. Svensk miljöpolitik och teorier om policypro-

duktion. 385 sid. 2010. 177. Tanja Olsson Blandy. The Europeanisation of gender equality. The unexpected case of

Sweden. 196 sid. 2010. 178. Wendy Maycraft Kall.The Governance Gap. Central–local steering & mental health re-

form in Britain and Sweden. 400 sid. 2011. Skrifter utgivna genom Li Bennich-Björkman, Jörgen Hermansson och Joakim Palme

179. Emma Lundgren Jörum. Beyond the Border. Syrian Policies towards Territories Lost. 254 sid. 2011.

180. Josefina Erikson. Strider om mening. En dynamisk frameanalys av den svenska sexköp-slagen. 189 sid. 2011.

181. Johanna Söderström. Politics of Affection: Ex-Combatants, Political Engagement and Reintegration Programs in Liberia. 402 sid. 2011.

182. Julia Jennstål, Traits & Talks. Lessons about Personality and Deliberation from the Negotiations between Nelson Mandela and FW de Klerk. 276 sid. 2012.

183. Emma Björnehed, Ideas in Conflict. The effect of frames in the Nepal conflict and peace process. 257 sid. 2012.

184. Jenny Jansson, Manufacturing Consensus. The Making of the Swedish Reformist Work-ing Class. 296 sid. 2012.

185. Per Helldahl, The Challenge from Nationalism. Problems of Community in Democracy. 159 sid. 2013.

186. Ludvig Norman, From Friends to Foes. Institutional Conflict and Supranational Influ-ence in the European Union. 240 sid. 2013.

187. Anna Danielson, On the Power of Informal Economies and the Informal Economies of Power. Rethinking Informality, Resilience and Violence in Kosovo. 273 sid. 2014.

188. Sofia Knöchel Ledberg, Governing the Military, Professional Autonomy in the Chinese People's Liberation Army. 260 sid. 2014.

189. Kristin Ljungkvist, The Global City 2.0, An International Political Actor Beyond Eco-nomism? 346 sid. 2014.

190. Michael Jonsson, A Farewell to Arms, Motivational Change and Divergence Inside FARC-EP 2002–2010. 324 s. 2014.

191. Jaakku Turunen, Semiotics of Politics, Dialogicality of Parliamentary Talk. 396 sid. 2015.

192. Niklas Nilsson, Beacon of Liberty, Role Conceptions, Crises and Stability in Georgia’s Foreign Policy, 2004–2012. 280 sid. 2015.

193. Oscar Larsson, The Governmentality of Meta-governance, Identifying Theoretical and Empirical Challenges of Network Governance in the Political Field of Security and Beyond. 204 sid. 2015.

194. Jessica Giandomenico, Transformative Power Challenged, EU Membership Condition-ality in the Western Balkans Revisited. 237 sid. 2015.

195. Hanne Martinek, Mothers’ Social Citizenship. The logics and effects of the German and Swedish welfare states. 202 sid. 2016.

196. Svante Nycander, The History of Western Liberalism. 411 sid. 2016. 197. Jonas Hultin Rosenberg, The All-Affected Principle and its Critics. A Study on

Democratic Inclusion. 142 sid. 2016. 198. Magnus Johnsson, Strategic Colonels. The Discretion of Swedish Force Commanders in

Afghanistan 2006–2013. 282 sid. 2017. 199. Evert Vedung, Gustav Jakob Petersson och Tage Vedung (red.), Secession och diplo-

mati. Unionsupplösningen 1905 speglad i korrespondens mellan UD, beskickningar och konsulat. xii + 876 sid. 2017.

Skrifter utgivna genom Li Bennich-Björkman och Joakim Palme

200. Marcus Wangel, Deep Roots and Tangled Branches. Bureaucracy and Collaboration in Natural Resource Governance in South India. 262 sid. 2018.

201. Karin Leijon, National Courts as Gatekeepers in European Integration. Examining the Choices National Courts Make in the Preliminary Ruling Procedure. 213 sid. 2018.

Skriftserien ingår från och med nr 94 i serien Acta Universitatis Upsaliensis.