¹ Treaty Establishing the European Coal and Steel Community, art. 31, Apr. 18, 1951 (“La Cour assure le respect du droit dans l'interprétation et l'application du présent Traité et des règlements d'exécution.”) (now TEU art. 19).Google Scholar

² See, e.g., Alter, Karen J., Who are the ‘Masters of the Treaty'?: European Governments and the European Court of Justice, 52 Int'l Org. 121 (1998); Weiler, J.H.H., The Transformation of Europe, 100 Yale L.J. 2403 (1991); Craig, Paul & Gráinne de Bürca, EU Law 63–66 (2011).Google Scholar

³ John J. Barceló, Precedent in European Community Law, in Interpreting Precedent 407, 417 (D. Neil McCormick et al. eds., 1997). Barceló notes that the Court never explicitly refers to its previous judgments as “precedents”. Id. However, the Court has acknowledged that the General Court's judgments can “constitute a precedent for future cases.” Case C-197/09 RX-II, M v. EMEA, 2009 E.C.R. 662, para. 62; Case C-334/12 RX-II, Jaramillo et al. v. EIB, 2013 E.C.R. 134, para. 50.Google Scholar

⁴ See, e.g., Case C-409/06, Winner Wetten GmbH v. Bürgermeisterin der Stadt Bergheim, 2010 E.C.R. I-8015, paras. 36, 39, 53, 58. On the importance of settled case law in proceedings before the CJEU, see Lasser, Mitchel, Judicial Deliberations—A Comparative Analysis of Judicial Transparency and Legitimacy 107-12 (2004).Google Scholar

⁵ See, e.g., Case C-176/12, Association de médiation sociale v. Union locale des syndicats CGT et al., para. 50 (“[A] party injured as a result of domestic law not being in conformity with European Union law can nonetheless rely on the judgment in Joined Cases C-6/90 and C-9/90 Francovich and Others [1991] ECR I-5357.”).Google Scholar

⁶ In a forthcoming article, Mattias Derlén & Johan Lindholm, Peek-a-boo, It's a Case Law System (forthcoming 2015), we explore the question of how CJEU case law is an important source of law.Google Scholar

⁷ These claims are presented in the beginning of Sections C-E. See infra Sections C-E. While the study's main contribution is to legal science, we draw knowledge and experience from a wide range of sciences when it comes to methodology. We realize that many legal readers may be more interested in the findings than the method, but we feel that full methodological disclosure is necessary.Google Scholar

⁸ See Barceló, supra note 3, at 415-16 (providing an overview of existing scholarship on this topic); see also Toth, A.G., The Authority of Judgments of the European Court of Justice: Binding Force and Legal Effects, 4 Y.B. Eur. L. 1 (1984).Google Scholar

⁹ See, e.g., Sweet, Alec Stone, The European Court of Justice, in The Evolution of EU Law 121 (Craig, Paul & Gráinne de Bürca eds., 2011); Shapiro, Martin, The European Court of Justice, in The Evolution of EU Law 321 (Craig, Paul & Gráinne de Bürca eds., 1999); Vesterdorf, Bo, A Constitutional Court for the EU?, 4 Int'l J. Constitutional L. 607 (2006).Google Scholar

¹⁰ See also infra Figure 1.Google Scholar

¹¹ Schauer, Frederick, Precedent, 39 Stan. L. Rev. 571, 572-73 (1987).Google Scholar

¹² To further complicate things, new citations change the content and structure of the law. See also infra note 36 and accompanying text.Google Scholar

¹³ Barceló, supra note 3.Google Scholar

¹⁴ Cf. Yonathan Lupu & James H. Fowler, Strategic Citations to Precedent on the U.S. Supreme Court, 42 J. Legal Stud. 151 (2013). Obviously, persuasive power in this sense differs quite distinctly from the persuasive force of judgment in a normative sense more commonly discussed in legal scholarship. The latter often employs the concept of persuasive force in relation to the traditional view of formally binding precedent, see, for example, Bronaugh, Richard, Persuasive Precedent, in Precedent in Law 217 (Laurence Goldstein ed., 1987). As part of this discussion the persuasiveness of a single judgment depends on a number of factors, including the position of the court in the hierarchy and the soundness of the reasoning, see further Aleksander Peczenik, The Binding Force of Precedent, in Interpreting Precedent 461 (D. Neil MacCormick et al. eds., 1997). In contrast, our concept is purely empirical, measuring the extent to which a judgment is embedded in previous case law by way of citations.Google Scholar

¹⁵ See, e.g., Peczenik, supra note 14.Google Scholar

¹⁶ Lupu, Yonatan & Voeten, Erik, Precedent in International Courts: A Network Analysis of Case Citations by the European Court of Human Rights, 42 Brit. J. Pol. Sci. 413, 417 (2011). The extreme counter-position would be that a judgment's authority exclusively follows from the position of the rendering court.Google Scholar

¹⁷ See Tridimas, Takis, The Court of Justice and Judicial Activism, 21 Eur. L. Rev. 199, 203 (1996).Google Scholar

¹⁸ Notwithstanding the fact that the CJEU expects national courts to either follow its findings or request a preliminary reference. See Joined Cases 28/62, 29/62 & 30/62, Da Costa en Schanke v. Nederlandse Belastingadministratie, 1963 E.C.R. 61.Google Scholar

¹⁹ Davies, Gareth, Activism Relocated. The Self-Restraint of the European Court of Justice in its National Context, 19 J. Eur. Pub. Pol'y 76, 76 (2012).Google Scholar

²⁰ The reasoning of the CJEU has been heavily criticized as lacking in transparency. For an overview see Derlén, Mattias, Multilingual Interpretation of CJEU Case Law: Rule and Reality, 39 Eur. L. Rev. 295, 296-98 (2014).Google Scholar

²¹ Cf. Lawrence M. Friedman et al., State Supreme Courts: A Century of Style and Citation, 33 Stan. L. Rev. 773, 794 (1981); Hansford, Thomas G. & James F. Spriggs II, The Politics of Precedent 16–23 (2008); Jacob, Marc, Precedent and Case-Based Reasoning in the European Court of Justice 100 (2014).Google Scholar

²² See, e.g., Fowler, James H. et al, Network Analysis and the Law: Measuring the Legal Importance of Precedents at the U.S. Supreme Court, 15 Pol. Analysis 324 (2007).Google Scholar

²³ Kleinberg, Jon M., Authoritative Sources in a Hyperlinked Environment, 46 J. ACM 604 (1999).Google Scholar

²⁴ In the HITS algorithm, “importance” is measured as Authority score and calculated on the basis of the judgments citing it. A good Authority is a node pointed to by many good Hubs and a good Hub is a node that points to many good Authorities.Google Scholar

²⁵ In a network of case law, authorities and hubs can be translated as influential judgments and judgments that are well founded in law. See Fowler, supra note 22, at 331. For a discussion of important hubs in the case law of the CJEU, see Derlén, Mattias & Lindholm, Johan, Goodbye van Gend en Loos, Hello Bosman? Using Network Analysis to Measure the Importance of Individual CJEU Judgments, 20 Eur. L.J. 667, 685 (2014).Google Scholar

²⁶ Landes, William M. & Posner, Richard A., Legal Precedent: A Theoretical and Empirical Analysis, 19 J. L. & Econ. 249, 250 (1979). It is important to remember that precedential power in this sense, viz. to what extent a judgment has actually affected subsequent judgments, differs quite distinctly from the more common, normative legal concept of precedence that, for example, provides that courts should cite and follow judgments rendered by certain institutions.Google Scholar

²⁷ The chance of a teleport, in the algorithm expressed as the damping factor (d), affects how far back in a chain of citations the Random Walker will travel. Because of the comparatively limited size of the CJEU's case law network, we use a factor of 0.5 (compare with the original where d=0.85). We want to emphasize that while the exploration is random, the outcome is not. The Random Walker follows actual citations between judgments and given enough exploration reveals the structure of the citation network.Google Scholar

²⁸ See Brin, Sergey & Page, Lawrence, The Anatomy of a Large-Scale Hypertextual Web Search Engine, 30 Computer Networks & ISDN Sys. 107 (1998); Lawrence Page et al., The PageRank Citation Ranking: Bringing Order to the Web (1998), http://ilpubs.stanford.edu:8090/422/1/1999-66.pdf.Google Scholar

²⁹ An obvious alternative to PageRank is Authority Score, which is the other value, besides Hub Score, produced by the HITS algorithm, see supra chapter B.II, used for example, in James H. Fowler & Sangick Jeon, The Authority of Supreme Court Precedent, 30 Soc. Networks 16 (2008). The main advantage of PageRank over Authority as a measure of precedential value is that PageRank allows, by use of a Random Walker and modifiable through the damping factor, see supra note 27, precedential weight to be given when appropriate to a judgment that was not cited directly but that served as the basis for a judgment that was cited. See Derlén, Mattias et al., Coherence Out of Chaos: Mapping European Union Law by Running Randomly Through the Maze of CJEU Case Law, 16 Europarättslig Tidskrift 517, 520-24 (2013). Thus, Authority Score will tend to disadvantage foundational cases, for instance, cases on which further case law is built, without necessarily having accumulated many citations. An example of this is the well-known Case 14/83, von Colson, the foundation of the principle of indirect effect. When using PageRank, the case is one of the top ten cases, but the case has a limited Authority Score (.00082, as measured in 2011), placing the case at place 60. See also Derlén & Lindholm, supra note 25, at 678. Unfortunately, there is no comparable, alternative centrality measurement for Hub Score.Google Scholar

³⁰ See, e.g., Foster, Nigel, EU Law 64 (2014).Google Scholar

³¹ Derlén & Lindholm, supra note 25.Google Scholar

³² Derlén & Lindholm, supra note 6.Google Scholar

³³ But see Jacob, supra note 21 (conducting a limited empirical study).Google Scholar

³⁴ When we calculate Hub Score and PageRank for individual judgments, we take the entire dataset into account. Thus, the examination is limited to judgments rendered 1992–2011, but when we measure the PageRank and Hub Score of those judgments, we take the entire network into account. To do otherwise would produce strange results. Say, for example, that we are interested in the Hub Score of a group of recent judgments, and that they cite the seminal cases 8/74 Dassonville and 120/76 Cassis de Dijon, two of the most significant authorities in the network. See Derlén & Lindholm, supra note 25, at 685–86. If judgments before 1992 were not considered when calculating Hub Score, we would erroneously conclude that the judgments discussed have very low Hub Score, despite being firmly embedded in previous case law.Google Scholar

³⁵ See Derlén & Lindholm, supra note 25, at 675; Derlén & Lindholm, supra note 6 (providing greater detail on the matter). By 1989, average out-degree per case had reached 2.25, exceeding 2 for the first time and only ever increasing thereafter. Even more importantly, the yearly percentage of cases citing at least one previous case reached eighty percent by 1989, never to dip below that threshold again, instead approaching a hundred percent in recent years.Google Scholar

³⁶ Naturally, this is a snapshot of CJEU case law as it stood in 2011. One challenge when exploring case law using network analysis is that the network constantly changes. This is particularly true when using Hub Score, PageRank, and other feedback centrality measures as every new link added affects the centrality of all nodes in the network. This is not merely a methodological problem. From a legal perspective, a court citing, interpreting, and developing a previous decision represents a legal shift; the law is not the same as it was just moments before. We seek to explore this issue and how to overcome it in future studies.Google Scholar

³⁷ For example, that the case was brought as a preliminary reference (actions), by the Commission (actors), or concerned the free movement of goods (area).Google Scholar

³⁸ See, e.g., Albert-László Barabási & Albert, Réka, Emergence of Scaling in Random Networks, 286 Sci. 509 (1999). Consequently, the traditional measurement of standard deviation becomes unhelpful, and is not employed in this study. The standard deviation for Hub Score as well as PageRank is higher than the mean (.000172 as compared to .000131 and .000167 as compared to .000103, respectively) due to the Power-Law Distribution. For more on the distribution of links in the network, see Derlén & Lindholm, supra note 6. That citations follow a Power-Law Distribution is not as such problematic or grounds for criticism; they naturally arise in, for example, business networks, social networks, transportation networks, academic citation, and the World Wide Web. See, e.g., Barabási & Albert, supra note 38.Google Scholar

³⁹ Aaron Clauset et al., Power-Law Distribution in Empirical Data, 51 SIAM Rev. 661, 662 (2009).Google Scholar

⁴⁰ To use the same example, it can be fruitful to compare average city population in Sweden and in the United States.Google Scholar

⁴¹ TFEU art. 267.Google Scholar

⁴² See, e.g., Alter, supra note 2, at 126–29, 122Google Scholar

See also Sweet, Alec Stone & Brunell, Thomas L., Constructing a Supranational Constitution: Dispute Resolution and Governance in the European Community, 92 Am. Pol. Sci. Rev. 63, 65–66 (1998).Google Scholar

⁴³ Craig, Paul, The Jurisdiction of the Community Courts Reconsidered, in The European Court of Justice 177, 182-83 (Gráinne de Bürca & J.H.H. Weiler eds., 2001).Google Scholar

⁴⁴ Jacob, supra note 21, at 19. Regarding what constitutes an important decision, see Derlén & Lindholm, supra note 25.Google Scholar

⁴⁵ Schepel, Harm & Blankenburg, Erhard, Mobilizing the European Court of Justice, in The European Court of Justice supra note 43, at 9, 41–42.Google Scholar

⁴⁶ See infra Figures 2 & 3.Google Scholar

⁴⁷ Of course, this does not mean that no such cases have high precedential or persuasive power. See, e.g., Case C-365/97, Comm'n v. Italy, 1999 E.C.R. I-7773; Case C-103/00, Comm'n v. Greece, 2002 E.C.R. I-1147; Case C-168/03, Comm'n v. Spain, 2004 E.C.R. I-8227. However, most such cases are primarily cited in other successful infringement proceedings of limited precedential importance (most of which are not published in full text) and on points of procedure, for example, what constitutes the relevant time for assessing a failure to fulfill obligations and the placement of the burden of proof in an infringement proceeding.Google Scholar

⁴⁸ Measured as the top ten percent by PageRank.Google Scholar

⁴⁹ Case C-415/93, URBSFA v. Bosman, 1995 E.C.R. I-4921; Case C-379/98, PreussenElektra AG v. Schhleswag AG, 2001 E.C.R. I-2099; Case C-279/93, Finanzamt Köln-Altstadt v. Schumacker, 1995 E.C.R. I-225; Case C-55/94, Gebhard v. Consiglio dell'Ordine degli Avvocati e Procuratori di Milano, 1995 E.C.R. I-4165.Google Scholar

⁵⁰ See, e.g., the Court's often-cited judgments in Greece, Case C-103/00; Spain, Case C-168/03; Case C-23/05, Comm'n v. Luxembourg, 2009 E.C.R. I-124, which the Court routinely cites in infringement proceedings (many of which are summary publications) on the point that it will not consider any changes made subsequent to the period laid down in the Commission's reasoned opinion.Google Scholar

⁵¹ Individual parties’ success rate in direct actions is part of the dataset, but not presented in full in this article.Google Scholar

⁵² Another possible and combinable explanation for this finding is that the Court is more concerned with its internal standing vis-à-vis the other EU institutions and thus more careful to legitimize its findings when it goes against the Commission than against a Member State. However, our findings regarding party constellations in direct actions suggest that this is not the case. See infra Chapter E.Google Scholar

⁵³ Barceló, supra note 3, at 419.Google Scholar

⁵⁴ Foster, supra note 30, at 76–77.Google Scholar

⁵⁵ Davies, supra note 19, at 79–80.Google Scholar

⁵⁶ Cf. Jacob, supra note 21, at 16–18 (arguing that in the absence of legislation and codification, “[t]he task of putting flesh on the bones” that is the Treaties fall to the CJEU).Google Scholar

⁵⁷ Lupu & Voeten, supra note 16.Google Scholar

⁵⁸ Landes & Posner, supra note 26, at 249.Google Scholar

⁵⁹ The average number of subject matters has remained more or less constant since 1971.Google Scholar

⁶⁰ However, as explained further below, many of those cases fall under the subject matter Principles, objectives and tasks of the Treaties. Google Scholar

⁶¹ For example, thirty-four subject matters always occur together with the subject matter Agriculture, including Beef & Veal, Cereals, Dry Fodder, Flax & Hemp, Milk Product, Sugar, and Wine. We find no plausible reason why there would be a legal difference between these categories of agricultural products.Google Scholar

⁶² The lowest being Safety at Work and Elsewhere which is only a subject matter in two judgments. We have excluded subject matters raised in less than fifty cases (n<50), as they constitute too small a sample.Google Scholar

⁶³ See supra Section C.Google Scholar

⁶⁴ See infra Table 1.Google Scholar

⁶⁵ See supra Section C.Google Scholar

⁶⁶ Flynn, Leo, Coming of Age: The Free Movement of Capital Case Law 1993-2002, 39 Common Mkt. L. Rev. 773, 804 (2002).Google Scholar

⁶⁷ Case C-319/02, Manninen, 2004 E.C.R. I-7477.Google Scholar

⁶⁸ Case C-35/98, Staatssecretaris van Financiën v. B.G.M. Verkooijen, 2000 E.C.R. I-4071.Google Scholar

⁶⁹ Joined Cases C-46/93 & 48/93, Brasserie du Pecheur SA v. Germany, The Queen v. Sec'y of State for Transport, 1996 E.C.R. I-1029.Google Scholar

⁷⁰ Case C-129/96, Inter-Environnement Wallonie ASBL v. Région Wallonne, 1997 E.C.R. I-7411.Google Scholar

⁷¹ Case C-85/96, Martínez Sala v. Freistaat Bayern, 1998 E.C.R. I-2691.Google Scholar

⁷² Case C-144/04, Mangold v. Helm, 2005 E.C.R. I-9981.Google Scholar

⁷³ Case C-112/00, Schmidberger, Internationale Transporte und Planzüge v. Austria, 2003 E.C.R. I-5659.Google Scholar

⁷⁴ The second to highest average is Customs Union with 2.25 references.Google Scholar

⁷⁵ Cf. Lupu & Voeten, supra note 16, at 421.Google Scholar

⁷⁶ Jacob, supra note 21, at 18–19.Google Scholar

⁷⁷ Until the European Union's enlargement in 2004, the Full Court would decide certain—presumably more important—cases collectively, whereas other—presumably less important—cases would be heard by smaller chambers. During the time period examined in this study, the Full Court consisted of twelve to fifteen members. Since the enlargement, the Court rarely sits as a full court, instead relinquishing its role to the Grand Chamber, a composition comprising of a minimum of thirteen judges.Google Scholar

⁷⁸ Ranked by PageRank, making them overrepresented by 2.5 times.Google Scholar

⁷⁹ They have a mean PageRank of .000195.Google Scholar

⁸⁰ They have a mean Hub Score of .000226. Judgments made by the Grand Chamber (i.e. after 2004) cite on average 10.59 previous judgments. That can be compared to an average of 6.10 citations for all judgments between 1992 and 2011.Google Scholar

⁸¹ See supra Section C.Google Scholar

⁸² See infra Table 2. We have only actions involving the larger institutions in the table, excluding, for example, the Community Plant Variety Office and European Maritime Safety Agency.Google Scholar

⁸³ See supra Section B.IV.Google Scholar

⁸⁴ Successful infringement proceedings are excluded from consideration in this part.Google Scholar

⁸⁵ Case C-352/98 P, Bergaderm SA & Goupil v. Comm'n, 2000 E.R.C. I-5291.Google Scholar

⁸⁶ Case C-185/95 P, Baustahlgewebe GmbH v. Comm'n, 1998 E.R.C. I-8417.Google Scholar

⁸⁷ Rules of Procedure of the Court of Justice, EUT (L 265) 29.09.2012, art. 37.3., at 1–42.Google Scholar

⁸⁸ Legal language is dependent on national legal traditions and culture, which is why legal concepts can vary in meaning between countries with the same national language and even within a country. See Pozzo, Barbara, Multilingualism, Legal Terminology and the Problems of Harmonising European Private Law, in Multilingualism and the Harmonisation of European Law 3, 6-8 (Pozzo, Barbara & Valentina Jacometti eds., 2006) (providing an introduction to this subject).Google Scholar

⁸⁹ See infra Table 3.Google Scholar

⁹⁰ Italian, Portuguese, and Spanish.Google Scholar

⁹¹ Danish, Finnish, and Swedish.Google Scholar

⁹² Protocol (No. 3) on the Statute of the Court of Justice of the European Union, art. 23, para. 2.Google Scholar

⁹³ See infra Figure 4.Google Scholar

⁹⁴ As a result, data for Commission observations and any observation mirror each other.Google Scholar

⁹⁵ See infra Figure 4. Obviously, Hub Score is still only calculated on the basis of cases cited by the CJEU in the judgment. In another article, the authors of this article explore the relationship between the opinions expressed by the Member States in their observations and the embeddedness of the final judgment in previous case law. See Larsson, Olof, Daniel Naurin, Mattias Derlén & Johan Lindholm, Speaking Law to Power. The Strategic Use of Precedent of the Court of Justice of the European Union (forthcoming).Google Scholar

⁹⁶ See Broberg, Morten & Fenger, Niels, Preliminary References to the European Court of Justice 357-58 (2014).Google Scholar

⁹⁷ See Derlén & Lindholm, supra note 6 (discussing the average life time of a CJEU judgment).Google Scholar

⁹⁸ Case C-457/12, S & G v. Minister voor Immigratie, Integratie en Asiel, EU:C:2014:136. At the time of writing, the judgment has never been cited.Google Scholar

⁹⁹ See supra Section B.I.Google Scholar