Litigant Success: How Litigant Configurations Relate to Legal Outcomes

Annulment actions with different litigant configurations are associated with different legal outcomes. Chances of legal defeat for the defending EU institution are higher whenever litigant configurations are complex rather than simple. This is because success rates in cases with simple litigant configurations are substantially lower than 50%. In cases with complex litigant configurations, success rates move closer to this 50% benchmark. Complex litigant configurations thus emerge in cases that are less predictable. This is because different political and legal conflict situations give rise to different litigant configurations. Times of institutional turbulence, especially, create legal uncertainty that not only brings more actors to take an active role in litigation but also makes Court rulings harder to predict. Explanations of legal success rates thus cannot treat litigant configurations as exogenous factors. Instead, we have to integrate the emergence of different litigant configurations into our models of judicial behaviour.


Litigant Configurations and JudiCiaL suCCess: What We KnoW
With the increasing empirical relevance of courts and litigation described under the well-known labels of judicialisation and legalisation, judicial behaviour has increasingly come under the focus of social science research (see Chapter 2). In this regard, the relationship between litigant configurations and the content of court rulings has become an area of particular scholarly interest. From a strict legalistic perspective, litigant configurations should not be an important influence on judicial decisions. Instead, decisions should be based on the legal merits of the case, not on the question of who presents the case. Consequently, any observed covariance between litigant configurations and legal outcomes should be purely coincidental. From this perspective, any aggregate-level variation of legal success might simply be the product of chance, untouched by the structural characteristics of member state litigants and their strategic interactions with the Court. Instead, what matters is the plain meaning of the legal texts, the intention with which the legal texts were written, existing case law, and precedents that determine judicial decisions (Segal and Spaeth 2002, 48). While court decisions can have political consequences and litigation, as we have also argued throughout this book, can be politically motivated, rulings as such are apolitical decisions; for the legitimacy of court decisions, a lot depends on whether the decisions are perceived as politically neutral. Regarding the Court of Justice of the European Union (CJEU), it thus comes as no surprise that legalist approaches deny 'the existence of ideological and socio-political influences on the Court's jurisdiction' (Burley and Mattli 1993, 45). From this perspective, any pro-integration bias of the Court results directly from treaty asymmetries (Scharpf 2002(Scharpf , 2007, and CJEU case law might reflect the 'inevitable working out of the correct implications of the constitutional text' (Shapiro 1980, 538).
Such legalistic conceptions have been heavily criticized by proponents of an attitudinal model of judicial decision making. This attitudinal model proposes that legalistic considerations 'serve only to rationalize the Court's decisions and to cloak the reality of the Court's decision-making process' (Segal and Spaeth 2002, 53). Rulings do not emerge automatically from existing law. No matter which legal method of interpretation one uses, rulings are always based on interpretations of the law. This process of the interpretation of more-or-less (un-)clear words, phrases, situations, and potential precedents adds the attitudes and ideological predispositions of judges to the equation (Segal and Spaeth 2002, 86-97; for the CJEU, see Höpner 2011;Vauchez 2012). Based on this understanding of judicial decision making, potential litigants-and researchers, for that matter-try to predict court decisions based on the ideological predispositions of judges. For the CJEU, this has proved particularly difficult because of the non-transparent decision-making process of the Court, where positions of individual judges cannot be identified. Abstracting from the attitudes of individual judges, researchers have assumed the Court to be generally very favourable of European integration. Otherwise, the argument goes, the Court's drive towards legal integration could not be explained. While this reasoning is hard to reject, recent attempts to open the black box of CJEU decision making have revealed that CJEU judges do not necessarily have uniform preferences regarding the development of the European Union (EU)'s body of law (Malecki 2012).
A third prominent theoretical perspective on judicial decision making is commonly summarized under the label of strategic approaches to judicial decision making. While this perspective does not deny that judges might hold relevant policy-related and institutional preferences, it emphasizes that judges are hardly able to act freely on those preferences. Instead, judges are constrained by the anticipated reactions to their rulings. As a result, decisions reflect strategic interactions between judges, within courts, between the court and the litigant, and between courts and public opinion (Segal and Spaeth 2002, 100).
In the EU context, this perspective supports intergovernmentalist accounts of European integration. As such, intergovernmentalist scholars have emphasized the strategic relationship between the CJEU and member state governments. Garrett et al. (1998) provide a formalized and strategic model of the relationship between member state governments and the CJEU in which the authors claim that the Court is a strategic actor that works to protect its institutional authority (Garrett et al. 1998, 174). This authority rests on both the perception of the Court's impartiality and integrity and on its ability to adopt rulings that are not overruled by subsequent legislation and are obeyed. More recently, Larsson and Naurin (2016) have demonstrated a strong correlation between the CJEU's rulings and the political signals it receives from member states. Ultimately, in both studies, the authors argue that CJEU decisions reflect strategic assessment (Garrett et al. 1998;Larsson and Naurin 2016).
In contrast to this strategic approach to judicial behaviour, the general litigation literature coined by US scholars has highlighted other reasons for which certain actors will be more successful in court than others. In this literature, the question about who has success in court has been on the research agenda since the 1980s. Concerning litigant groups, the evidence is quite clear-cut. Governments and public actors come out first, followed by businesses and other organized interests, while individuals only reach the lowest success rates in comparison with the other groups (Farole 1999). Usually, litigant success is associated with arguments about judicial constraint or litigants' capacity. In a comparative study of social activists' ability to succeed in court, Epp (1998) emphasizes economic resources available to the claimant as the most important support factor (for similar results on EU preliminary rulings, see Tridimas and Tridimas 2004). This purely economic factor partly overlaps with Galanter's (1974, 97;similar McGuire 1995;Haire et al. 1999) prominent repeat-player argument. Accordingly, resource rich claimants can afford to appear in court regularly and thereby gain the experience necessary to increase their chance of success in court. The relevance of capacity is also emphasized by studies analysing governmental litigation in the context of the World Trade Organization's system of dispute resolution. Only governments with a high degree of executive effectiveness are found to be able to navigate the complex procedures, learn effectively from experience, and keep up with the constantly changing body of case law (Kim 2008;Davis and Bermeo 2009).
Authors that consider ideological closeness to the Court as the more relevant strategic factor oppose such arguments, which stress (economic) characteristics of claimants. In a much cited study on federal or state courts in the United States, Sheehan et al. (1992) find that across different litigant groups, the ideological complexion of courts was more important to explain success than other factors. More recently, Skiple et al. (2016) also found substantial explanatory power of Supreme Court judges' ideological orientation-via appointment mechanisms-to matter for outcomes on economic conflicts.
At a more general level, judicial constraints can systematically affect litigant success. Studies along this line adopt a principal-agent perspective and assess whether national governments are able to effectively constrain the CJEU, which tries to avoid non-compliance and legislative overriding of its rulings. From this perspective, active participation in judicial proceedings by more powerful member states is likely to constrain the Court in its rulings because threats of legislative override and non-compliance with rulings from that side are more credible (Garrett and Weingast 1993;Carrubba et al. 2008). Given that more powerful states are seen to be less susceptible to the reputational costs resulting from non-compliant behaviour, the probability of winning, that is the probability with which the European Court of Justice should be found to agree with the litigant government, increases with this government's degree of political power.
With regard to multiple litigants, it has been argued that a threat of legislative override is reduced where member states appear to be divided over the legal question. In the EU context, which regularly demands high degrees of consensus or even unanimity in the Council, voting jointly becomes less likely in such cases. Whether or not there is empirical support for these theoretical propositions remains heatedly debated, however (Carrubba et al. 2008(Carrubba et al. , 2012Stone Sweet and Brunell 2012).
But the ratchet effect created by the arithmetics of decision making in the Council is not the only aspect heatedly discussed within this controversy. Some authors claim that the model is based on a misconception of the CJEU as an agent of the member states when it should really be considered to be a trustee (Stone Sweet and Brunell 2013). The trustee role, the argument goes, is distinctly different from the role of an agent and is characterized by three different aspects, '(1) the court is recognized as the authoritative interpreter of the regime's law, which it applies to resolve disputes concerning state compliance; (2) the court's jurisdiction, with regard to state compliance, is compulsory; and (3) it is virtually impossible, in practice, for contracting states to reverse the court's important rulings on treaty law' (Stone Sweet and Brunell 2013, 62). As trustee, the Court's decisions would rather reflect a logic of majoritarian activism. This means that it tries to produce rulings that reflect standard practices in many member states and are characterized by a high level of state consensus (Stone Sweet and Brunell 2013). This brief literature review can hardly do justice to the vast existing and emerging literature on the CJEU, let alone on judicial behaviour. Nevertheless, it hopefully serves to highlight that there is a controversial debate over the ability of powerful political actors to influence judicial decision making. While legalistic, attitudinal, and neo-functional approaches to CJEU decision making refute this claim, adherents of the strategic model argue that litigant configurations are an important influence on judicial behaviour. Accordingly, strategic models of court behaviour argue that when many powerful member states support a specific legal argument, the Court becomes more inclined to follow this argument than when many powerful member state governments oppose this particular argument. Therefore, the threat of member state non-compliance or legislative override is conceptualized with the help of member state's political power and the number of member states supporting or opposing particular arguments. This approach yields so called net-weighted observations (Carrubba et al. 2008(Carrubba et al. , 2012, which basically counts the number of legal observations of member state governments on either side of the legal argument and weights this number by the political power of the respective member state. Neo-functionalist accounts have heavily criticized this approach (e.g. Burley and Mattli 1993). Neo-functionalist accounts of European integration through law argue that the CJEU has-with the help of private litigants-promoted European integration well beyond the preferences of member state governments. Most importantly, the critique against strategic models of the CJEU emphasizes that because of the high number of member states and the heterogeneity of their preferences, any sort of threat of legislative override is hardly ever credible. Therefore, this threat should not be measured on a continuous scale. Only when the vast majority of member states were clearly opposed to a particular legal interpretation would this threat be credible. In all other cases, the threat would be absent (Stone Sweet and Brunell 2012).

Litigant Configurations and endogeneity: a neW approaCh
We use this chapter to highlight one further problem inherent in the empirical evaluation of strategic models of court behaviour that strongly rely on observed litigant configurations. Essentially, authors such as Carrubba et al. (2008) treat litigant configurations as factors that are exogenous influences on judicial decision making. The emergence of different litigant configurations is not explicitly theorized within such models. Empirical evaluations of these models thus rely on the assumption that cases that include many powerful actors are not systematically different in any relevant way from cases that do not include powerful actors, except for the different participant configurations. Therefore, a correlational relationship between litigant configurations and patterns of legal success can be interpreted as supporting the theoretically assumed causal relationship between these variables. We believe, however, that patterns of correlation between litigant configurations and legal success are better understood when litigant configurations are endogenized within the analysis and conceived in relation to the nature of the conflict. We argue that it is not necessarily the litigant configuration that produces certain rulings. The causal chain is much longer. As we have argued in the previous chapter, different litigant configurations emerge-to a substantial extent-as a result of different characteristics of the underlying situation. More specifically, we have argued that complex litigant structures tend to emerge in situations of institutional turbulence. Such institutional turbulences often trigger active litigation by several actors who not only have a stake in the outcome of conflicts but, importantly, also perceive the legal situation to be sufficiently unclear-the legal merits of the case being open to different interpretations-as to consider it worthwhile joining the case. As seen in Chapters 5 and 6, actors are typically more likely to engage in costly litigation where the chances of legal success are substantial, which is the case in situations of greater legal uncertainty. Success rates in annulment actions are rather low (around 25% across all cases and litigant configurations). Situations of greater legal uncertainty, that is, when issues disputed before the Court are not clearly predetermined by previous case law, should translate into a higher success rate, close to 50%. Greater legal uncertainty thus typically means higher chances of success for potential applicants and thus a higher likelihood to see complex litigant configurations as a result of additional actors joining the case in favour of the applicant. This is particularly obvious for private actors who generally litigate out of financial motivations and where the chance of success is a critical element of the risk-benefit analysis underpinning the decision to litigate. Yet legal uncertainty is also an important factor to public actors, such as national governments. As repeat players before the Court (Galanter 1974;McGuire 1995), they are unwilling to risk their reputation as serious partners in the legal discourse regarding EU law and European integration by pushing conflicts without legal merits. Besides, litigation before the CJEU consumes key human resources that need to be managed wisely (state attorney units are typically relatively small), which requires prioritization among possible cases where the legal uncertainty criterion does play a role. Member states or other public authorities are thus rather likely to launch annulment cases or join them in support of an applicant where chances of success are higher, that is, in situations of greater legal uncertainty.
Interestingly, the same should be true for actors potentially interested in joining the case in support of the defending EU institution. In situations of great legal certainty (i.e. great predictability of the court ruling), the necessity to intervene in support of the defending institution does not appear as important. As the defending EU institution is more likely to win anyway, the actors interested in the success of the defendant adopt a free-rider approach and refrain from investing resources into the conflict. By contrast, if the chances of the defending institution are lower, an actor interested in the defeat of the action might perceive its intervention in the case as being potentially able to tip the scales and help to obtain a favourable ruling. The incentive to join the case is thus higher. Hence, situations of legal uncertainty are also more likely to see at least one actor intervening in support of the defending EU institution than situations of high legal predictability.
All this is not to say that complex litigant configurations will never emerge in cases with a marginal degree of legal uncertainty and consequently an expected ruling. Yet the emergence of complex litigant configurations in this context is less likely as compared to the emergence of simple litigant configurations. Put differently, if conflicts with little legal uncertainty-that is, when the outcome of court rulings is rather predictable-do lead to annulment litigation at all, they tend to lead to simple rather than complex litigant configurations.
In sum, we argue that treating litigant configurations as exogenous factors is problematic when trying to analyse their impact on judicial behaviour. It is not necessarily the litigant configuration that produces certain rulings. Instead, it is specific characteristics of the underlying conflict situation that promotes specific litigant configurations and triggers respective rulings. Accordingly, any correlation between litigant configurations and legal outcomes is not necessarily the result of the litigant configurations' causal effect. Instead, the correlation is a reflection of the different character of underlying cases.
In an earlier study, we have made a similar point (Adam et al. 2015). While we found correlational evidence supporting arguments of judicial constraint, we emphasized that the characteristics of the litigant inform us not only about the abilities of this litigant to constrain or influence the Court in its decision making. Instead, the characteristics and motivations of the litigant tell us something about the kind of cases the litigant will bring to the Court's attention. In this particular context, we argued with the help of case studies and regression analysis that member state governments, which face strong subnational governments with a high degree of authority, are more likely to initiate annulment litigation against the European Commission for other reasons than trying to win the legal case. Instead, annulment litigation for those governments is often part of a two-level game. Adverse rulings might not harm them politically and in fact might even have positive electoral effects. This is particularly the case where such adverse rulings can be used as normative levers legitimizing domestic reform processes. These characteristics and motivations help us understand why national governments facing strong regional governments are substantially less successful in winning annulment cases than national governments operating within centralized political environments. They do not possess a lower level of legal expertise and they are not necessarily less likely to constrain the Court politically. Instead, they are somewhat more often inclined to initiate litigation in cases with only meagre chances of success in a legal sense, because they more often choose cases for their political rather than legal merits.
In this chapter, we make a similar argument. Yet it is not only the characteristics of litigants that contain information about the kind of cases litigants bring to the Court's attention. More generally, litigant configurations in specific cases contain information of the underlying conflict situations that the Court has to settle. Building directly on the arguments presented in the previous chapter (see Chapter 6), complex configurations tend to arise more often in contexts in which court behaviour is difficult to predict and less certain, that is, in situations of greater legal uncertainty. This is subsequently reflected by the Court's rulings fluctuating around a 50:50 chance of winning or losing in complex configurations, whereas annulment actions with a simple litigant configuration (simple applicant constellation v. simple defendant constellation) succeed in only about one in four cases.

Litigant Configurations and LegaL reasoning
In line with Chapter 6, we argue that success rates for cases with complex litigant configurations should be around 50% because these configurations tend to emerge in situations of lower legal certainty. In this section, we put forward an additional mechanism through which complex litigant configurations not only emerge in situations of higher legal uncertainty, but even contribute to increasing the legal uncertainty of the case under consideration. It is in this sense that we argue that litigant configurations can have a causal impact on court rulings. This is a second reason that we expect chances of legal success in cases with complex litigant configurations to be closer to 50% than in cases with simple litigant configurations.
Since the Court has to engage with the arguments brought forward by the litigants, the merits of the different legal arguments advanced by the parties are important. In this regard, complex litigant configurations seem not only to be reflective of the legal stock of a case but also to potentially affect the diversity of arguments presented in court. Therefore, since complex litigant configurations tend to increase rather than decrease the heterogeneity of legal perspectives presented to the Court, court decisions in these situations are again more difficult to predict.
While the legal merits of a case are obviously important, so are the arguments that build on this legal stock and their presentation in court. It requires adequate pleas and reasoning to present these arguments in a way that will convince the Court. Thus, participation by different litigants is anything but merely symbolic. On the contrary, our interviews indicate that the belief of being able to influence the Court's decision making with the help of convincing legal arguments is an important factor that brings public actors to participate in annulment litigation. The logic of an intervener's plea is to support arguments by adding new ways of reasoning and 'to place emphasis on a point that is particularly important' (COM_1, own translation; similarly COM_2; MIN_D_4; COMP_2; MIN_GA_2). In contrast, interviewees attributed little relevance to legal constraints on the Court. They argued that the Court is rarely impressed by the political weight that member states put behind certain demands or arguments. An interviewee's explanation that in horizontal annulments, high numbers of interveners are indicators of an uncertain defendant, supports this view. 'All the time the Council has a problem, there are large numbers of member state interveners-but this does not impress the Court' (COM_1, own translation). Instead, interveners matter because even when formally limited in the length of their pleas, they are able to add legal arguments, information, and nuances to the debate and thereby provide the Court with a wider array of pieces to choose from. What is more, they avail themselves of more time to do so, since their pleas can be submitted after the case is launched before the Court. This is important because-as we outlined in Chapter 3actions for annulment are subject to a tight time line and can be filed only within two months after a legal act is published. In this sense, more actors means more legal perspectives. To be clear, it is not the number of issues discussed in court that increases, but rather the number of different perspectives on a somewhat fixed number of legal questions (formally no additional aspect can be raised by interveners).
Of course, we cannot predict the diversity of legal arguments presented to the Court simply based on the complexity of litigant configurations. In our interviews, we have questioned litigants about their judicial strategies in cases involving other litigants on their side of the conflict. The answers indicate that the diversity of arguments is often reflected in complex configurations-but not always. This has a lot to do with the efforts of coordination between different litigants.
Across the multiple possible complex constellations of litigants, we found quite different efforts to coordinate. Public claimants often organized along existing networks of national legal experts, where member state officials are sometimes approached directly by mail to draw attention to an upcoming or submitted case and a related request for a friendly intervention (MIN_D_4). Coordination by phone or email assures that substantial support is forthcoming. Moreover, such cooperation avoids 'being in front of the Court and saying different things' (EP_1). 1 Finally, we came across cases where a given EU measure was being challenged in parallel by several applicants who never entered in contact with each other. This seems to be more likely for private actors who frequently do not avail themselves of the same inter-and transnational networks. In the renewable energy case (cases T-134/14 and T-47/15 presented in Chapter 5), the EU measure was attacked by fifty-one companies. One team of lawyers defended the interests of about ten of these companies. The legal arguments for these ten companies were de facto very similar to the legal arguments of law firms defending the remaining companies. However, the respective lawyers had no contact whatsoever with the lawyers defending the remaining forty-one companies involved in the conflict (LAW_5).
It is important to note that such strategic interaction before or during the process does not necessarily create convergence of positions and arguments. We found cases where litigants exchanged information on their respective cases and legal strategy that did not lead to an alignment of substantial arguments. In the milk quality case (T-683/15), the Commission attacked a Bavarian practice whereby investigations of the quality checks in the milk industry were paid by a fund of the Bavarian state. While the dairy industry pays into the fund, the fund itself is a public instrument. The Commission thought that companies themselves must pay the investigations and that the involvement of the Bavarian fund constituted illegal state aid. Both the Bavarian government and the association of Bavarian milk producers raised an annulment action against the Commission's decision. They exchanged information, but they put different arguments at the core of their reasoning. On the one hand, the association claimed, in line with existing CJEU case law, that the measure could not be classified as state aid, because no state resources were involved. The Bavarian government, on the other hand, was not convinced that they could win based on this argument. Instead, they argued that the scheme had already been in place before EU state aid law became applicable, thus falling into the category of the so-called existing aid, which is subject to a specific and less restrictive procedure. When the Commission finds an existing state aid to be in breach of EU state aid rules, it cannot ask the member state to recover the aid granted but rather asks it to put an end to the measure. In this case, while the dairy industry association claimed that the contested measure was not state aid, the Bavarian government acknowledged that it was state aid, arguing instead that it was a particular kind of state aid (MIN_BA_1). Here, we have two legal arguments that are contradictory. This is not necessarily a bad thing for the litigant, however, since it offers two alternative legal perspectives to the Court. If one does not convince the judges, the other may.
By contrast, the BMW case (T-671/14) is an example of a high level of substantial argumentative alignment among litigants. In that case, BMW raised an annulment action against a decision of the Commission that declared illegal aid by the German state of Saxony for BMW, which was meant to encourage the creation of a new factory in their region as contrary to EU state aid law (see discussion in Chapter 5). The Saxon government intervened in support of an action raised by BMW. Saxony hired expensive lawyers in order to bring additional argumentation to the Court. While BMW and Saxony had different lawyers, there was a lot of coordination among them. Both teams of lawyers shared all the information on the cases, met in Berlin, and developed strong ties. Every argumentation was exchanged and checked by the other claimant. While they had slightly different views-some points were more important for Saxony, some more for BMW-they did not disagree on the points put forward by the other party. The common argumentative strategy was to present to the Court a convincing story, with complementary legal narratives contributing to the same common interest (MIN_SA_1).
Complex litigant configurations enhance rather than reduce the diversity of legal arguments provided. If the arguments presented to the Court are to have any effect on the rulings, then a greater variety of legal arguments should, in principle, make rulings less predictable. In any conflict, a greater variety of legal interpretations provides the Court with more possibilities to diverge from its preestablished legal interpretations. In other words, independent of the legal stock preceding a given case, where a complex litigant configuration brings about a higher diversity of legal arguments, this increases the Court's capacity to diverge from the preexisting case law. Thereby, complex litigant configurations can add to the level of legal uncertainty: where more complex litigant configurations bring more heterogeneous legal arguments as well as varying perspectives to legal cases, CJEU rulings are again more difficult to predict than for simple litigant configurations with just one applicant and one defendant.

empiriCaL patterns of Litigant Configurations and LegaL suCCess
We now turn to empirical evidence on success in annulment cases. When going to Court, litigants seek a decision on a particular conflict. The Court can decide that the applicant is inadmissible and consequently reject making any decision. Typically, this is the case where the grounds raised are invalid. In all other cases that are not dropped or withdrawn, a judgement will eventually be made on the conflict. The plaintiff can be successful or lose the claim. We coded legal success based on the eventual decision of the Court taken at the end of the respective proceedings and clearly stated at the very end of the text of its judgements. Specifically, we consider applicants to have been successful whenever the Court completely or partially annuls the contested legal act. All other outcomes are treated as unsuccessful cases. Figure 7.1 depicts the share of successful annulment litigation over time. Figure 7.1 shows that on average, the success rate lies somewhere between 18 and 35%. The volatility of success rates was higher in the early years of integration. Note, however, that in these early years, overall numbers of annulments were substantially lower and therefore one single ruling influences the overall success rate substantially. Nevertheless, success in court seems to have been somewhat more likely for plaintiffs in the early decades than it is today. Despite highs and lows between single years, the overall trend has been rather stable since the number of annulments has started to increase in the 1980s.
With this picture in mind, we now return to the constrained court argument put forward in the debate. In essence, this argument considers success to be a function of the threat for legislative override. In contrast, we posit that success is a function of the characteristics of the conflict situation from which legal cases emerge. If we look at the patterns of litigant success for different litigant configurations, we find further support for our argument. To see how, consider Fig. 7.2 in light of strategic approaches to judicial decision making; in situations in which the Court settles conflicts with private applicants on the one side and an EU institution on the other side, political constraint should play hardly any role in the Court's decision-making process. In these situations with simple litigant configurations (n = 939), we observe that EU institutions lose only 22% of cases. Alternatively, to put it the other way around, private litigants win 22% of cases against EU institutions. If we compare this success rate to the success rate in which an EU institution faces challenges from private actors and from at least one member state government at the same time, the Court should-if anything-be more constrained and thus more sensitive towards the applicant's concerns. In fact, the higher success rate for the applicants in this complex litigant configuration of 32% appears at the first glance to support this proposition. Yet considering our argument about endogeneity of litigant constellations, we should not jump to quick conclusions. First of all, an analysis of variance (ANOVA) 2 that tests for significant differences in the mean success rate between the four different litigant groups, displayed in Fig. 7.2, indicates that the success rates of the first two groups (private v. EU and private/ MS v. EU) are not significantly different from each other. Second, and maybe more importantly, the other success patterns do not really fit the constrained court narrative. This may result from the fact that the constrained court narrative remains theoretically underspecified by failing to explicitly theorize the emergence of different litigant constellations. To begin with, it is difficult to see why such a constrained court would be less responsive to applicants when these applicants are just member state governments (25%) as opposed to member state governments combined with private actors (32%)-as the latter cannot threaten legislative override (see Fig. 7.2).
Similarly, and perhaps even more interestingly, we find that when litigating against an EU institution, a member state has more chances to succeed when the defending EU institution is supported by another member state. This finding is completely at odds with the expectations of the strategic approach, which would predict a lower success rate when a defending EU institution is supported by a member state. Instead, we observe that member state applicants are much more likely to succeed in court when they face a defence alliance consisting of the contested EU institution and at least one other member state, compared to an EU institution acting on its own; this cannot be brought in line with the argument about court constraint, either.
We conduct a one-way ANOVA to determine whether these nominal differences in success rates are statistically significant (see Table 7.1). The differences between the first three groups are not statistically significant. We do find, however, that the success rate of group four (MS v. EU/  Fig. 7.2 (MS v. EU/MS), where member state governments appear on both sides of the conflict. In line with the arguments presented in the previous chapter, these configurations tend to arise more often in cases that are not only perceived as important by member state governments (for whatever reasons). They also include a sufficiently unclear legal situation as to make the Court's ruling quite difficult to predict. This is reflected in a success rate of 51% for the applicant member state government(s). In situations of institutional turbulence, it is more likely that complex constellations comprise actors that are drawn to the conflict by different motivations. Put differently, cases that fall in the MS v. EU/MS category are not simply characterized by a different litigant configuration than cases that fall in the category MS v. EU. Rather, they represent a different underlying situation. The former constellation is more likely to emerge in situations of turbulence and greater legal uncertainty. Accordingly, member state governments do not have a better chance of winning when facing an alliance between an EU institution and (an)other member state government(s). Instead, in these cases, court behaviour is simply harder to predict. Consequently, the success rate for these cases is closer to 50:50.
We believe that this argument also helps to explain the patterns displayed in Fig. 7.3. When court rulings are easier to foresee, cases are unlikely to attract a high number of litigants. Actors in anticipation of losing are reluctant to invest the necessary resources just as actors anticipating sure success will rather free ride on the outcome of the case. Therefore, simple litigant configurations are more likely to emerge when the outcomes of court rulings are easy to foresee. In contrast, conflicts and related legal questions that invite complex applicant and defendant configurations are typically characterized by a high level of political relevance and a high level of legal uncertainty; legal uncertainty and emerging policy junctures ensure that plausible legal arguments can be brought forward on both sides of the conflict, possibly making a difference in the ruling.

distributive effeCts and feedbaCK of Winning or Losing a Case
Now that we have explored the sequence that links policy conflict with litigation, litigation with litigant configurations, and litigant configurations to judicial outcomes, we come back to the origin. What is the feedback effect of the Court's rulings on the political context out of which the legal action emerged in the first place? Our multilevel policy approach to litigation compels us to reflect on the policy impacts of rulings. How do the Court's rulings feed back into policy arrangements and institutional settings that had led to conflict? What kind of redistributive effects among policy actors can rulings have? We saw that actors generally initiate annulment actions for material gain, to protect or improve decision making competences, to maximize ideological preferences, or to improve political trust. To what extent do court rulings contribute to the achievement of these objectives that motivated policy actors to engage in litigation in the first place? Typically, when referring to winners and losers, the literature frequently sets success in court equal to winning. Far less attention has been devoted to distributive effects of CJEU rulings (but see Cappelletti et al. 1986). We contend that courts also have distributive impacts that may or may not correspond to success or failure in the judicial conflict. As we have argued in an earlier publication, 'it's not always about winning' (Adam et al. 2015). Losing the legal argument can, at times, be irrelevant for the political utility associated with litigation. Even more, losing can even be positively related to the political utility of rulings. Acknowledging this compels us to explore how court rulings contribute to the objective pursued by policy actors when engaging in litigation.

Winning Can be aLigned With LegaL suCCess
Success in court and winning of the underlying conflict can be, and most of the time is, inherently connected. This is particularly the case when ideological opposition to specific supranational legal acts drives litigation, when litigation is motivated by direct material concerns, and when litigation is motivated by the wish to obtain or protect decision-making competences.

Material Gains
The connection between judicial success and the maximization of the utility of litigation is particularly evident where litigation is motivated by a concern for material resources. Most of the private claimants seek direct material benefits when going to court. Clearly, where they have success in court, they win. We saw in Chapter 5 that public actors also hold direct economic interests that can be pursued via annulment actions. Virtually all annulment actions in the area of agriculture and regional funds involve financial issues. Here, member states litigate against the Commission's decisions to impose financial correction, consisting in refusing to reimburse to the member states a sum they spent under the Common Agriculture Policy or EU cohesion policies when the Commission's auditors find that national authorities have committed irregularities. EU rules on how EU funds must be spent are complex.
This increases the likelihood of irregularities. When a member state challenges a Commission decision denying the transfer of funds corresponding to the irregularity found, the member state expects that the Court will force the Commission to proceed to the transfer. When the Court rules that the member state is right, the Commission's decision denying the transfer is annulled and the Commission has to reimburse the money spent. Here, the correspondence between judicial success and policy objective is very clear.

Institutional Competences
We have seen that a public actor may litigate in order to protect its decision-making competences from an over-reaching EU institution. There are many cases in state aid, for example, where the member states or regional governments raised an annulment action against a Commission decision that was perceived as a clear instance of competence creep. Often, the member state loses. In Chapter 5, we mentioned the Leipzig-Halle and Dresden Airport cases (T-396/08, T-215/9), where the Saxon government litigated in order to put a halt on what was perceived as a competence creep of the Commission. In this case, for the first time, the Commission considered the construction of airport infrastructure as an economic activity in the sense of EU state aid law and posited that the Commission thus needed to be notified. This was new; up to this moment, the Commission had never put out this interpretation. Saxony's decision to litigate was mainly driven by the attempt to fend off the Commission's attempt to become more strongly involved in regional infrastructure projects. Yet it remained unsuccessful.
The ruling had a highly significant effect on the redistribution of competences between the member states and the Commission in the field of infrastructure construction. After the ruling, Joaquim Almunia, then the Commissioner for Competition, declared that the Court's judgement was applicable to all kinds of infrastructure, independent of the sector concerned. Such an extrapolation from airports to any kind of infrastructure was unexpected because state aid law is sector specific. This led to a huge increase in the number of state aid case notifications submitted to the Commission, which became overwhelmed by the increase in workload (MIN_SA_1). Nevertheless, the Commission further consolidated its new and enhanced ability to influence regional infrastructure projects by adopting several on sector-specific guidelines for regional infrastructure projects. 3 Particularly in the area of state aid, annulment litigation was not able to sustainably limit the Commission's competence creep. A series of cases dealing with a Spanish tax scheme that would grant tax deductions to Spanish companies that acquire shares of companies located outside the EU (T-219/10, T-399/11). In these cases, Autogrill SA, Banco Santander, and other companies attacked the Commission for declaring this provision in the Spanish tax code to be incompatible with EU law. According to the EU's state aid law, aid granted selectively to some companies but not to others is generally illegal unless the aid meets a number of exceptional criteria. In a very broad interpretation of this provision, the Commission saw this criterion to be fulfilled since only companies that acquire foreign companies benefit from the tax benefit but not companies that acquire Spanish firms. The applicants perceived this to be an erroneous application of the selectivity concept and went to court. Specifically, the applicants went to the General Court in first instance. The General Court shared the view of the applicants and developed stricter conditions that would have to be met by the Commission when trying to assess the selectivity of tax measures. For the private companies, legal success in court meant a clear victory since it helped them to realize their tax benefits. Also, the Spanish government welcomed this ruling as it effectively constrained the Commission's ability to intrude on its national tax policy.
Yet the Commission appealed the decision of the General Court before the Court of Justice. Given the high redistributive effects at stake for the member states, three of them-Spain, Germany, and Irelandjoined the case in support of the private companies in appeal. The Court overturned the ruling of the General Court based on the argument that the General Court had misapplied the selectivity criterion. 'For state aid specialists and tax lawyers, this decision was bound to be a landmark case, whatever it would turn out to hold'. 4 Legal defeat for the Commission would seriously compromise its impact on state aid provided through national tax measures, while legal success would pave the way for stronger influence in this area. The case was effectively seen as widely stretching the concept of selectivity, 5 which, mechanically, leads to a wide stretch of the Commission's capacity to use EU state aid law to veto national fiscal mechanisms. This redistribution of competences between the Commission and the member states in the field of tax policy was directly connected to legal success and legal defeat in this particular case.
Another policy area where annulment litigation has resulted in clearcut competence gains for the Commission, here acting as a plaintiff, is external policy. Importantly, the first annulment case ever launched between EU institutions falls into this category. Traditionally, member states would negotiate international agreements in all areas that do not fall within the EU's competence, such as in external trade deals. The ERTA case (C-22/70), launched by the Commission against the Council, deals with this right of the Commission to negotiate international agreements (see our discussion of this case in Chapter 5). The Council had authorized member states to negotiate and conclude an international transport agreement that included social rules for the protection of drivers (today, Article 95 TFEU). It did so by claiming that transport was an area of member state competence. The Commission felt that the Council had overstepped its competences and launched an annulment to shift the legal base so negotiation powers would fall on the Commission (Article 207, ex. Article 133 Treaty on European Union). The Court agreed that the existence of acquis communitaire harmonizing social provisions in transport (Council Regulation [EEC] No. 543/69 6 ) necessarily vested any international agreement that concerned transport in community powers, consequently excluding concurring powers of member states. Winning this case enabled the Commission to expand external policy competences to areas where the Community holds internal competences. This became known as the principle of implied powers and was further developed and fixed in the Nice Treaty (Cremona 2011). Success in court thus meant winning competences beyond the more narrow right to negotiate trade agreements and altered the relationship of the EU institutions.

Ideological and Policy Preferences
Furthermore, winning tends to be closely connected to legal success in court if the motivation to litigate is dominated by the wish to challenge a supranational legal act due to ideological opposition. Where a supranational legal act directly interferes with core beliefs of what is right and wrong as well as with policies that are built on these normative beliefs, then winning a conflict is inherently tied to the ability to win the legal case and thereby abolish the ideological threat.
A good example is the passenger rights case (cf. cases C-317/04 and C-318/04, discussed in Chapter 5). The European Parliament (EP) challenged agreements that the Council and the Commission had negotiated internationally. The Parliament considered these agreements to violate fundamental individual rights related to data privacy. In an attempt to avoid litigation, the Commission actively approached members of Parliament by trying to convince them that litigation 'was not a good idea' (COM_2). Yet the Commission failed to dissolve ideological concerns over data privacy simply based on strategic concerns. The EP won this case in court. This legal success was also a political success as it helped the Parliament to push forward the legislative agenda on the General Data Protection Regulation. 7 An example of an unsuccessful attempt to promote ideological preferences is the Spanish coal case (T-57/11), discussed in Chapter 6. This case was triggered by a decree adopted by the Spanish government that altered the rules of the Spanish electricity market in favour of Spanish coal mines. This was seen as one important step in trying to save the Spanish coal sector from further decay. The Commission considered that the Spanish measure was not contrary to EU state aid law and authorized the measure. This decision sparked controversy within Spain. Castelnou, a small company that felt discriminated against by the Spanish decree, raised an annulment action against the Commission's authorization, and several national actors, both private and public, joined the conflict to intervene in favour of either the applicant or the defendant. Greenpeace intervened in support of Castelnou in the hope of further containing attempts to subsidize the exploitation of fossil fuels in the EU. 'Nongovernmental organizations like Greenpeace are only very rarely given an opportunity to argue before the EU Courts. This case is therefore a special opportunity to challenge some three-quarters of energy subsidies in the EU that still go towards fossil fuels' (Simons 2014). Greenpeace's legal objective was thus to promote a rebalancing of the interaction between environmental rules and state aid law in favour of the former. Yet the Court's conclusion was unfavourable to this ideologically motivated view promoted by Greenpeace. In the ruling, the Court specified that where the Commission 'assesses an aid measure which does not pursue an environmental objective, the Commission is not required to take account of EU rules on protection of the environment', and 'limits the verification of compliance with the rules, other than those relating to State aid, solely to those rules capable of having a negative impact on the internal market' (General Court 2014, 2). This ruling clearly circumscribed the extent to which environmental objectives can trump other policy objectives in the context of state aid law. While the Commission had not pushed to extend its decision criteria to evaluate state aid in this specific context, the Court's judgement creates a barrier to do so in the future.

Winning Can be disConneCted from LegaL suCCess
Policy actors may benefit from a ruling of the Court in which they lose. Cases abound where the redistributive effect of the rulings on policy actors do not coincide strictly with the Court's decision about who the winner of the judicial conflict is. This disconnection between redistributive effect and ruling is particularly clear when the annulment action is raised with a view to maximize political trust domestically or to clarify (in addition to maximize) the distribution of institutional competences. This has specific implications for complex litigant constellations. As argued above, complex constellations are more likely to comprise compound motivations, particularly in cases with different types of actors involved. But where actors seek different objectives, the very same ruling can produce more than one winner or more than one loser on these different grounds. Consequently, annulment conflicts characterized by complex actor constellations are likely to be more nuanced in terms of winners and loser. In a way, rather than producing only winners or only losers, success and defeat are likely to be multidimensional.

Material Gains
When a member state loses a case over state aid against the Commission, legal defeat typically implies that the money flows back into the pockets of the state. In such cases where the state's aid measures are declared illegal retrospectively, the ministry responsible for budget easily feels like a winner (MIN_D_3). While we tend to treat member states as unitary actors before the Court because of their status as litigant, they host important internal tensions and policy conflicts. Here, we see that a court ruling rejecting the action of the state can have a redistributive effect between ministerial departments. A concrete example is the Apple case (T-892/16) we discussed in the introduction to this book. While we can assume the Irish Ministry of Finance to be delighted to see huge amounts of tax money being flushed into its coffers, the Ministry of Economic Affairs should certainly be less happy about gloomy investor prospects.

Institutional Competences
Also, cases in which decision-making competences are disputed, success in court might not always generate clear-cut winners. In Council v. Commission (C-409/13), the Council and Commission were at loggerheads over the right to withdraw proposals from the legislative process. Along with its quasi monopoly to propose legislation, the Commission has the right to withdraw legal acts from the inter-institutional decision-making process. Installed to allow checks and balances in the EU political system, this right has seldom been used. Therefore, the Council was astonished when the Commission withdrew its proposal for a regulation on general provisions for macro-financial assistance to third countries that face short-term balance-of-payment difficulties after it had altered its content. The Council criticized the Commission by noting that the Commission's right to initiate legislative initiatives did not imply a symmetrical right to withdraw proposals, particularly not for mere political reasons. Consequently, the Council challenged the Commission's decision to withdraw the proposal in court. The Court's judgement had ambivalent effects. One the one hand, the Commission won the legal case and successfully defended its right to withdraw this specific legislative proposal (COM_1). On the other hand, the Court also narrowed the Commission's ability to withdraw proposals in the future by 'clarifying that while this right exists, the Commission cannot invoke this right under any condition and in any kind of way it wants' (EP_1). The Commission's short-term success has thus also had a restricting effect in the long-term as the Court used the ruling to clarify the conditions under which the Commission would effectively be able to withdraw legislative proposals. There was thus a disconnection between imminent success in court and the winning of the general competence conflict.
Furthermore, disconnections between legal outcome and political success emerge in situations in which legal clarification is seen as a second best in cases motivated by the willingness to maximize decision-making competences. In those cases, litigants typically have a clear preference for winning the case. However, they also considered that losing this particular case would not be so bad either, as long as the judgement would clarify unclear rules and ambiguous distributions of competences (MIN_ ES_7). This way, even legal defeat is not conceived as a complete loss since the legal certainty produced by the judgements will help to avoid unnecessary conflict and work in the future.
Actually, this situation often emerges after treaty changes (Bauer and Hartlapp 2010). Here, the simple manifestation of a ruling can, in fact, produce winners across the board of litigants. For example, the new comitology system coming into force with the Lisbon Treaty troubled the Commission (Brandsma and Blom-Hansen 2017). Here, the EP claimed that any comitology act had to be considered an implementing act and thus had to be agreed upon according to a procedure endowing it with relatively more influence. The Council, in contrast, had an institutional interest in declaring all comitology acts to be delegated acts. Sitting uneasily between these two options, the Commission feared that it would be exploited by the two as 'a kind of a bargaining chip' (COM_1). A number of annulment actions were launched by the EP and the Commission on the usage of delegated or implementing powers (see, for example, cases C-427/12; C-65/13; and C-88/14). The situation was characterized by general disagreement over the concept of implementation versus delegation as well as uncertainty about what rules would apply. The Commission claimed, for example, that the concept of implementing powers would apply to the online platform EURES that allowed for EU-wide job advertisements and job searches to promote job mobility. It claimed that this was the case despite the narrowing of its usage under the Lisbon Treaty. While the Commission was not successful in court, it gained legal certainty for further practices. 'The judgement was useful to that extent, because the Court-actually it was dismissed so we lost the case-but we obtained clarification and the Court said very clearly what an implementing act can do and what it cannot do' (EP_1). The ruling also had consequences at the level of administrative practices. In an attempt to be better able to secure its substantial interests in the altered setting, the Commission secretariat general with the legal service and the secretary general issued an internal guideline advising Commission staff on how to deal with the situation from now on (COM_1). Thus, this case shows how CJEU rulings (indirectly) affect the daily practice of the EU institutions in their interaction.

Political Trust
Finally, where litigation is motivated by gains in political trust, the ruling itself might be of little relevance to determine winners and losers. In these cases, the value of the legal conflict for a plaintiff can be independent of legal success in court. Analytically, these conflicts are best conceptualized as being part of a two-level game (Putnam 1988), where actors can be indifferent towards legal defeat at the supranational level because the mere act of going to court helps them carry home political benefit at the domestic level. The decoupling of the decision to take an active part in judicial proceedings from expectations about the outcome of such proceedings is thus a result of the long duration of the proceedings and the short time horizons of elected officials who are able to reap immediate benefits from the initiation of litigation. When politicians opt for legal conflict based on its value for populist signalling and not on the probability of winning, legal defeat is likely to be more frequent.
A case that illustrates this is the Austrian transit (C-356/01) conflict. In this case, the Austrian government found it worthwhile to initiate annulment litigation against the Commission for refusing to reduce the quota of freight trucks that could legally transit through Austria. This quota had been introduced before Austria joined the EU in order to reduce the country's environmental burden from haulage companies travelling back and forth between Germany and Italy. Generally, this provision held that if the number of trucks in transit through Austria exceeded a certain threshold in any one year, the quota would be lowered for the following year to compensate for the excess. To implement this provision, the Austrian authorities installed a system whereby trucks in transit were counted electronically. In 2001, Austria demanded that the Commission lower the quota based on this data. However, the Commission refused to accommodate Austria's application because it had doubts about the correctness of the data the country had presented. The CJEU subsequently supported the Commission's decision. Following public protest, including several blockades of the Brenner motorway (the most important transit route through the Austrian Alps that connects Germany and Italy) in 1998 and 2000, the formation of social movements such as Transitforum Tirol and the involvement of environmental interest organizations (e.g. Alpenforum), transit traffic became a highly politicized issue in Austria. With the electorate organizing around this issue, Austria's government likely feared that accepting the Commission's position would endanger its perceived integrity, particularly in the affected regions. The annulment actions thus enabled the government to communicate its loyalty and commitment to national constituents and made the legal conflict public via the media. Signalling commitment to affected constituents was a more dominant rationale for the initiation of litigation than the prospect of legal success. Austria's transport minister, Hubert Gorbach, calmly explained in 2003 that the Court's dismissal of Austria's action for annulment had no implications for the country anyway, and thereby openly voiced the government's indifference towards the legal outcome of the dispute. 8,9 ConCLusion This chapter addressed the last step in our sequential chain argumentation. Having explored different motivations underlying judicial conflict and how they translate into litigant constellations, we turned to the question of how such complex configurations relate to legal success and broader policy and institutional outcomes.
In line with strategic models of judicial behaviour, we also argued that success rates in court vary systematically across different kinds of litigant configurations. Importantly, however, we developed a radically different interpretation of the relationship between litigants' configurations and ruling outcomes. While litigant configurations are typically treated as exogenous factors, we highlighted the need for models of judicial behaviour to endogenize these configurations. Only when we take into account that different litigant configurations tend to emerge in different legal conflict situations, will we be able to understand the empirical association between litigant configuration and court rulings.
More specifically, we argued that complex constellations are more likely to be characterized by higher legal uncertainty and greater variety of legal perspectives presented to and evaluated by the Court when adopting its ruling. This is what explains the substantially greater share of claimant success in complex as opposed to simple conflict configurations. Our statistical analyses, as well as the case study evidence, support this argument.
This argument reflects a similar idea to the one promoted by Davies (2018). Davies claims that the CJEU's recent tendency to side with member states rather than private litigants over questions of rights associated with EU citizenship cannot easily be attributed to a changing judicial perspective or to increasing member state influence. Rather, one has to consider that it is not the Court that has changed, but the cases. With less meritorious cases brought by private litigants, lower success rates are inevitable.
This chapter also brought new insights to the relationship between litigation and broader policy and institutional impacts. It is important to assess the different kinds of impacts that rulings may have, beyond the immediate success or defeat before the courts (Scheingold 1974;Lobel 1994;McCann 1994;NeJaime 2011). We followed up on this approach, extending it to all types of actors raising annulment actions in the EU, while linking it more explicitly with the multilevel policy conflict from which litigation emerges. Starting from the premise that there are a variety of political goods sought by litigants when raising an annulment action (material gains, institutional power, ideology, and political trust), we assessed the relationship between a favourable ruling and the litigant's achievement of the primary goal underpinning its decision to litigate. We found that legal success is often directly related to the litigant's broader objectives; especially when seeking the maximization of material gain, institutional power, and ideological preferences. Legal success can, however, also be entirely disconnected from the litigant's genuine victory in terms of it achieving the objectives underlying its motivation to go to court. Since 'it is not always about winning' (Adam et al. 2015), a litigant may be perfectly satisfied, even in case of legal defeat, as long as the action has allowed it to reach its objectives. We posit that this disconnection between legal success and genuine achievement is particularly likely when, first, litigants use litigation as a way to maximize political trust. Here, litigation is conceived as a symbolic act, more than as a way to change the legal order. Second, this is the case when litigants are seeking, as a secondary objective, legal clarification in situations of unclear distribution of competences. By emphasizing this relationship between legal success and a rulings' broader redistribution impacts, we link back ruling outcomes to the policy conflict from which litigation emerges in the first place, thereby closing the cycle through which litigation and courts intervene in multilevel policy conflicts in the EU. In the following final chapter, we summarize the individual sequences of our argument and our main findings. Moreover, we discuss the implications our study may have for further research in this area.

Cases Cited
See