Legal Methods for the Study of EU Institutional Practice

Methodological choices in the legal study of the role of EU institutions – The so-called doctrinal legal method is appropriate, provided that it includes the analysis of key elements of non-legal institutional practice – Simple distinction between the study of ‘law in the books’ and that of ‘law in action’ to be qualified – Doctrinal legal scholarship is meaningful only when it acknowledges and incorporates a certain amount of ‘law in action’


I
The dominant mode of research and writing on EU law is doctrinal research based on adherence to legal positivism: writing about EU law means exposing which norms of EU law exist on a given subject, how they interact, and how they are put into practice by the legal system. One reason for this continuing adherence to a positivist view of EU law is that the academic study of European law is still strongly fragmented along national lines, meaning that expectations of how EU law should be studied differ along those national lines. Given this fragmentation, reliance on the positivist legal tradition allows for the use of a common 'language' between those national communities and, thus, for a meaningful exchange of ideas. It is no wonder, for instance, that the legal services of the EU institutions all adopt a strongly positivist approach to EU law, as the use of that common positivist language allows for an effective discussion among them and with the member states' legal offices. 1 The argument of the paper is that this positivist approach is appropriate for the study of EU law, provided that a sufficiently broad view is taken of what constitutes a legally relevant practice. I will argue that legal positivism and its corollary, the doctrinal legal method, should not be narrowly limited to the study and exposition of written legal norms, but should also include the study of institutional practices that are not, or only partially, based on legal norms. The reason for this is that positive EU law cannot properly be understood (and described) without knowledge of key institutional practices; the 'black letter' of EU law remains a dead letter without reflecting in its analysis the various practices which EU institutions develop in shaping and implementing the formal legal norms. These practices are not legal norms themselves, but are not unlawful either (unless declared so by the Court of Justice). Knowing and understanding them is necessary for giving a useful account of positive law. They are 'legally relevant practices'. 2 By arguing for such a broad positivist approach, this contribution responds to the overall concern of this special issue to reflect on what is relevant for the purpose of developing a better understanding of EU law. 3 That being said, the legal positivist method, even taken in this broad sense, surely does not exclude a critical look at how EU law is being made and interpreted by the institutions, nor does it exclude the use of what can broadly be termed as law-in-context methods. Depending on the nature of the research question, the use of contextual methods (that is, of insights drawn from neighbouring scientific disciplines) can be useful or even indispensable. The important thing is not to adopt a zero sum approach: the study of EU law in context does not imply a rejection of legal positivism but is, on the contrary, made possible and meaningful by building on a positivist grammar of sources of law, legal concepts and relations between legal systems. This argument about the use of methods in EU legal research will be developed, in this paper, with reference to the role of actors in EU institutional law. Given the limited scope of this contribution, the focus will be on a particular category of actors, namely the non-judicial institutions of the European Union, and on their roles in the EU's decision-making process, as well as on the member states' role in the EU's decision-making process. I will not primarily consider the role of institutions and bodies tasked with a control or accountability function, ranging from the European Court of Justice to the Court of Auditors and the Ombudsman, although the Court of Justice will frequently be mentioned for its role in defining the legal significance of informal institutional practices.
The paper will primarily address the question of the extent to which the institutional practices of these actors are a significant element of, and for, the EU legal order, and therefore part of what a doctrinal approach to EU law should study. This enquiry is based on the view that EU law (like other legal systems) is produced and reproduced through a variety of legal practices, such as: law-making, law-application, adjudication, advocacy (in court or outside) and legal scholarship. Knowledge of those legal practices is part of having a comprehensive knowledge of the law. The legal roles of the EU institutions (i.e. the specific object of this paper) are primarily those of law-making and law-application. They perform these roles and their specific tasks by means of routinised ways of behaviour which we may call institutional practices. Some institutional practices are clearly part of the legal system, as they are directly prescribed by the text of the Treaties; for example, the basic rules on the appointment of the members of the Commission are laid down in Article 17(7) of the Treaty on European Union (TEU). Other practices, while not prescribed by the Treaties, have been formally incorporated into the EU legal system on the initiative of the institutions themselves by means of interinstitutional agreements. 4 The most interesting questions, from an epistemological point of view, arise with regard to practices that have only been partially formalised in written rules, or that have not been formalised at all. I will argue that some of these practices, while not being legally prescribed, are essential for knowing and understanding the EU legal systemand must therefore be understood and acknowledged also by legal scholars using the doctrinal method. Among those practices, to be examined below, are the trilogues taking place during the ordinary legislative procedure, and the conclusions of the European Council.
In addressing this question, I will distinguish between the expository and evaluative modes of doctrinal scholarship, 5 which will be discussed respectively in the following two sections. The former answers descriptive questions about the way the legal world is, whereas the latter subjects legal norms to an appraisal based on their coherence with other legal norms or their compatibility with higher legal norms situated, for example, in a national constitution or an international treaty. The evaluative mode, in this definition, remains internal to the law, as the criteria for evaluation are given by the legal system itself. The section after that will engage with a specific sub-question about where to draw the limits of doctrinal research, 4 Interinstitutional agreements started as an informal practice but the Lisbon Treaty revision recognized them as a formal source of EU law (Art. 295 TFEU). namely the extent to which 'borderline practices' situated at the fringe of the EU legal order should be addressed in the doctrinal study of EU law, both from an expository and evaluative perspective. The last section of the paper briefly deals with the use of non-doctrinal research methods as a complement, rather than an alternative, to the doctrinal method when studying EU institutional actors. For the sake of convenience, those non-doctrinal methods will be grouped under the umbrella term of 'law-in-context'. They allow for a deeper understanding of institutional practice than even the broadly defined doctrinal method, advocated in this paper, would allow. E  :          Even though the EU Treaties, compared to national constitutions, contain relatively detailed rules on the tasks of the EU institutions and on the ways in which they should perform their tasks, it is still true to say that some essential elements of the EU's law-making process cannot be understood by reading the text of the EU Treaties. Three such essential elements can be mentioned by way of example: (i) although the Treaties state that the Commission has an exclusive right of legislative initiative in almost all policy domains, we know that, in practice, the Commission's formal initiatives frequently respond to agenda-setting decisions and concrete drafts proposed by others; 6 (ii) although the Treaties establish a firm distinction between cases where the Council decides by qualified majority and by unanimity, we know that in practice the national delegations in the Council and its preparatory bodies prefer not to outvote each other and, instead, aim at reaching an overall consensus before deciding a matter; 7 and (iii) although Article 251 TFEU gives a detailed description of the stages of the ordinary legislative procedure, we know that trilogues have become an essential element of that procedure without a basis in the text of Article 251.
To what extent is such broadly shared knowledge about institutional practices part of legal knowledge?
A first part of the answer to that question is that those practices have, in fact, been partly formalised in legal rules, even though mostly not found in the Treaties. To understand the meaning of those partial rules, one necessarily needs to understand the broader practice which they reflect. Let us take the example of the trilogues. They originally developed as a clear example of what political scientists call 'informal institutional change', namely the development of 'mostly unwritten rules of behaviour which are not subject to formal third party dispute resolution (court decisions)'. 8 By now, some 30 years after the emergence of the codecision procedure, some of the elements of the trilogue process have been codified. The Interinstitutional Agreement on Better Law-Making contains some provisions relating to 'transparency and coordination of the legislative process', 9 but these are quite abstract and rather tangential to the core operation of the trilogues. The European Parliament's Rules of Procedure contain fairly detailed guidance on how European Parliament members should act in the context of trilogues, 10 but this reflects only the way the European Parliament participates in trilogues; there are no corresponding provisions in the Council's Rules of Procedure, which are entirely silent about trilogues. So, there are some legal rules dealing with trilogues, but they clearly refer to a broader reality which is not comprehensively described in any written legal text. This is, then, a case where a doctrinal legal analysis of EU law-making requires a basic knowledge of informal institutional practices. To put it in other words, a legal description of the ordinary legislative process that would limit itself to rendering the content of Article 251 TFEU, completed with the European Parliament's Rules of Procedure, would be unacceptably superficial. When advocating that legal doctrinal analysis should incorporate and acknowledge the existence of trilogue practice, I do not, in fact, stretch that much the boundaries of doctrinal analysis. Indeed, legal writing has traditionally dealt with the origins of a statute or individual legal rules by examining the so-called travaux préparatoires, which are nothing other than institutional practices. Trilogues are part of the travaux préparatoires of EU legislation, although the results of their activity are perhaps less easily accessible than most preparatory materials of national laws. The EU institution whose activity is least regulated by formal legal rules, and where institutional practice is particularly important, is the European Council. That institution has over the years developed a rich spectrum of practices, of which only some find a clear basis in the text of the Treaties. Whereas the powers and functions of the other institutions are described in some detail in the Treaties, thereby constraining their discretion to develop new institutional practices, this is not the case with the European Council. This presents a doctrinal study of EU institutional law with a choice: either to stick to the 'official' legal rules regulating the activity of the European Council and leaving it to others (namely, political scientists) to study the 'invisible transformation' 11 of that institution; or else, to study the full range of institutional practices of the European Council because they impact on how EU law is currently being made and because ignoring those practices would give a truncated view of 'living' EU law. The latter attitude is the right one, as I will try to argue by reference to some recent events.
The European Council's general institutional role, as described in Article 15 TEU, is to 'provide the Union with the necessary impetus for its development and define the general political directions and priorities thereof '. The next sentence adds: 'It shall not exercise legislative functions'. This priority-setting activity is laid down in the European Council's Conclusions, which typically contain calls for action by others, whether the other EU institutions or the member states. When doing this, the European Council sets a political direction but the necessary legislative or executive action that follows from there must be taken by others, and those others are not bound by the European Council's Conclusions. The Court of Justice confirmed this in a judgment from 2018, Poland v Parliament and Council. It ruled that it would not annul a decision adopted by Council and Parliament that diverges from conclusions of the European Council. Otherwise, 'the Parliament and the Council's powers would be compromised in favour of following the political will expressed by the European Council'. 12 So, whereas the power to give political directions is legalised (as it is mentioned in Article 15 TEU), the actual exercise of that power does not take the form of law. It could at most be considered as a form of soft law.
So much is clear. However, Article 15 TEU gives only a partial and therefore distorted picture of the real role played by the European Council. As argued by Luuk Van Middelaar in his book Alarums & Excursions, the European Council plays, in fact, five different roles in the EU's political system, depending on 11 See the use of that term in a political science study of the codecision procedure: H. the circumstances: the European Council acts as strategist (which is the role described in Article 15), but also as crisis tamer, as impasse-breaker, as shaper and as spokesperson. 13 The European Council's activity in 2020 provides two interesting examples of this broader institutional role, which is not covered by the simple wording of Article 15 TEU. 14 The first example is the European Council Conclusions of July 2020. They were adopted after one of the longest meetings in the history of that institution. The Conclusions contained a very detailed political agreement on the content of the Next Generation EU package (the name given to the EU's Covid recovery plan), as well as on the Multiannual Financial Framework for the years 2021-2027, some 65 pages of text altogether. The main points of contention during the European Council discussions had been the overall size of the recovery fund, and the appropriate mix of loans and subsidies within the fund, and mainly pitted the 'frugal' states against the others. This event can be described as an example of the European Council acting as 'shaper'. When using that term, Van Middelaar and others before him (such as Simon Bulmer in an article from 1996) 15 refer to the key role of the European Council in deciding revisions of the European treaties or accession of new member states. These decisions shape the nature of the European integration process. The July 2020 meeting fits in that category, I would argue.
Although no Treaty revisions were made, or even discussed, the European Council endorsed a new interpretation of some key constitutional norms of EU public finance and gave the green light to what amounts to a major expansion of the EU budget in order to deal with the Covid crisis. The content of the Conclusions of July 2020 is remarkable in two ways. First, by the way in which the Conclusions set out in great detail (including by means of tables containing precise financial amounts) the content of the Next Generation EU and the Multiannual Financial Framework; and second, by the peremptory way in which this happens: the paragraphs of the Conclusions are replete with the verbs 'will' and 'shall', indicating that the European Council considered that it could decide what the Union was going to do through the Next Generation EU programme and through the Multiannual Financial Framework. No words were spent on even a polite reminder that the formal adoption of this political agreement would require action by the other EU institutions. And yet, the formal adoption of the various elements of the package still had to happen according to the procedures set out in the various legal bases. In fact, the remaining five months of the year 2020 were spent on fine-tuning the July agreement within the Council, and negotiating its content with the European Parliament, which had codecision or consent powers for major elements of the policy package. The European Council Conclusions of July 2020 were not the final word on the recovery plan and the Multiannual Financial Framework, and the European Parliament managed to force through a number of changes to the package, although not on the essential points (with which the Parliament agreed anyway).
The second example is what happened a few months later, during the final stages of those same negotiations. A major political incident occurred when the governments of Hungary and Poland threatened to veto the whole package (which they legally could) unless one element of the Multiannual Financial Framework was removed from it, namely the Regulation establishing a general regime of conditionality for the use of EU fundsthe famous rule-of-law regulation. As the adoption of that Regulation was subject to codecision, the European Parliament had managed, during the trilogue, to strengthen somewhat the rule of law conditionality compared to what had been written in the European Council Conclusions of July, and Hungary and Poland were unhappy with that. At the European Council meeting on 10 and 11 December 2020, difficult discussions happened and eventually a compromise was found and expressed in the European Council Conclusions. The Conclusions sought to tweak the meaning of the Conditionality Regulation so as to make it more palatable to Hungary and Poland. 16 This then paved the way for the adoption of the Multiannual Financial Framework and the various elements of the recovery package, which Hungary and Poland had threatened to veto. What the European Council did here is to perform another of Van Middelaar's five functions, namely to act as 'impasse-breaker': the other institutions had reached a dead end in their deliberations, there was the possibility that the whole recovery plan would fail, and intervention by the Heads appeared to be the only way to solve the problem.
What is clear from these episodes is that the European Council occasionally seizes the initiative to break an impasse, or to shape a major new development in EU policies, and it does so in ways which are not expressly foreseen in the Treaty provisions on decision-making. This protagonist role reflects the fact that the European Union is still dependent on unanimity and single country vetoes for a number of important policy decisions. Given this institutional rigidity, the Union needs the direct involvement of the collective heads of government to take matters forward on crucial occasions, after which the formal institutional balance designed by the Treaties can take its course again. But that 'normal' operation is 16 Conclusions of the European Council meeting of 10 and 11 December 2020, EUCO 22/20, points 1-4. affected by the intervention of the European Council, and in order to understand the meaning of later laws produced by the EU's formal decision-making process, one needs to have a grasp of the European Council's political practice. In this particular case, knowledge of the legal content of the Next Generation EU programme, of the Multiannual Financial Framework and of the Rule-of-law Conditionality Regulation requires an examination of the non-legal practice of the European Council.

E  :      
In the law of international treaties, there is some doctrinal controversy as to whether treaties can be modified, not only by means of their formal amendment, but also by implication and without their text being changedeither by the emergence of a new customary law norm or (more likely) by means of subsequent practice of the contracting parties that supersedes the original text of the treaty. 17 In EU law, that possibility has been rejected by the Court of Justice. A clear ruling in this sense came in the Defrenne judgment of 1976, in which the Court stated that 'apart from any specific provisions, the [EEC] Treaty can only be modified by means of the amendment procedure carried out in accordance with Article 236'. 18 The Court's central concern might well have been that, if one were to accept the possibility of an informal Treaty amendment, the governments of the member states could unilaterally modify the foundations of the legal order without the checks and balances (in particular, the involvement of EU institutions) provided by the normal Treaty amendment procedure. Along the same lines, the Court of Justice insisted in a number of cases that 'a mere practice' of the EU institutions cannot override the provisions of the EU Treaties. 19 The Court made those statements in cases where one or other of the EU institutions claimed that 'this is how we have consistently been doing things in practice', whereas the other institution, or a member state, claimed that the practice was contrary to the Treaty text and therefore illegal. In such disputes, the Court always took the latter view, in the 17 This unorthodox form of treaty amendment is not expressly codified in the Vienna Convention, but some legal writers consider it to be covered by the general wording of Art. 39 Vienna Convention, stating that 'a treaty may be amended by agreement between the parties'.  name of preserving the institutional balance or the division of powers between the EU and its member states, as set out in the Treaty text.
Institutional practices are not only constrained by norms in the TEU and TFEU but also by unwritten institutional principles which the Court of Justice has developed over the years, of which the most important was the principle of sincere cooperation. That principle was eventually turned into two written rules of primary law, laid down in Article 4(3) TEU (the vertical declination of the principle, concerning the relations between the EU institutions and the member states) and in Article 13(2) TEU (the horizontal declination, in relations between the EU institutions).
The EU institutions are thus not free to develop new institutional practices of decision-making in the way they wish. They must respect the limits of the powers conferred upon them by the subsequent versions of the Treaties as well as other norms of EU constitutional law. The Court of Justice can, in principle, sanction any violation of primary law norms. Indeed, the Court often insists on the fact that the EU institutional order is entirely 'judicialised'. In the Les Verts judgment of 1986, it held that all acts of the Community institutions were, in principle, amenable to judicial review, and that the EEC Treaty had 'established a complete system of legal remedies and procedures designed to permit the Court of Justice to review the legality of measures adopted by the institutions', even though the Treaty text seemed to contain some lacunas in this respect. 20 The Court itself frequently highlights the emblematic nature of that judgment by repeating its key sentences again and again in later cases. 21 There is no doubt that this judicial review power constrains the practice of the EU institutions. Even when a practice does not take a legal form, it can, naturally, still be in breach of the law. Supported by their respective legal services, the Commission, the Council and the European Parliament are spurred by their legal services to respect the limits of the powers conferred on them by the Treaties, and the institutional balance established by the Treaties, because they know that their actions could otherwise be challenged before the Court. Sometimes, though, the institutions are taken by surprise by an unexpected Court judgment that affects 20 ECJ 23 April 1986, Case 294/83, Les Verts v Parliament, ECLI:EU:C:1986:166, para. 23. The textual lacuna which was at stake in this case was the absence of judicial remedies against acts of the European Parliament. The Court filled that lacuna by reading into Art. 173 EEC (the current Art. 263 TFEU) an implied right to bring actions for annulment against binding acts of the European Parliament. well-established institutional routines. This happened, for instance, in the De Capitani case, when the Court struck down the secrecy with which the institutions 'protect' the files of an ongoing trilogue negotiation. 22 Most of the time, though, the existence of judicial review does not inhibit the adoption of what could euphemistically be called 'creative institutional solutions'. In being legally creative, the institutions are aware that the Court of Justice is sympathetic to the smooth functioning of the EU's institutional system and will not unduly constrain the room for manoeuvre of the institutions. The Court of Justice's acceptance of institutional creativity is particularly visible when all three key institutions agree to engage in a new institutional practice that is then challenged before the Court by a member state or by a private party, through an action for annulment or a preliminary reference on the validity of the act. Two examples from the case law illustrate the Court's flexible attitude in such situations.
In the ESMA judgment of January 2014, it acceptedagainst the view defended by the UK governmentthat the EU legislative organs could lawfully create forms of EU-level implementation beyond the regime of delegated and implementing acts mentioned in Articles 290 and 291 TFEU. 23 In the case at hand, the European Securities and Markets Authority, an EU agency, had been granted the power to adopt binding emergency measures under an EU regulation in the financial services sector. The text of the TFEU nowhere mentions the possibility to attribute administrative decision-making powers to EU agencies, but the Court of Justice considered this to be an oversight of the Treaty authors; since the Treaty authors had, elsewhere in the TFEU, stated that acts of agencies can form the object of an action for annulment, this presupposes that agencies (at least some of them) can adopt binding decisions. By allowing this practice, the Court may be said to have tinkered with the central administrative role of the Commission recognised by the Articles 290 and 291 TFEU, leading to what has been termed 'a significant adaptation of the principle of the institutional balance to the new realities of European governance'. 24 The leniency shown by the Court can be explained by the fact that, unlike in cases of inter-institutional disputes, the ESMA case was one in which all EU institutions (including the Commission) positively supported the granting of decision-making rules to the agency, and the opposition came from one single member state.
A second example of the Court giving its blessing to an unorthodox institutional practice is the Pringle case, decided in November 2012. In that well-known judgment, the Court of Justice accepted, among other things, that the EU institutions can be 'borrowed' by the member states when the latter conclude a separate international agreement on matters related to the functioning of the EU. 25 In this particular case, the Treaty establishing a European Stability Mechanism provided that the Commission and the European Central Bank were to play a role in the preparation, implementation and review of financial rescue decisions adopted by the European Stability Mechanism. Mr Pringle had invoked Article 13(2) TEU as an argument to challenge this arrangement. According to that Treaty provision, the EU institutions shall act within the limits of the powers given to them under 'the Treaties' (meaning: the TEU and the TFEU, andarguably at leastno other treaties). On a literal interpretation, this could mean that it is not possible to give any additional functions to the Commission or the European Central Bank (or indeed the Parliament and the Council) under separate international agreements such as the European Stability Mechanism Treaty. Yet, the Court had accepted, in a couple of low-profile judgments of the 1990s dealing with development aid, that the Commission could perform tasks entrusted to it by the Member States under a separate international agreement. The EC Treaty, it had held in the Bangladesh case, 'does not prevent the member states from entrusting the Commission with the task of coordinating a collective action undertaken by them on the basis of an act of their representatives meeting in the Council'. 26 The importance of the tasks entrusted to the Commission and the European Central Bank in the European Stability Mechanism Treaty, compared to the rather anodyne administrative tasks of the Commission which had been challenged in the Bangladesh case, prompted the Court to add an extra condition for any institutional borrowing; the extra tasks should not 'alter the essential character of the powers conferred on those institutions by the EU and FEU Treaties'. 27 In applying the test, the Court foundwithout surprise that the European Stability Mechanism tasks entrusted to the Commission and the European Central Bank did not affect the essential character of those institutions' powers. And yet, whenever the member states make use of the EU institutions within such a satellite organisation (as they did in the European Stability Mechanism Treaty), they actually strengthen the powers of those EU institutions that are being 'borrowed', to the detriment of the ones (especially the European Parliament, in this case) that are not borrowed.
Both these examples show that the text of the Treaties is not only an incomplete, but also an unreliable, guide to EU institutional law. Even though the Treaties describe the tasks and duties of the law-making institutions in considerably greater detail than a national constitution normally does (or, maybe, precisely for that reason), there is still considerable room for practices that experiment with solutions that do not seem to be contemplated by the Treaty text. Those practices were declared legally permissible in the two Court cases mentioned above, and doctrinal knowledge of the EU institutional system therefore requires knowledge of those practices, and acceptance of the fact that institutional practices are permissible in the EU legal order even when they are not covered by written primary lawor may even seem, at first sight, to contrast with it.
Of course, this does not mean that doctrinal scholarship should obediently endorse the Court's generous acceptance of unorthodox institutional practice. It is a traditionally accepted part of the doctrinal method to critically examine the extent to which the institutions (and the member states) respect primary law when developing institutional practices, and also to critically comment on the way in which the European Courts deal with such questions. Thus, the Pringle judgment was criticised by a number of legal writers for not having exercised a sufficient judicial review of the institutional practices of the EU institutions and the member states. More recently, a doctrinal controversy arose about the legality of the European Council's impasse-breaking role in relation to the adoption to the Rule-of-Law Conditionality Regulation. 28 Did the European Council Conclusions of December 2020 constitute a massive interference with the institutional balance, as some commentators wrote after the meeting? 29 It was argued that the European Council had interfered with the EU's legislative, executive and judicial power. The alleged interference with the legislative power was that the European Council had attached a lengthy gloss to the text of the Regulation as it was adopted, shortly afterwards, by European Parliament and Council. The alleged interference with the Commission's executive power was that the European Council Conclusions stated that the Commission intended to adopt guidelines on how to apply the Regulation and would do so only after 28 See the previous section of the paper. 29 A. Alemanno and M. Chamon, 'To Save the Rule of Law You Must Apparently Break It', Verfassungsblog, 11 December 2020, 〈https://verfassungsblog.de/to-save-the-rule-of-law-youmust-apparently-break-it/〉, visited 14 November 2022. the Court of Justice had decided the action for annulment of the Regulation that Hungary and Poland had introduced in the meantime. 30 The alleged interference with the judicial power was that the European Council, by constraining in this way the Commission's actions, had tried to give a suspensive effect to an action for annulmentsomething which is not foreseen in the Court's statutes or rules of procedure. These arguments may not be entirely convincing, but they are a good illustration of the way in which doctrinal scholarship, in its evaluative mode, canand shoulddeal with informal institutional practices.
H     :     '-' The legal positivist account of EU law is based on a fundamental distinction between law and non-law but, as the preceding examples show (and many more examples could be given), that distinction must accommodate the existence of informal institutional practices that either 'cross the line' by becoming legal (as with the interinstitutional agreements or the decision-making powers given to agencies), or do not cross that line but are nevertheless crucial to a proper understanding of positive EU law (as with trilogue routines and European Council conclusions). In this section, I would like to draw attention to another border that legal positivist accounts care about (and should continue to care about), namely that between EU law and other forms of law (be it international law or national law). Here again, institutional practices straddle this boundary and, hence, pose a challenge to doctrinal studies of EU law as to how to deal with such practices.
Throughout the history of the European Union, a curious institutional phenomenon recurred again and again, namely meetings that take place between representatives of the governments of the EU member states to discuss 'their own business' in the margin of an official EU meeting. Government representatives meet, of course, very regularly within two EU institutions (namely, the Council and the European Council) and their preparatory bodies; but they also, occasionally, leave that EU framework without leaving their physical meeting place. They then continue their deliberations as a 'conference of the representatives of the governments of the Member States', which is not an EU body but an international law body whose members act in the exercise of the foreign relations 30 The language used in the European Council Conclusions was that 'the Commission intends to develop guidelines' and that 'until such guidelines are finalised, the Commission will not propose measures under the Regulation'. In other words, the Commission agreed to do this, not because it was ordered to do so by the European Council but out of its free political will which was apparently conveyed by its President who, as we know, is a member of the European Council. powers given by (and constrained by) their national constitutions. Within such a conference, they adopt legal acts that are variously named but often constitute, in effect, binding agreements under international law. These 'acts of the representatives', as I will call them for the sake of brevity, are a subgroup of the larger category of agreements concluded between EU member states in connection with the operation of the EU. 31 What distinguishes this subgroup is that the acts are not adopted in a full-fledged and separately organised diplomatic conference but in the margins of a meeting of an EU institution, which may be either the Council or the European Council. The logistic closeness between EU action and international cooperation between the EU member states may result in some of the participants to the meetings not being fully aware that, in legal terms, they are hovering between the EU part and the non-EU part of the composite European constitution.
The fact that the 27 members of the Council of the European Union or the European Council can decide to 'switch hats' and become representatives of their countries at a non-EU diplomatic gathering is evocative of the quick adaptability of the chameleon, switching its colours from black to green according to changes in its environment or in its emotions. 32 One example of this institutional practice happened in December 2016. The Conclusions of the European Council meeting of 15 December 2016 include an Annex containing a 'Decision of the Heads of State or Government of the 28 Member States of the European Union, meeting within the European Council, on the Association Agreement between the European Union ( : : : ) and their Member States, of the one part, and Ukraine, of the other part'. 33 The switching of hats was not announced on the official agenda of the European Council meeting. 34 At some moment of that European Council meeting (but we do not know when exactly), the Leaders (as they like to be called these days) ceased to function as members of the EU institution and became (maybe without personally realising the nature of this legal transformation) representatives of their states concluding an executive agreement under international law.
The reason for this chameleonic event was the problem that had arisen in the Netherlands, where a consultative referendum had rejected the approval of the association agreement concluded by the EU, and its member states, with  193 at p. 222-234. 32 In chameleons, colour change is determined by environmental factors such as light and temperature as well as by emotions such as fright (Encyclopaedia Britannica,15 th edn,Vol. 3,Micropaedia,p.69 Ukraine. In order to allow the Netherlands to go ahead with the ratification of the agreement, and thus to save the entry into force of this mixed agreement, it was considered necessary to adopt a binding interpretative agreement that clarified that the association agreement did not do anything that the Dutch no-voters seemed to have feared. 35 The legal service of the Council had usefully given an opinion in which it argued that the Decision 'should : : : be regardedalthough it does not require the accomplishment of the formalities generally needed for selfstanding agreementsas an instrument of international law, by which the EU Member States agree on how they understand and will apply, within their competences, certain provisions of an act by which they are otherwise all bound'. 36 In other words, the Decision was a binding one, but it did not modify the content of the association agreement with Ukraine. This hat-switching trick enabled the Dutch ratification of the Ukraine agreement. Although this act itself is not part of EU law, it cannot be ignored by doctrinal studies of EU external relations law, since it was clearly instrumental to the operation of the EU-Ukraine agreement. In this case, the act of the representatives supported the operation of the EU legal system: the member states acted collectively to reassure the Dutch government (and, through it, the Dutch voters who had rejected the Ukraine agreement) that the agreement would not cause mass immigration from Ukraine or EU membership of that country.
Nothing was lost in making those superfluous statements and something important was gained, namely the entry into force of the association agreement. In other cases, the capacity of the member state representatives to quickly switch hats so as to 'leave' and 're-enter' the EU legal order can be detrimental to the integrity of the EU legal order. If the members of the Council or the European Council can at any time transform themselves into an impromptu diplomatic conference, the institutional balance established by the EU Treaties, and the categorical distinction between EU law and non-EU law, made by the Court of Justice, 37 could become very fragile. The question therefore arises The ECJ has consistently declared itself incompetent to interpret or enforce international treaties concluded by the member states (except, of course, for the founding Treaties), or collective decisions adopted by representatives of those member state, for instance when they appoint members of the Court of Justice. See, most recently, ECJ 16 June 2021, Case C-684/20 P, Sharpston v Council and Conference of the Representatives of the Governments of the Member States, ECLI:EU:C:2021:486, para. 39. whether such chameleonic behaviour is compatible with the states' obligations under EU law.
The potentially negative effects of this behaviour on the integrity of the EU legal order are particularly visible when the acts of the representatives have farreaching policy implications. One example stands out in this respect, namely the EU-Turkey statement of 2016. 38 The document with this name was published on the Council's website 39 but, according to the General Court, it was not an act of the EU at all, but an informal agreement reached by the collective member states with Turkey. 40 The General Court based itself, for that conclusion, on the preparatory documents of the meeting (especially the agenda) which appeared to show that a meeting of the European Council on 17 March 2016 had been followed, the next day, by a meeting of the Heads of State or Government of the member states together with their Turkish counterpart.
In other words, the 28 heads of government had 'switched hats' overnight, even though the press release made it wrongly appear that the action had been taken by the European Council. Acknowledging this switch, the General Court declined jurisdiction when a complaint was made that the EU-Turkey deal breached the fundamental rights of asylum seekers crossing from Turkey to Greece. Both the chameleonic behaviour itself, and the General Court's acceptance of it, were widely criticised in academic commentary. The member states, supported by the Commission, decided that the switch was instrumental in securing a sorely needed deal with Turkey that would help to stem the mass immigration into the EU from that country and to facilitate the EU's migration management policy. But many observers found this action to be detrimental, not only to the rights of asylum seekers, but also to the constitutional integrity of the EU legal order. 41 This is, once again, a good example of how doctrinal legal scholarship of EU law can, and must, examine institutional practices that lie across the formal dividing line between EU law and other legal systems.
L--     ;    In the preceding sections, I have argued that legal positivism and its corollary, the doctrinal legal method, should not be narrowly limited to the study and exposition of written legal norms (so-called 'black letter law') but should also include the study of institutional practices that are not, or only partially, legalised. This general contention is particularly relevant, arguably, for the study of EU institutional law, as the European Union institutions (and the member states that interact with those institutions) have over time developed a large number of institutional practices 'in the shadow of the law'. However, the additional use of non-doctrinal methods helps to get an even better understanding of institutional practices. Admittedly, the line between a practice-oriented doctrinal analysis and a rulesensitive contextual approach to the law is not very neat. In fact, in the preceding sections, that border may sometimes have been straddled. For instance, when discussing the roles of the European Council, this paper relied on the insights of Van Middelaar who is not a legal scholar but a historian and political theorist. These insights helped to clarify certain European Council practices that occurred in 2020, and allowed for a richer understanding of the way in which EU legal instruments emerged during that time; they also helped, in combination with a doctrinal analysis of the legal concept of institutional balance, to reflect on whether the European Council's creative practices in 2020 were actually compatible with the Treaties. Was this then a practice-oriented doctrinal analysis or rather a law-in-context approach? Admittedly, the methodological boundary between the two cannot always be clearly drawn.
Generally speaking, what law-in-context adds is the possibility to rely on the findings of other disciplines (or, in its more ambitious form, the possibility to engage in interdisciplinary research oneself ) in order to offer a better and richer understanding of legal questions. 42 In the present case, that richer understanding could include answers to questions about EU institutional law that a doctrinal analysis (even the broad one advocated in this paper) does not answer well, or at all, such as: why did an EU legal norm or institutional practice develop in 42 See, generally, Dumont and Bailleux, supra n. 5; and, for the specific case of EU law, the editorial by Francis Snyder in the first issue of the European Law Journal, defining the new journal's main purpose as being 'to express and to develop the study and understanding of European law in its social, cultural, political and economic contexts': F. Snyder, 'Editorial', 1 European Law Journal (1995) p. 1. the way it did, and how did actors shape that norm or practice either openly or behind the scene? How did that legal norm or practice impact on social reality? And how should we evaluate that norm or practice from a normative or moral perspective? The materials to be used for answering those different questions are obviously diverse and they will, consequently, belong to different scientific disciplines, ranging from sociology and history to economics and political science, and to political philosophy.
I cannot engage in detail with this here, but would like to offer just a few examples of how a law-in-context approach can enrich the legal scholar's understanding of the role of EU institutional actors. Sociological studies can help us understand how an EU institution actually functions 'from the inside'; see for example the ethnographic work by Frédéric Mérand about the operation of the Moscovici cabinet in the Commission, and its relations with other parts of the institution, in the years 2015 to 2019. 43 Works of historical sociology tracing the emergence of the role of 'Euro-lawyers' in the EU's institutional framework can help us understand how legal reasoning permeates the whole functioning of the institutions. 44 When it comes to the institutional functioning of the EU, political science is a privileged source of knowledge. 45 Every legal scholar knows that, when a treaty legal basis provides for unanimous decision-making in the Council, each national government possesses a formal veto power; however, it may not always be possible, in practice, for that veto power to be exercised. In-depth studies by political scientists help us to understand when and why a formal veto power gives a national delegation an effective power of obstruction, and when it does not. 46 Political science studies can, similarly, shed light on why the European Parliament voted in a certain way about a particular piece of EU legislation both through a micro-study on that particular legislative file or through a macro-study analysing the general voting patterns of the European Parliament. Political science studies can, more generally, help the legal scholar understand how informal institutional change impacts on legal reality even when the formal legal rules do not change. 47 43 F. Mérand, Un sociologue à la Commission européenne (Presses de Sciences Po 2021); also available in English: The Political Commissioner: a European Ethnography (Oxford University Press 2021). Returning to the general theme of this paper, the use of a law-in-context approach is often appropriate for legal researchers working on the EU institutional system. Much depends, though, on the way those legal scholars formulate their research questions and also on the format of their writing. Whereas a contextual approach may usefully complement the doctrinal method in a monograph or doctoral dissertation, a short comment on a court case may not lend itself so easily to a contextual approach, due to space limits or, in some cases, the expectations of the readership of the publication. However, even for scholars choosing to adopt a doctrinal method, the application of that method must allow for an understanding of institutional practice that goes beyond the analysis of formal legal rules. The reason for this, as I have tried to argue in this paper, is that legal norms of EU institutional law are often incomprehensible unless the researcher is prepared to engage with informal ('non-legal') institutional practice. The paper has also sought to qualify the simple distinction between the study of 'law in the books'which would be the domain of doctrinal legal scholarshipand that of 'law in action'which would be the domain of socio-legal scholars. I argue that doctrinal legal scholarship is meaningful only when it acknowledges and incorporates a certain amount of 'law in action'; the law is not only what is to be found in the books but also what happens in the real world and is not mentioned in the books.