Introduction
A steadily growing literature has sought to explain how high courts and international courts reason and thereby make law (see Tiller and Cross Reference Tiller and Cross2006; Lax Reference Lax2011; Lupu and Voeten Reference Lupu and Voeten2012; Fox and Vanberg Reference Fox and Vanberg2014; Beim Reference Beim2017; Callander and Clark Reference Callander and Clark2017; Larsson et al. Reference Larsson, Naurin, Derlén and Lindholm2017; Ainsley, Carrubba, and Vanberg Reference Ainsley, Carrubba and Vanberg2021). This article contributes to the research on judicial reasoning by focusing on a tool that is available to many courts, but nevertheless has received little attention in the literature: explicit engagement with amicus curiae (“friends of the court”) briefs.
Such briefs by third-parties contain proposals as to how relevant legal issues should be solved and disputes should be adjudicated, along with supporting legal reasoning. Scholars have studied extensively the willingness of third parties to file briefs and the influence of briefs on judicial decisions (Epstein Reference Epstein1992; Kearney and Merrill Reference Kearney and Merrill2000; Collins, Corley, and Hamner Reference Collins, Corley and Hamner2015; Collins and McCarthy Reference Collins and McCarthy2017; Collins Reference Collins2018; Dederke and Naurin Reference Dederke and Naurin2018; Hazelton and Hinkle Reference Hazelton and Hinkle2022). There is also a substantial literature on how courts use references to legal sources as a tool for judicial reasoning (Hume Reference Hume2006; Lupu and Voeten Reference Lupu and Voeten2012; Larsson et al. Reference Larsson, Naurin, Derlén and Lindholm2017). However, few studies have theorized and empirically studied the function that references to briefs may have for a court (two exceptions are Canelo Reference Canelo2022 and Bagashka, Chapa, and Tiede Reference Bagashka, Chapa and Tiede2024). Citing briefs is different from citing legal sources: while briefs may contain arguments that persuade judges to adopt and repeat them in support of their decisions, they have no legal authority and courts generally have no obligation to refer to them.
The relative dearth of research on the function of courts’ references to amicus briefs is surprising given how common such references are. Kearney and Merrill (Reference Kearney and Merrill2000, 757) documented a steady rise in the citation of amicus briefs by the US Supreme Court (SCOTUS) in the period 1946–1995, with about one third of cases citing briefs in the latter part of the period. More recently, Franze and Anderson (Reference Franze and Anderson2020) report that the SCOTUS cites amicus briefs in more than half of its rulings. In one of the few studies of the impact of amicus briefs outside the US, Bagashka, Chapa, and Tiede (Reference Bagashka, Chapa and Tiede2024, 12) found that between 2000 and 2012, the Bulgarian Constitutional Court cited more than half of all amicus briefs. As we will show, the Court of Justice of the European Union (CJEU) also regularly cites briefs submitted by the European Commission and Member States, including those not directly involved in the dispute. In the period 1995–2011, the CJEU explicitly referenced 33% of the Member States’ briefs and 40% of the Commission’s briefs.
Especially intriguing are negative references to amicus briefs – that is, instances where a court explicitly signals that it disagrees with the brief. Positive references to briefs that support the court’s decision are likely to fill a persuasive function, demonstrating to potentially skeptical observers that the court was not alone in its assessment of the legal issues at stake. But why do courts regularly draw attention to alternative legal assessments that contrast with their own?
We develop and test several theoretical propositions of how negative references to briefs may be useful to courts: As a tool for law making, such references may increase the precision of the case law as they provide courts with an opportunity to clarify their reasoning and expand the scope of its precedent. As a tool for fostering legitimacy, negative references may be used to demonstrate responsiveness and signal to concerned third parties that their voices have been heard. As a tool for judicial impact, negative references may function as a fire alarm to raise attention to possible problems with the implementation of the court’s decision and to provide compliance constituents with legal ammunition to put pressure on implementing authorities.
Our theoretical account of how courts engage with briefs in their reasoning integrates legal motivations with concerns for legitimacy and policy implications. It thereby resonates with the research agenda in the political science literature on judicial politics that strives to take law seriously (see Friedman Reference Friedman2006; Lax Reference Lax2011; Clark Reference Clark2016). Our theory is rooted in the assumption that judges strive to craft “good law” – interpretations of the law that are supported by sound legal arguments – while at the same time producing outcomes that are complied with and generally perceived as fair and legitimate.
While the literature on amicus briefs focuses overwhelmingly on American courts (see Collins Reference Collins2018, 223), our focus is on the European Union. In our empirical application, we rely on original data covering all preliminary references lodged with the CJEU between 1995 and 2011. Our data offers detailed insights into how the Court engages with the arguments submitted in briefs by third-parties (in EU jargon known as “observations”), notably the European Commission and the Member States. For each legal question the CJEU considered in more than 4,200 preliminary references, we know which Member States submitted observations and the sentiment with which the Court engaged with each observation.
Our empirical analysis lends support to some, but not all, of our theoretical expectations. We find at least some support for the proposition that the Court uses negative references to expand its precedent by increasing the precision of the law. We find strong support for the proposition that negative references to briefs are used when the Court perceives a need to strengthen the legitimacy of its decision in the eyes of the most concerned Member States, but not that they are used as a fire alarm to boost compliance.
To the best of our knowledge, we offer the first analysis of the CJEU’s references to Member State observations, and thereby contribute to a well-established and ever-expanding strand of literature, exploring how the CJEU’s jurisprudence shapes the course of European integration (see for example Alter Reference Alter1996, Reference Alter1998, Reference Alter2014; Garrett, Kelemen, and Schulz Reference Garrett, Kelemen and Schulz1998; Carrubba, Gabel, and Hankla Reference Carrubba, Gabel and Hankla2008; Davies Reference Davies2012; Stone Sweet and Brunell Reference Stone Sweet and Brunell2012; Martinsen Reference Martinsen2015; Larsson and Naurin Reference Larsson and Naurin2016; Larsson et al. Reference Larsson, Naurin, Derlén and Lindholm2017; Blauberger and Martinsen Reference Blauberger and Martinsen2020; Krehbiel and Cheruvu Reference Krehbiel and Cheruvu2022). Further, by incorporating both legal and political constraints on judicial behavior into our theoretical expectations, we contribute to bridging the divide between political scientists and the legal academy, two strands of scholarship that study the same subject but who have often found it difficult to speak the same language.
We proceed as follows: The next section outlines the primitives of our theory of the functions of negative references to amicus briefs in judicial reasoning. We then translate these into concrete expectations for the CJEU’s engagement with briefs submitted in the course of preliminary reference proceedings. Our empirical application follows, employing a statistical analysis of the CJEU’s negative references to briefs. We conclude by discussing key findings from our analysis and their implications for our understanding of judicial reasoning and precedent-setting.
Theorizing negative references to amicus briefs
Many high courts and international courts allow third parties to submit briefs. On many courts, briefs are permitted both by government actors and interest groups, while in the EU preliminary reference procedure, written observations are only allowed by Member States and EU institutions. Previous research on both national (Collins, Corley, and Hamner Reference Collins, Corley and Hamner2015; Collins Reference Collins2018; Hazelton and Hinkle Reference Hazelton and Hinkle2022; Bagashka, Chapa, and Tiede Reference Bagashka, Chapa and Tiede2024) and international (Carrubba, Gabel, and Hankla Reference Carrubba, Gabel and Hankla2008; Cichowski Reference Cichowski2016; Larsson and Naurin Reference Larsson and Naurin2016) courts have generally found that briefs influence courts’ decisions.
What is less well understood is why judges sometimes choose to explicitly refer to amicus briefs in their judgments. Sometimes a positive reference may simply be an acknowledgement that the brief provided the opinion writer with useful “fodder” for their argument (Epstein Reference Epstein1992, 650). Kearney and Merrill (Reference Kearney and Merrill2000, 811) propose that cited briefs contain more quality information than non-cited briefs, but do not test this proposition empirically. Another intuitive explanation for positive references to briefs is that these increase the persuasiveness of the judgment. Scholars have found that when international courts face a higher risk of negative political reactions to a judgment, they are more likely to embed their reasoning in more (and more authoritative) case law (Lupu and Voeten Reference Lupu and Voeten2012; Larsson et al. Reference Larsson, Naurin, Derlén and Lindholm2017). A likely reason for that is to increase the legal authority of a contested judgment by pointing to a wealth of previous legal reasoning that fits with the present reasoning. Similarly, showing that third parties have put forward legal arguments in their briefs that are in line with the court’s judgment may be a way for a court to demonstrate that it is making a reasonable decision accepted by others.
While intuitive, the few studies that exist on how courts refer to amicus briefs have largely failed to find any clear patterns. In her study of SCOTUS’s citation practices, Canelo (Reference Canelo2019) found that references to briefs were less common in salient cases, which seems counterintuitive to the idea that references play a positive legitimizing role. In a survey experiment, Canelo also failed to find any positive legitimacy effect in public opinion of a court citing amicus briefs. Bagashka, Chapa, and Tiede (Reference Bagashka, Chapa and Tiede2024) found that the Bulgarian Constitutional Court used language from briefs submitted by powerful political actors more often than less powerful actors, which resonates with other research on the varied influence of third parties (see Box-Steffensmeier, Christenson, and Hitt Reference Box-Steffensmeier, Christenson and Hitt2013; Collins, Corley, and Hamner Reference Collins, Corley and Hamner2015). However, they were not able to find any systematic explanations for the variation in the Court’s willingness to explicitly reference briefs, compared to borrowing language without citing the brief.
Furthermore, as far as we know, there is no existing study that attempts to explain the use of negative citations to briefs. As we will show, for the period that we studied, the CJEU referenced at least one brief negatively in 14% of the legal issues it addressed. We lack information about how common such references are in other courts.Footnote 1 At first sight, negative references to amicus briefs are even more mysterious than positive references: why would a court want to raise attention to legal assessments that contrast with its own?
To understand the function of negative references to briefs, we believe it is useful to draw insights from the literature on the role of dissenting judicial opinions. Much of this literature focuses on the possible gains and harms that dissents may bring to the legitimacy of a court and its judgments (Zink, Spriggs, and Scott Reference Zink, Spriggs and Scott2009; Salamone Reference Salamone2014; Vitale Reference Vitale2014; Naurin and Stiansen Reference Naurin and Stiansen2020; Dunoff and Pollack Reference Dunoff and Pollack2022). Some argue that dissents may damage a court’s legitimacy by weakening the authority of its judgments (Naurin and Stiansen Reference Naurin and Stiansen2020) or undermine the clarity of a court’s case law by creating legal uncertainty (Vitale Reference Vitale2014). A common view is that dissents have positive implications for the quality of a court’s case law. In particular, it is argued that the anticipation of a dissenting opinion forces the majority to sharpen its arguments, clarify, refine, and possibly modify its opinion, which leads to more balanced and precise legal doctrine (Brennan Reference Brennan1985; Ginsburg Reference Ginsburg1990; Scalia Reference Scalia1994).
Moreover, a dissenting opinion can clarify the majority’s views “by throwing them into sharper relief” and “making it clear what the majority does not stand for by providing useful contrast” (Lynch Reference Lynch2003, 726, 743; see also Bergman Reference Bergman1991; Alder Reference Alder2000; McIntyre Reference McIntyre2016). Dissenting opinions “can send signals to appellate bodies, future litigants, and scholars that invite the development of alternative lines of argument” (Dunoff and Pollack Reference Dunoff and Pollack2022, 343).
This is where we find the analogy to negative references to briefs particularly useful: both negative references to briefs and dissenting opinions engage with an alternative holding and reasoning compared to the majority opinion. Just like a dissenting opinion, a negative reference to a brief represents “a roadmap of a juridical path not taken” (Dunoff and Pollack Reference Dunoff and Pollack2022, 346). Thus, dissents and negative references to briefs both deliberately raise attention to contrasting legal avenues and disagreement which may serve to clarify the majority opinion.
There are also important differences between dissenting opinions and negative references to briefs that will affect their impact. Whereas the majority has limited control over their colleagues writing dissenting opinions, negatively referencing amicus briefs is a deliberate act on behalf of the majority. Furthermore, unlike dissenting judges, amici hold no authority on the court, and negative references to their briefs are therefore less likely to harm legitimacy or create more legal uncertainty. To the contrary, as we will argue below, negative references to briefs are more likely to improve legal certainty by increasing the precision of the case law and discouraging outside actors from pursuing that line of reasoning. Thus, while dissenting opinions may open doors to alternative legal avenues, negative references to briefs close them.
Against this background, we propose that negatively referencing briefs serves three potential positive functions for precedent-setting high courts and international courts: They may be used as a tool for judicial law making, as a tool for fostering legitimacy, and as a tool for judicial impact.
Negative references as a tool for judicial law making
First, negative references may function as a tool for judicial law making by giving the court the chance to calibrate the precision and scope of the legal precedent. In any legal order, it is the high courts’ job to reduce uncertainty of what the law is. High courts and international courts should provide authoritative interpretations that are not only applicable to the facts of an individual dispute before them, but generally guide societal behavior and the application of law to future situations.
A challenge for these courts is that case law is often made under a shadow of uncertainty. The court is required to commit to a path that will affect situations that have not yet occurred, which means that it cannot fully know what the societal impact of its decision will be (see Sunstein Reference Sunstein1996, Reference Sunstein1999; Fox and Vanberg Reference Fox and Vanberg2014). If a court delivers a more precise and stringent answer, its constraining effect on future cases increases, whereas vague answers leave more room for interpretation by future courts – who will have more information, and therefore ability, to make better assessments (for a similar argument see Staton and Vanberg Reference Staton and Vanberg2008; Schauer Reference Schauer2006). As a response to this challenge, a court may calibrate the scope of its answer. It can choose everything from a “minimalist approach,” producing the narrowest rule necessary to decide the specific dispute before it and protecting it against setting “out a rule that is wrong as applied to other cases not before it” (Sunstein Reference Sunstein1996, 14–15; see also Schauer Reference Schauer2006; Clark Reference Clark2016), to a “maximalist approach” that involves declaring a broad rule governing a wider set of future scenarios (see Fox and Vanberg Reference Fox and Vanberg2014).
Engaging with amicus briefs that are contrary to the court’s reasoning is a relevant tool in this regard because it provides the court with an occasion to clarify what the law is not, in addition to what it is. In other words, the court can use negative references to briefs to expand the scope of its interpretation and increase the precision of the precedent it sets. Thus, while a dissenting opinion forces the majority to clarify its reasoning, a conflicting brief provides an opportunity for the court to do the same.
An illustrative example from the CJEU of how negative references may have a clarifying effect can be found in Hendrikman and Feyen (Case C-78/95). The case concerned the recognition and enforcement in the Netherlands, under the Brussels Convention, of a German court’s judgment against two Dutch residents who claimed that they had not been properly served, and therefore were not able to defend themselves. The German government filed an observation as a third party in the proceeding. In its judgment, the CJEU explicitly rebukes the German government’s observation, defending its national court’s decision: “The German Government submits that the rights of the defence are observed even if a lawyer who is not authorized to act appears for the defendants because the court must rely on what is stated by that lawyer until such time as he is shown to have no authority.” (para. 16) The CJEU rejected this interpretation of the Convention (para. 17) and held that: “Where proceedings are initiated against a person without his knowledge and a lawyer appears before the court on his behalf but without his authority, such a person is quite powerless to defend himself [and] must therefore be regarded as a defendant in default of appearance…” (para. 18). In practice, the dismissal of the German government’s argument amounts to closing a door toward a less stringent precedent with regards to the right of representation. The CJEU has subsequently relied on that holding when deciding other cases (see e.g., Case C-112/13, A v. B and Others, para. 56).
Another example can be found in Case C-57/96, Meints, which concerned a German agricultural worker’s right to social security benefits in the Netherlands after he had become redundant. After answering the questions posed by the referring Dutch court, the CJEU chose to address an additional issue raised by the governments of France and the Netherlands in their observations. Disagreeing with these Member States’ claim that frontier workers could not rely on the regulation in question to “export” social advantages (para. 49), the CJEU held that the regulation provided such rights to all workers who are nationals of a Member State, including frontier workers (para. 50). Since the referring Dutch court had not asked about this issue, and the CJEU is under no obligation to address additional issues raised in the submitted third-party observations, the Court could have left the proper interpretation of the scope of the regulation open. Instead, it decided to use a negative reference to the French and Dutch governments’ observations to preclude alternative interpretations of the scope of these rights. The CJEU has invoked this holding in several subsequent decisions (see e.g., C-212/05, Hartmann v. Freistaat Beyern, para. 24 and Case C–238/15, Verruga and Others v. Ministre de l’Enseignement supérieur et de la Recherche, para. 39). These cases illustrate how negative references to briefs, essentially obiter dicta, can generate authoritative interpretations of the law.
Negative references as a tool for fostering legitimacy
Second, negative references to amicus briefs may be used as a legitimation tool when a court takes a controversial decision. As argued by Shapiro, because courts resolve conflicts, “judges must have something to tell the loser” (Reference Shapiro1994, 155). A basic function of judicial reasoning is to make it easier for disappointed parties to accept adverse judgments (see e.g., Dyzenhaus and Taggert Reference Dyzenhaus, Taggert and Edlin2007, 148). Again, it is sometimes argued that dissenting opinions may also be helpful in this regard. SCOTUS Justice Sandra Day O’Connor has suggested that the most important role of a dissenting opinion is to show “those who disagree with the Court’s disposition of a case that their views, while they did not prevail, were at least understood and taken seriously” (O’Connor Reference O’Connor2004, 121).
In a similar way, a negative reference shows that the court has taken the submission of a concerned third party seriously. Explaining why the argument in the brief was not persuasive to the court may have a legitimizing effect by signaling that – although the court was not convinced – it has at least engaged with the content of the brief. On the one hand, this indicates that the court is attentive to detail and that it evaluates all the arguments put forward. On the other hand, it also potentially triggers the mechanism found in the research on procedural fairness, which has shown that decision-makers who demonstrate responsiveness to critical voices during a policy process may increase acceptance among those dissatisfied with the outcome (Tyler Reference Tyler1994; Rohrschneider Reference Rohrschneider2005, but see Esaiasson et al. Reference Esaiasson, Persson, Gilljam and Lindholm2019).Footnote 2
The Zambrano case of the CJEU is instructive in this regard (C-34/09). In this case, the Court found that non-EU parents of children with EU citizenship must have a right to reside in the EU for their children to be able to fully enjoy their citizenship rights. The judgment expanded the scope of EU citizenship rights beyond situations involving cross-border movement. The reception of the judgment shows how being caught not fully engaging with briefs may generate legitimacy costs. The CJEU was strongly criticized in the legal scholarly community after the decision, where the Court first acknowledged that “[a]ll governments which submitted observations to the Court and the European Commission argue that … the provisions of European Union law … are not applicable to the dispute in the main proceedings” (Zambrano, para. 37), only to go on to explain its own legal reasoning, leading to the opposite position that EU law indeed is applicable in the case “without discussing the arguments of the states and the Commission” (Dawson, de Witte, and Muir Reference Dawson, de Witte and Muir2013, 3).
Negative references as a tool for judicial impact
Third, besides providing an opportunity for the court to clarify its case law and demonstrate responsiveness to concerned parties, negative references to briefs may also contain an element of coercion. Courts famously lack the power of the purse and the sword, making them dependent on other actors for effective societal impact. Just like dissenting opinions, negative references may signal to actors with powers over the implementation of the case law that there is a potential sensitivity to this case that is worth paying attention to. However, while dissents are often held to weaken the authority of a judgment, thereby providing an excuse to contain compliance, negative references are more likely to have the opposite effect.
We envision two ways by which negative references may increase compliance pressure on recalcitrant public or private actors. Research on judicial impact has emphasized the importance of transparency of judicial decisions in order to activate support for a court in civil society and public opinion (Vanberg Reference Vanberg2005; Staton Reference Staton2010; Krehbiel Reference Krehbiel2016). One effect of highlighting and denouncing the arguments of a third party with implementation powers is to raise awareness of potential compliance problems down the road. In this way, negative references may function as a fire alarm. Potential compliance constituencies – actors with an interest in making sure that the law is effective on the ground (Alter Reference Alter2014) – are made aware that they have reason to increase their monitoring of the implementation of the judgment. Furthermore, a negative reference provides such compliance constituencies with legal arguments against the third party’s alternative legal position, which may be useful in the advocacy process and may thereby function as a source of additional legal ammunition for compliance constituencies (Müller and Slominski Reference Müller and Slominski2019).
The role of compliance constituencies as enforcers of international law is often emphasized in the international relations literature (Simmons Reference Simmons2009; Alter Reference Alter2014). There is also empirical evidence from the European Union that the Commission takes cues from CJEU cases to gain information about possible risks of non-compliance. Hilpert (Reference Hilpert2022) demonstrates that the Commission is more likely to start infringement cases against member states who initiate or support annulment cases (Article 263 of the Treaty on the Functioning of the European Union, in the following TFEU) in order to challenge the validity of an EU legal act. Thus, in the EU context, submitting an observation that signals opposition implies a risk of “awakening a sleeping bear.” When the CJEU explicitly dismisses an EU Member State’s argument, compliance constituencies are provided with a legal argument vetted by the Court itself that specifically addresses and invalidates the contrary argument made by the government. Such a clear legal focal point may be useful in the compliance bargaining that precedes the judgment (see Blauberger Reference Blauberger2012).
In sum, negative references to amicus briefs is a potential tool that courts of precedent may employ to increase the precision of its case law and address concerns about legitimacy and compliance. The implications of this tool may be compared to the practice of dissent. On the one hand, dissenting opinions and negative references of briefs have in common that they raise attention to legal alternatives to the precedent set by the majority, and therefore are likely to incentivize the majority to increase the quality and precision of its legal reasoning. On the other hand, negative references lack the potential negative impact on the legal certainty and authority of the precedent set by the court, and are more likely to contribute positively to the legitimacy of the decision. In the next section, we contextualize our argument in the case of the CJEU and specify testable empirical implications.
Negative references to briefs at the CJEU
What applies to precedent-setting courts generally also applies to the CJEU. The CJEU is entrusted with providing the correct application and interpretation of EU law (see Article 19 (1) TEU), and one of its key means for fulfilling this task is by answering questions posed by national courts through preliminary references (see Article 267 TFEU). A preliminary reference presupposes legal uncertainty since national courts do not need to (and are encouraged not to) refer a question if the CJEU has previously answered a question that is “materially identical” or “where the correct interpretation of EU law is so obvious as to leave no scope for any reasonable doubt” (see Case C-561/19, Consorzio Italian Management and Catania Multiservizi SpA v. Rete Ferroviaria Italiana SpA, paras. 36 and 39).
Despite this “uncertainty threshold,” the legal questions presented to the CJEU vary significantly in terms of legal uncertainty, complexity, and the scope of the precedent that the Court is asked to provide. The CJEU lacks docket control and must address all questions that are submitted to it. Sometimes it may be asked to make a limited interpretation of a single word in a specific piece of EU legislation, or a request for detail to a long and rich line of preexisting case law. Other times, the questions referred may concern broad and vague concepts, the application of general principles, or matters of “first impression.” The opportunity for reducing legal uncertainty through judicial precedent is greater in the latter cases where the blank spot on the map covered by the matter is larger.
It is the national court that formulates the question referred and thus determines the initial scope of the Court’s potential answer and its room for precedent setting. However, there are ways by which the Court can expand the matter that it addresses, and consequently, the scope of the case law as well. The tool that we focus on is engaging with the legal arguments in “observations,” which is the official term for written submissions to the CJEU. Observations may be submitted by the parties, the twenty-seven EU Member States, the European Commission, and (”where appropriate”) other government institutions and agencies within two months after the case is registered at the Court (see Article 23, Statute of the CJEU).
Before we specify empirically testable hypotheses about how the CJEU engages with observations, we should make clear our first priors with regards to the preferences and strategic behavior of the Court. We assume that the main motivation of judges at the CJEU is to fulfill the function of their institution in a professional way.Footnote 3 Judges at the CJEU are likely to strive for developing a coherent and clear legal system that works for its purpose of creating a common market and adjacent policies by means of sound legal reasoning. A core task of the CJEU is to fill the gaps in the EU Treaties. CJEU judges therefore seek to reduce legal uncertainty in EU law by providing clear guidance to national courts and other actors on what the law is – and what it is not – and what they have to do not to violate the law.
Besides aiming to develop a coherent and clear legal system based on sound reasoning, the Court has also been found to be aware of and attentive to the fact that it is situated in a social-political context, where its decisions may have very significant impacts on society (see Blauberger Reference Blauberger2012; Blauberger and Schmidt Reference Blauberger and Schmidt2017; Schmidt Reference Schmidt2018). Along with much of the previous literature, we assume that the Court wants to avoid triggering negative political reactions, such as non-compliance and harsh public critique, that would threaten its legitimacy and the coherence and uniform application of the law (see Garrett, Kelemen, and Schulz Reference Garrett, Kelemen and Schulz1998; Carrubba, Gabel, and Hankla Reference Carrubba, Gabel and Hankla2008; Carrubba and Gabel Reference Carrubba and Gabel2015; Larsson and Naurin Reference Larsson and Naurin2016; Larsson Reference Larsson2020). We believe that the Court wants to avoid ‘messing with people’ unnecessarily, that is, inadvertently inflicting negative social or economic impacts on European citizens, businesses, government agencies, political parties, and civil society.
We build on these assumptions about judicial preferences to generate predictions with regards to the Court’s engagement with third parties’ observations in its reasoning, and in particular, with respect to dismissals of arguments made in those briefs. We will refer to such dismissals throughout the following as “negative references.”
We are focusing in particular on negative references of the observations submitted by the European Commission and EU Member State governments, who are arguably the Court’s most powerful interlocutors. The European Commission always submits an observation, and the legal service of the Commission is generally perceived as one of the most significant “legal minds” in the EU law community. Although preliminary reference proceedings concern national courts’ questions in relation to a legal dispute in a particular Member State (or in the case of joined proceedings, very similar legal disputes in more than one Member State), Member State governments that are not immediately involved in the case also frequently submit their positions on these questions to the CJEU in observations. Member States have long understood that the CJEU’s answers to preliminary references originating from other Member States may have a significant impact on their own domestic politics, polities, and policies, and thus seek to make their voices heard, nudging the Court to issue a favorable interpretation (see Blauberger and Schmidt Reference Blauberger and Schmidt2017; Dederke and Naurin Reference Dederke and Naurin2018).
We posit that the Court will use negative references to observations as a tool for expanding the precedent and increasing the precision of the law. By explicitly and publicly dismissing a position put forward by the Commission or by a Member State, judges can close the door on legal arguments and interpretations that run counter to their preferred understanding of the law. Observations thus effectively become a vehicle for the Court to expand on its precedent beyond the question raised by the national court.
We expect that the CJEU is more likely to use this tool when there is a high demand for legal clarity and when serious alternatives to its chosen path are available. In cases with high legal uncertainty there is a higher demand for precision from the Court. Furthermore, legal uncertainty also generates good faith confusion about what the law is. When the Commission and the Member States have good faith confusion about what the law is, Member States’ observations are more likely to contain serious legal alternatives for the Court to engage with. This leads us to our first hypothesis.
Hypothesis 1 (H1). Negative references of observations are more likely when legal uncertainty is high.
Engaging with observations may also affect the reception of the Court’s judgment. We posit that negative references of Member States’ observations may have a positive effect on both the perceived legitimacy of the judgment and the likelihood that the new precedent will have an effective impact.
The legitimacy function of a negative reference stems from the perception of the judgment among Member States who have submitted observations with alternative holdings compared to that of the Court. By engaging explicitly with the submission of Member States who lost the argument – explaining why the Court did not agree with some of the content in the observations – the Court demonstrates that it has at least taken the arguments of these Member States seriously.
The judicial impact function of a negative reference, on the other hand, is derived from the potential mobilization of actors that may be willing and able to put pressure on a Member State government that is dragging its feet with regards to compliance with the Court’s case law. By referencing an observation in the judgment, the Court is moving information that would otherwise be accessible only to the Member State governments, the Commission, and the parties to the case, into the public realm. The content of Member States’ observations are not publicly accessible, and since 1995, the CJEU no longer publishes reports containing summaries of the observations. In the period 1995–2011, which we study here, summary information on the substance of the observations could only be gained by requesting access to the Reports for the Hearing, which were produced by the Reporting Judge in advance of the oral hearing in Luxembourg. The Reports would then be accessed only in the language of the case, that is, in the language of the Member State from where the case originated (Larsson et al. Reference Larsson, Naurin, Schroeder, Ghavanini, Björnberg, Coster, Näsström, Otero, Lindholm, Naurin, Šadl, Ghavanini, Brekke, Fjelstul and Hermansen2024).
Hence, it is not immediately obvious to even close observers which Member State argued in favor of which interpretation of the legal issues at stake. References to Member State observations in the judgments make this information easily available. Furthermore, where the Court negatively references an observation, it also becomes obvious that the Court disagrees with the Member State’s argument. We therefore propose that a negative reference may function as a “fire alarm,” sending a signal to domestic and international compliance constituencies, such as opposition parties, interest groups, legal academics, and the European Commission, that this particular Member State government may be reluctant to comply with the law in this area. Besides sounding the alarm, a negative reference also provides compliance constituencies with legal arguments against the position argued by the Member State.
We expect that the Court will use negative references to Member States’ observations as a tool for increasing the legitimacy of the ruling and the compliance pressure in particular in cases where Member States signal serious concerns with the ruling. Thus, both the legitimacy function and the fire alarm function lead us to the second hypothesis.
Hypothesis 2 (H2). Negative references of Member States’ observations are more likely when the Court takes decisions that contrast more seriously with the positions submitted by the Member States.
To distinguish the legitimacy and “fire alarm” mechanisms, we formulate a third hypothesis that specifically tests the latter mechanism. We posit that if the Court is motivated by activating compliance constituencies rather than increasing the perceived legitimacy of the ruling in the eyes of the concerned Member State it will have reason to make a difference between Member States from where the case originates and other Member States who act as third parties. The reason is that the “fire alarm” rings louder for the latter, who have not already publicly shown their cards. If the Court is mainly interested in legitimacy concerns of the Member State government that submitted the observation, they have no reason to make that distinction. The “fire alarm” mechanism of negative references is therefore less relevant for the Member State from which the case at hand originates compared to the other Member States. In effect, the former Member State has already been “outed” by the request for a preliminary reference from one of its national courts. This leads to our final hypothesis.
Hypothesis 3 (H3). The Court is more likely to negatively reference a Member State’s observation when that Member State’s position is not already public information.
To our knowledge, the theoretical expectations outlined in the preceding paragraphs are the first attempt to explain variation in the CJEU’s explicit engagement with Member State observations in its judicial reasoning. In the following, we bring novel, original data to bear on these hypotheses.
Empirical application
We make use of the IUROPA CJEU Database and center our attention on the CJEU’s judgments in preliminary reference procedures lodged with the Court between 1995 and 2011. For this time frame, we have access to data on the characteristics of the national courts’ questions referred to the CJEU as well as information on the positions that EU institutions and EU Member States took on these questions.
Preliminary rulings typically deal with more than one legal question referred by a national court. Given our interest in the effects of Member States and EU institutions’ positions on specific questions, we construct the following two datasets based on information supplied by the IUROPA CJEU Database.Footnote 4 For our first dataset, the units of observation are not judgments, but the 5,038 legal issues – that is, individual questions referred by national courts – which the CJEU resolved in preliminary reference proceedings.Footnote 5 The data comprises information on issue characteristics – for example, the area(s) of law it relates to, the size of the panel that heard the case, as well as the number of observations submitted by Member States and the Commission. For our second dataset, the units of analysis are the 12,647 individual observations submitted by EU Member States to the Court. This dataset allows us to identify how the Court engages with individual observations and to test whether certain characteristics of an observation have a tangible effect on the likelihood of the Court negatively referencing it in its response to a national court question.
In the first step of our analysis, we bring evidence from our issue-level dataset to bear on hypotheses H1 and H2. In a second step, we then evaluate whether evidence from our submission-level data corroborates any patterns identified at the issue-level. Finally, we draw on our submission-level data to test H3 regarding the CJEU’s incentives to publicize Member State positions. Throughout the following, our dichotomous outcome variable captures the CJEU’s reference to an observation with a negative sentiment (i.e., Negative reference = 1 and Negative reference = 0 otherwise). When considering legal issues, our outcome variable indicates whether the Court referenced at least one observation submitted by a Member State or the Commission negatively. When the units of analysis are Member State observations, the outcome variable indicates whether the submitted observation in question was referenced negatively. Descriptive statistics for the main variables introduced below, along with information on the protocol for coding the CJEU’s references to observations, are discussed in the supplementary materials. The supplementary materials also include analyses with a categorical outcome variable, including the CJEU’s positive and neutral references to observations, given that any effects identified in our main analyses may not be specific to negative references, but generally apply to the Court’s decision to reference observations.
Legal issues as units of analysis
Our first dataset allows us to identify whether the Court’s use of negative references is driven by the uncertainty surrounding a legal issue, as well by concerns about the legitimacy of its rulings and Member States’ subsequent compliance with the Court’s decisions. To capture the effects of variation in legal uncertainty (H1), we identify whether the Advocate General (AG) and the European Commission disagreed on the answer to the legal question referred by a national court, Conflicting positions: AG/Commission. The AG is an institutional unit within the Court, conducting independent investigations into each case and presenting its opinion to the panel of judges deciding the case. We expect that disagreement among these actors, the AG as the Court’s chief legal advisor and the Commission relying on the expertise of its legal service, reflects a higher degree of uncertainty on what the correct answer to a referred question should be. Following the same logic, we record whether the Court itself disagreed with the AG’s opinion. In addition, we consider the density of relevant existing CJEU case law as a proxy for legal uncertainty. We expect legal uncertainty to be higher in scenarios where fewer sources of relevant case law exist, suggesting that the legal issue touches on a subject that the Court had rarely encountered before. Previous research shows that the CJEU is strategic in its citations to legal sources (Larsson et al. Reference Larsson, Naurin, Derlén and Lindholm2017), therefore we count and then standardize the number of citations to different sources of case law in the AG’s opinion instead of relying on citation patterns in the Court’s ruling.
To test H2, we record how many Member States submitted positions that favored rulings that would further restrict states’ national autonomy for each legal issue, MS favoring restrictions, and vice versa, the number of Member States that opposed further restrictions, MS favoring no restrictions. We recorded the same information for the position voiced by the Commission, the AG, and the position issued by the Court itself. For some of our regression models, we considered the Net position of Member States, subtracting the number of Member States arguing in favor of further restrictions from the number of Member States arguing against restrictions.Footnote 6 For our analysis, we then standardized these variables. In line with H2, we expect the Court to make use of negative references when it restricts national autonomy against the interests of larger coalitions of Member States.
Empirical strategy
The units of analysis for the first part of our empirical application are the legal issues considered by the CJEU in preliminary reference proceedings. Including the CJEU’s own position on these legal issues in our regression models complicates their specification. We expect that Member States’ positions are directly and systematically linked to the Court’s use of negative references of observations, and we also expect that the CJEU’s position on a legal issue explains its use of negative references. Yet, in line with existing research (see Carrubba, Gabel, and Hankla Reference Carrubba, Gabel and Hankla2008; Larsson and Naurin Reference Larsson and Naurin2016; Ovádek Reference Ovádek2021), we have reason to believe that the Court’s position itself is endogenous to the positions submitted by Member States, the Commission, and the AG. Hence, Member State observations have both a direct and indirect effect on negative references.
To account for this relationship, we estimate a path analysis model, a variant of structural equation modelling. Figure 1 illustrates the structure of our path analysis. Predictors contained in the vector
$ {\boldsymbol{x}}_{\mathbf{1}} $
, namely the Commission and AG’s position on the legal issue, explain the CJEU’s decision to (not) restrict Member States’ autonomy. Yet, other predictors in the vector
$ {\boldsymbol{x}}_{\mathbf{2}} $
, comprising the positions signaled in Member State observations and the presence of disagreement among the Commission and AG have an effect on both the CJEU’s decision on the merits and its decision to negatively reference observations. Finally, the endogenous variable CJEU position itself has an effect,
$ {\beta}_{21} $
, on our main outcome variable, Negative reference.
Figure 1. Path diagram for path analyses in Models 3 and 4. The predictors’ variances and covariance are given by
$ {\phi}_{11},{\phi}_{22} $
and
$ {\phi}_{12} $
. Note: The residuals for the endogenous predictor
$ {y}_1 $
and the outcome
$ {y}_2 $
are given by
$ {\zeta}_1 $
and
$ {\zeta}_2 $
, with their respective variances given by
$ {\psi}_1 $
and
$ {\psi}_2 $
.
Estimating path analysis models allows us to account for the relationship between Member States and EU institutions’ positions and the CJEU’s ruling. However, another endogeneity issue is likely to affect our analysis. The CJEU’s ability to reference observations is conditional on Member States and EU institutions actually filing them. Dederke and Naurin (Reference Dederke and Naurin2018, 879) show that Member States submit their observations “both with an eye to influencing the broader development of European law and with the purpose of defending more immediate policy preferences.” Hence, we have reason to believe that Member States in particular are strategic about their decision when to submit observations. They are likely to do so when their interest is particularly salient or when they hope that signaling their position to the Court will make a difference for the latter’s ruling. Given Member States’ expectations of the CJEU’s actions influences their decision to file observations likely biases our coefficient estimates for the effects of Member States’ positions – the CJEU simply finds more “targets” for negative references in cases that restrict Member States’ autonomy, as these are the cases where governments predominantly file their observations.
Unfortunately, with the observational data available to us, we have no straightforward solution on hand to address this endogeneity issue. Across our models, we include a variety of control variables to capture important case characteristics, including case complexity via the number of subjects a case touches upon as listed in the EU’s official documentation service EUR-Lex, and the total number of observations submitted by Member States to capture a legal issue’s salience. Yet, we acknowledge that these controls cannot entirely mitigate our endogeneity concerns. We also control for the number of primary and secondary EU law sources cited in the AG’s opinions, and count the number of days it took the panel of judges to deliberate over the answer to a question after they had received the AG’s opinion (Deliberation time). Finally, we recorded whether the issue concerned primary and/or secondary EU law, and included fixed-effect controls for the year the proceeding was lodged, as well as the national area of law the issue concerned.
Effects of uncertainty and concerns on restricting autonomy
Figure 2 plots the coefficient estimates of our path analysis at the issue level. Model 1 includes the explanatory variables Member States favoring restrictions and Member States favoring no restrictions, while Model 2 includes the variable Net position of Member States instead. Both models are estimated with the Lavaan package for R (Rosseel Reference Rosseel2012). The top panel of Figure 2 plots coefficient estimates of predictors explaining the CJEU’s position on legal issues. The evidence is consistent with patterns uncovered in existing research: the CJEU is more likely to restrict Member States’ autonomy in its judgments when the Commission and AG advise the Court to do so, yet less likely to restrict national autonomy as the coalition of Member States signaling their opposition to restrictions grows.Footnote 7
Figure 2. Regression coefficient estimates with 95% confidence intervals for path analyses in Model 1 (N = 3,224) and Model 2 (N = 3,224). Notes: Both models include year and national area of law fixed effects. The top panel plots coefficient estimates for predictors predicting CJEU decision, the bottom panel plots coefficient estimates for predictors predicting Negative reference.
Turning to the bottom panel of Figure 2, a clear pattern regarding the drivers of the CJEU’s use of negative references emerges. Not only is the CJEU more likely to make use of negative references when it restricts national autonomy (see the positive coefficient estimate for CJEU position), the Court is also more likely to use negative references when it restricts autonomy against the interests of larger coalitions of Member States (i.e., the coefficient estimates for MS favoring no restrictions and Net position of MS are clearly distinguishable from zero and positive). This evidence is consistent with our expectation that the CJEU is more likely to engage with the positions it disagrees with when it issues rulings that starkly contrast with the preferences of Member States. In these scenarios, the CJEU has an incentive to explain in what ways and why it disagrees with Member States to bolster the legitimacy of its decision, and to point compliance constituencies to possibly recalcitrant Member State governments.
For both Models 1 and 2, we find no effect of disagreements among the AG and Commission on the substance of the CJEU’s answer (see top panel of Figure 2). However, we find positive, statistically significant effects for the variable Conflicting positions: AG/Commission (see bottom panel in Figure 2) on the CJEU’s choice to reference observations negatively in its answer. The Court is more likely to use negative references when key legal advisors hold different views on the legal issue. This pattern supports our H1, suggesting that the likelihood of observing negative references increases with legal uncertainty.
Our data allows us to dig deeper concerning the effects of the legal uncertainty surrounding an issue. Since reforms to the CJEU’s statute came into effect in April 2003, the Court has enjoyed discretion on whether or not to request an AG opinion on the questions referred by national courts (Statute of the Court of Justice, Article 20). We exploit this reform, coding a dichotomous variable AG opinion requested = 1 whenever the Court requested an AG opinion, and AG opinion requested = 0 otherwise for cases heard after this date. Following the Court’s reformed rules of procedure, judges ought to request an AG opinion whenever they believe that the national court’s question raises a novel point of law, and thus involves higher levels of legal uncertainty. Figure 3 plots coefficient estimates for Model 3, sub-setting our data for cases heard by the Court after the reforms had come into effect. Providing additional support to our first hypothesis, we find a statistically significant, positive coefficient estimate for AG opinion requested. The CJEU is more likely to use negative references when judges deemed it necessary to request an opinion from the AG, indicating a higher degree of legal uncertainty.
Figure 3. Logistic regression coefficient estimates with 95% confidence intervals for Model 3 (N = 2,305), sub-setting for cases after Nice Treaty reforms. Notes: Model 3 includes year and national area of law fixed effects. Coefficients’ standard errors are clustered by case proceeding.
To summarize, when taking legal issues as our unit of analysis, we find evidence that the Court’s use of negative references is tied to legal uncertainty, and a clear indication that negative references are a tool employed when the Court restricts Member States’ autonomy against the latter’s interests. In the following, we further investigate this second pattern. We draw on our dataset that takes Member State observations as its unit of analysis and investigate which of the observations submitted by the EU Member States are the targets of the Court’s negative references.
Member State observations as units of analysis
The structure of our second dataset is somewhat more complicated, given that our units of analysis are now Member State observations nested in legal issues, which themselves are nested in case proceedings. This hierarchical structure has implications for our regression analyses, which we further discuss below.
Empirical strategy
Our second dataset allows us to test our expectation that the Court uses negative references to Member State observations as a tool when its decisions starkly contrast with what Member States would have hoped for (see H2). It also helps us distinguish the mechanism behind these uses of negative references. We can set up a test for H3, expecting that the CJEU increases compliance pressure by referencing Member State positions negatively, which would otherwise not be public information.
To test H2, the variable Member State position captures whether the Member State in question voiced a position that opposes further restrictions to national autonomy (“No restrictions”), a position that favors further restrictions (“Restrictions”), or an ambivalent position on this dimension (“Ambivalent”).Footnote 8 In addition, we code the same information for the CJEU’s final position on the legal issue, CJEU position. We expect the Court to be more likely to negatively engage with a Member State observation whenever it restricts national autonomy against the expressed interests of a Member State.
Again, we can reasonably assume that the CJEU’s position itself is endogenous to the positions that Member States voiced in their observations. We therefore estimate a path analysis model, allowing Member State position to predict both the CJEU’s substantive position and the Court’s choice to negatively reference the Member State’s observation. The Lavaan package does not support multilevel modeling for binary outcomes, and we report results from our path analysis model without taking the multilevel structure of our data into account. For models not including the CJEU’s position as an endogenous variable predicting references, we estimate multilevel logistic regressions allowing intercepts to vary across both legal issues and case proceedings. Given the high number of estimated parameters, we rely on a Bayesian estimation of our logistic regression models, specifying non-informative normal priors for coefficient estimates and running four chains with 1,000 warm-up iterations and 20,000 sampling iterations.
Turning to H3, we employ two proxies. First, we code a dichotomous variable capturing whether the preliminary reference was lodged by a national court located in the Member State that submitted the observation, Case originated in Member State = 1, and Case originated in Member State = 0 otherwise. We expect the variable’s coefficient to be negative, as the Court should be more interested in highlighting potential future compliance issues in Member States that are not already in the focus of the case. Further, we code a dichotomous variable recording whether the Member State in question voiced its position during an oral hearing, Oral hearing = 1, and Oral hearing = 0 otherwise. Following H3, we expect the variable’s coefficient to be negative, as the Member State would have already publicly signaled its position on the legal issue.
Our multilevel logistic regression models include similar sets of control variables as in the previous section. We control for the size of the panel that heard the case, the number of subject matters the issue concerned, whether the issue dealt with primary and/or secondary union law, as well as national area of law and year fixed-effects. We first estimate a model including all submitted Member State observations in our data (Model 6) and another model sub-setting for Member State observations the CJEU eventually disagreed with (Model 7). In both of these models, we also control for Member States’ GDP per capita and include Member State fixed effects. Hence, our coefficient estimates allow us to compare nearly identical Member State observations, which differ with respect to a variable of interest, for example, whether the position was voiced at an oral hearing. Finally, Model 8 estimates our path analysis at the submission-level, including the CJEU’s position (ranging from “No restrictions” over “Ambivalent” to “Restrictions”) as an endogenous variable predicting the Court’s use of a negative reference. To reduce the number of parameters estimated in our path analysis, we exclude Member State fixed effects, but control for whether the observations were submitted by one of the four largest Member States (i.e., Germany, France, the United Kingdom, and Italy).
Results
Table 1 shows the estimated means of our Bayesian logistic regression coefficients along with their 95% highest probability densities. We observe two main patterns in our results. Starting with actors’ positions on restrictions to Member States’ national autonomy, we find clear evidence that Member States’ observations arguing against such restrictions are the most likely targets of the Court’s negative references. Coefficient estimates for the category “No restrictions” of our variable Member State position are positive and clearly distinguishable from zero in both Models 4 and 5.Footnote 9
Table 1. Regression Coefficients for ML Logistic Regression Models at the Submission Level
Notes: Outcome variable is the Court’s decision to reference an observation negatively. Coefficient estimates are posterior means with 95% highest probability densities. Model 5 subsets our data for positions conflicting with the CJEU’s decision. Both models include random effects, allowing intercepts to vary across legal issues and proceedings.
Turning to our coefficient estimates for the variables Case originated in Member State and Oral hearing, our evidence sheds light on which of the two mechanisms is behind the Court’s negative references to observations arguing against autonomy restrictions. Unlike the coefficient for Oral hearing, the estimate for Case originated in Member State is clearly distinguishable from zero, but positive, and hence contrasts with our expectation that the Court would be interested in drawing attention to potential compliance laggards that are not already in the spotlight. This evidence is at odds with our hypothesis that negative references are used as a tool to increase compliance pressure on recalcitrant Member States.
Now consider the results of our path analysis at the submission level. The Lavaan package treats ordered categorical variables as numerical, and for illustrative purposes we let the variables CJEU position, Commission position, and AG position range from “No restrictions” = −1 over “Ambivalent” = 0 to “Restrictions” = 1, while the variable Member State position ranges from “Restrictions” = −1 over “Ambivalent” = 0 to “No restrictions” = 1. The top panel of Figure 4 plots the coefficient estimates for our variables predicting the CJEU’s position on the legal issue with respect to Member States national autonomy. The positions of the Commission, AG, and individual Member States all have significant effects on the Court’s decision: the CJEU tends to follow the Commission and AG’s advice to restrict national autonomy, but is less likely to do so when a Member State argues against such restrictions.
Figure 4. Regression coefficient estimates with 95% confidence intervals for path analyses in Model 6 (N = 6,881). Notes: Units of analysis are individual Member State observations. Model 6 includes year and national area of law fixed effects. The top panel plots coefficient estimates for predictors predicting CJEU decision, the bottom panel plots coefficient estimates for predictors predicting Negative reference.
The bottom panel of Figure 4 plots the coefficient estimates for variables predicting the CJEU’s use of negative references and provides additional support to the conclusions we drew from our analysis of legal issues. The positive coefficient estimate for CJEU position shows that the Court is more likely to negatively reference a Member State observation when it restricts national autonomy in its decision on the merits. Further, the positive coefficient estimate for Member State position again highlights that it is the observations arguing against such restrictions that are the most likely target of these negative references. Put simply, the evidence from both our analyses, taking legal issues and Member State observations as units of analysis, strongly suggests that negative references are a tool the CJEU employs when it restricts Member States’ national autonomy, and it tends to engage with those observations that precisely oppose such restrictions in order to strengthen the legitimacy of its decision.
Discussion
Judicial law making by precedent-setting in high and international courts primarily takes place in the “grounds” section of a judgment. Here, a court develops its legal reasoning, explaining how and why it reached answers to the legal questions raised in the case and why these answers are justified. Although scholars of judicial politics and empirical legal research have long realized that courts’ power to allocate values in society lies in judicial reasoning, they have also had to grapple with the fact that systematically measuring judicial reasoning is harder than measuring outcomes in terms of wins and losses (see McGuire and Vanberg Reference McGuire and Vanberg2005; McGuire et al. Reference McGuire, Vanberg, Smith and Caldeira2009; Clark and Lauderdale Reference Clark and Lauderdale2010; Lauderdale and Clark Reference Lauderdale and Clark2014; Collins, Corley, and Hamner Reference Collins, Corley and Hamner2015).
This study contributes both to theorizing and empirically measuring an important and understudied element of judicial reasoning – the citation of amicus briefs. A large number of studies from different legal and political contexts have shown that amicus briefs can be an important source of information for courts and that they often have an impact on the outcome (Carrubba, Gabel, and Hankla Reference Carrubba, Gabel and Hankla2008; Collins, Corley, and Hamner Reference Collins, Corley and Hamner2015; Larsson and Naurin Reference Larsson and Naurin2016; Collins Reference Collins2018; Hazelton and Hinkle Reference Hazelton and Hinkle2022). However, despite the fact that referencing briefs is a common practice in precedent-setting courts, few studies have researched what role these references play in judicial reasoning, and the few studies that exist have not been able to systematically explain why and how courts engage in this practice (Canelo Reference Canelo2019; Bagashka, Chapa, and Tiede Reference Bagashka, Chapa and Tiede2024). We have focused our attention on negative references to briefs, which we argue are a particularly interesting potential tool for judicial law making. We propose that negative references to briefs give courts an opportunity both to enhance the legitimacy of its judgments and to develop its case law by increasing the precision of the precedent. By explicitly closing a door to a legal interpretation proposed in a brief, a court indicates not only what the law is, but also what it is not.
Our study of the CJEU, arguably the most powerful and politically significant court in Europe, provides important empirical evidence as to the validity of our suggested theoretical mechanisms. Overall, we find strong support for our claim that the CJEU negatively references briefs more often when it perceives a greater need to legitimize its decision. This is particularly the case when the Court takes a decision that constrains the sovereignty of Member States against their explicit will as submitted in their observations. Results presented in our supplementary materials show that these patterns are specific to the Court’s decision to negatively reference observations. We find no evidence that the CJEU seeks to legitimize decisions to restrict national autonomy with positive references to observations that support its position. Overall, these findings are in line with previous research on the CJEU that has found that the Court is conscious about the political context it is operating in and mindful of protecting its legitimacy and status (see for example Carrubba, Gabel, and Hankla Reference Carrubba, Gabel and Hankla2008; Martinsen Reference Martinsen2015; Larsson and Naurin Reference Larsson and Naurin2016; Larsson et al. Reference Larsson, Naurin, Derlén and Lindholm2017; Blauberger and Martinsen Reference Blauberger and Martinsen2020; Ovádek Reference Ovádek2021).
In addition, we find at least tentative support for our claim that courts use negative references of briefs to close off roads that it does not want its case law to travel. Some of our findings indicate that the CJEU is more likely to use negative references in situations where there is a higher degree of legal uncertainty, and thus a larger variety of possible interpretations and a higher demand for precision. These findings should be taken with some caution, however, as measuring legal uncertainty is challenging and not all of our proposed measures indicate significant effects.
At the same time, we have reason to believe that the CJEU is a most likely case to find evidence in favor of negative references as a function of legitimacy concerns, which may overshadow effects of legal uncertainty. Compared to domestic high courts, the CJEU is particularly dependent on the long run support of the Member States who often have high stakes in these preliminary references. The third-party briefs that the CJEU receives are therefore likely to carry more weight than the briefs from organized interests that has often been the focus of the SCOTUS literature. That literature has also found that the SCOTUS pays particular attention to the briefs submitted by the executive branch (the Solicitor General) and the state attorneys (Collins, Corley, and Hamner Reference Collins, Corley and Hamner2015). Therefore, the fact that the CJEU’s main interlocutors are the EU Member States is an important scope condition that should be considered when generalizing our findings to domestic high courts: legitimacy concerns possibly take priority in the minds of judges at the CJEU over attempts to address legal uncertainty. Nevertheless, the theoretical mechanism that we propose concerning the effects of legal uncertainty is generally applicable to precedent-setting courts, and future research will hopefully show whether this is a useful tool in other judicial contexts, in particular where the stakes for third parties submitting briefs and for the court are lower.
We did not find support for the proposition that negative references to briefs are also used as a fire alarm to alert compliance constituencies that the proper implementation of the case law may be at risk. It is possible that the CJEU does not perceive amicus briefs as useful in this regard, or that increasing information about compliance risks is not an important concern for the CJEU. We also need to acknowledge the possibility that our empirical measure – the distinction between home state briefs and third-party state briefs – is too crude to capture this mechanism.
Finally, although it may seem paradoxical that a judgment would be more favorably received when it is accompanied by a debunking of an actor’s argument, not explaining to a loser why they lost is likely to cause even more resentment. Similar to dissenting opinions, if a court perceives a need to respond to the arguments on the losing side, a likely positive implication is that it needs to sharpen its argument in the process.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/jlc.2025.4.
Acknowledgments
We appreciate the feedback provided on earlier drafts of the manuscript by participants in the IUROPA network and the Comparative Politics research seminar at LMU Munich, as well as those attending the workshop on “Litigants, Agendas, and Decisions at the European Court of Justice,” held at the University of Oslo in June 2023. We would also like thank the three anonymous reviewers and the editor for their useful suggestions on how to improve the manuscript. All replication material is available at the the Journal’s Dataverse archive.
Financial support
This research was conducted as part of the IUROPA project (www.iuropa.pol.gu.se), financed by the Swedish Research Council Project No. 2018-04215.
Competing interest
The authors declare no competing interests exist.
Data availability statement
All replication materials are available on the Journal of Law and Courts Dataverse archive.
Open access
