External Differentiated Integration: Legal Feasibility and Constitutional Acceptability

The concept of external differentiated integration has received only modest attention from the legal community. As such, the concept’s contours have been shaped primarily by political science. There is therefore a pressing need for a stronger legal orientation to underpin discussions of this concept. This paper seeks to construct an analytical framework for a legal exploration of external differentiation and to map its ranging landscape. In assessing the legal feasibility and constitutional acceptability of external differentiation, the paper distinguishes between external differentiation that is driven by the conclusion of an international agreement such as the EEA Agreement, and external differentiation that flows from the adoption of an internal EU act, such as the General Data Protection Regulation. Alongside the legal analysis, the importance of the paper lies in developing an analytical tool to measure the degree of ‘legalization’ that is inherent in international agreements giving rise to external differentiation and in the recognition that external differentiation can be achieved as a result of unilateral instruments as well as through the conclusion of international agreements.


Part I: Introduction *
According to the project proposal in InDivEU, external differentiated integration (external differentiation or external DI) arises when 'EU rules are legally valid in at least one non-member state for some time'. 1 While it places the notion of legal validity at the core of the definition, it does not explain what is meant by this concept. For the purpose of our discussion of external differentiated integration, we propose to adopt a broad understanding of legal validity to include situations in which EU law is applicable in some form within one or more third countries. We explain what we mean by this below.
In our overview of external differentiation, we distinguish between external differentiation that is driven by the conclusion of an international agreement between the EU and at least one third country (Part II), and external differentiation that may be considered to be unilateral because it flows from the adoption of an internal EU act (Part III). These two forms of external DI are very different from each other. Where external DI arises as a result of an international agreement, EU law is rendered applicable in some form in the specific third countries that have concluded the agreement with the EU. By contrast, where external DI arises as a result of a unilateral EU act, EU law is rendered potentially applicable to all countries in the world. While unilateral external DI is thus 'uniform' in this sense, in practice, EU law will only apply to the extent that a third country or an actor within a third country has an incentive such as gaining access to the EU's market -to comply with EU law.
In relation to external differentiation that is achieved by way of international agreement, we present this as a legal process through which a third country is required by the terms of the agreement to adopt EU law or, more commonly, to achieve a specified degree of alignment between EU law and third country law. Although we present legal alignment as a type of external differentiation, we recognize and demonstrate that within this single category, alignment obligations take a multitude of different forms. It is important to note that an obligation for a third country to adopt EU law or to achieve legal alignment with EU law are not mutually exclusive. As will be seen, both types of obligation may be included within a single agreement.
In relation to external differentiation that is achieved by way of an internal EU act, we identify two principal forms. In the first, the EU acts unilaterally to require legal alignment between EU and third country law. We label this legal alignment unilateral external differentiation. In terms of the objective pursued, this has much in common with external differentiation that is achieved by way of an international agreement. In the second form, the EU adopts an act that establishes an obligation that pertains to conduct that takes place outside of the EU (foreign or extraterritorial conduct). In this scenario, we consider that unilateral external differentiation arises when the EU regulator is required, as a matter of law, to take foreign conduct into account when assessing whether a natural or legal person has complied with obligations laid down in EU law. 2 In this second kind of unilateral external differentiation, the EU's goal is to influence the foreign conduct of natural or legal persons rather than the content of third country law. We label this conduct-related unilateral external differentiation.
For the purpose of this discussion, external differentiation will be considered to be unilateral when it results from the adoption of an internal EU act such as a regulation or directive. 3 However, as has been widely observed, unilateralism is a spectrum concept rather than a fixed point on a scale. 4 The degree of unilateralism inherent in a measure does not only depend on its legal form, but also on whether the objective(s) pursued by it, and the standards and enforcement mechanisms included within it, have been drawn up unilaterally or find support in instruments that have been multilaterally agreed. 5 It will be clear on the basis of these definitions that we view external differentiation as a legal phenomenon that is closely tied to the existence of a legal obligation laid down in EU law. It arises when EU law requires alignment between third country and EU law; or where EU law explicitly applies to conduct taking place within a third country. Thus, the boundaries of the concept of external differentiation are determined by the content of EU law.
We do, however, include a discussion of two phenomena which operate outside the margins of external differentiation so defined. These are labelled 'cooperation' and 'participation' in the discussion of external differentiation through international agreements in Part 2 below. We include these in part because they fall within the broader understanding of external differentiation that is adopted by Work Package 6 of this project. Here, external differentiation is defined as the participation of non-EU Member States in EU policies, institutions and programmes. 6 In addition, however, we also consider that cooperation and participation are closely related to our legal conception of external differentiation in that they often pave the way for third country alignment with EU law. 7 Given that Work Package 2 is concerned with the legal feasibility and constitutional acceptability of differentiated integration, it is essential for us to say something about these concepts. We understand legal feasibility to include two dimensions. First, it is concerned with the compatibility of EU measures with sources of law that constrain the EU in the exercise of its competences. We do not include analysis of the legal constraints that EU Member States may face as a result of national law. Second, we also include within the ambit of legal feasibility, considerations relating to the viability of enforcing such measures. As we will see, the issue of enforcement is especially relevant when considering unilateral external differentiation. Constitutional acceptability is concerned with the compatibility of EU measures with EU constitutional norms, including the rules, principles, objectives and values laid down in the EU Treaties, the Charter of Fundamental Rights of the European Union, 8 and in general principles of EU law. 9 3 The legal acts of the Union are defined by Article 288 TFEU. We include those that are binding within the scope of this chapter, namely regulations, directives and decisions. However, we exclude decisions concluding an international agreement in accordance with Article 218 TFEU. Such decisions are inextricably tied to international agreements of the kind that were discussed in Part II of this report. 4 Daniel Bodansky, 'What's so bad about unilateral action to protect the environment? ' (2000) 11 European Journal of International Law 339. 5 This is in accordance with the spectrum developed by Bodansky, ibid. He places national measures enforcing international law norms in the manner contemplated or authorized by international law at the least extreme end of the unilateralism spectrum. At the most extreme end are measures which are not only unilateral in form but unilateral also in terms of the objective pursued and the rules and standards included in the measure. 6 InDivEU, above n.1, p. 36. 7 For example, the DCFTAs with Georgia and Ukraine contain detailed sectoral cooperation clauses that make explicit reference to EU law and thereby set a course towards future alignment. For further detail, see section Part II, Section 3.5. 8 Article 6(1) TEU. 9 The CJEU has characterized the EU Treaties as the EU's constitutional charter since Case 294/83, Parti écologiste Les Verts v. European Parliament, para. 23, ECLI:EU:C:1986:16. The CJEU has also recognized that general principles of law have constitutional status; see e.g Case C-101/08, Audiolux SA e.a v Groupe Bruxelles Lambert SA (GBL) and Others and Bertelsmann AG and Other, para. 63, ECLI:EU:C:2009:626.

Introduction
The concept of external DI has received only modest attention from the legal community. As such, the concept's contours have been shaped primarily by political science. There is therefore a pressing need for a stronger legal orientation to discussions of external DI. In remedying this situation, the aims of this section are twofold. First, to construct an analytical framework for the legal examination of external DI and thereafter map its ranging landscape. Sections 2 and 3 of this part address this aim. Here, the focus is external DI that arises where a third country enters into a binding agreement with the EU, to adopt or align its laws with those of the EU, either to a greater or lesser degree. To this end, it will focus on external DI that is facilitated through EU international agreements. The second aim, addressed in Section 4, is to identify the legal constraints that may restrict the advancement of external DI by way of international agreement. In this regard, Part II identifies how both EU law, including EU constitutional law, as well as international law can constrain external DI. Thereafter, Section 5 draws preliminary conclusions on the continued legal feasibility of external DI.

Analytical framework for the assessment of external DI
To map the different international agreements of the EU, it is necessary to identify suitable parameters around which to do so. In constructing such a framework, the work of Abbott et al. in their seminal article 'The Concept of Legalization' is highly instructive. Their intention was to develop a lens through which the strength of international legal obligations could be assessed. 10 An important distinction is drawn between 'law' and 'legalization'. The former refers to the legal system and the body of legal rules, procedures, discourses, and institutions existing at a point in time. It is the static nature of things as they stand. In contrast, 'legalization' refers to the process of adding to, changing, or subtracting from the body of law and the legal system over time. 11 This latter describes the dynamic process though which law changes over time.
Abbott et al. suggest the process of legalization has three dimensions: (a) obligation: states or other actors are bound by a rule or commitment or by a set of rules or commitments; (b) precision: rules unambiguously define the conduct they require, authorize, or proscribe; and (c) delegation: third parties have been granted authority to implement, interpret, and apply the rules; to resolve disputes; and (possibly) to make further rules. 12 Alternative frameworks have been also been developed elsewhere. 13 Rather than represent binary categories, each dimension should be thought of as a continuum that ranges between weak and strong ideal types.   As explained, external DI is the process through which a third country either adopts EU law or aligns its law with the EU acquis. While comparable as a process to legalization because the law of a third country does change over time, external DI describes the transfer of the law of an international organisation to a third country that is not already Party to that organisation. This is a one-way process that changes the domestic law of non-EU member states. Accordingly, an assessment of external DI entails consideration of the extent and the terms on which a third country is obligated to adopt EU law or align its laws with the EU acquis. This is a different phenomenon to that described by Abbott et al., because their principal interest was the bindingness of international legal rules.
Notwithstanding the above, the analytical framework of Abbott et al. remains highly instructive. Our argument here is that it should be repurposed for the examination of external DI. 14 Specifically, its fundamental components, the distinct dimensions of obligation, precision and delegation, can be tailored to previously identified variables of external DI. Taken together, the modified dimensions provide an analytical lens through which we can examine external DI. As was the case in the original framework of Abbott et al., each tailored dimension represents a continuum of activity rather than a rigid dichotomy. Consequently, they are capable of capturing the extent to which the law of a third country is required to change. The modified versions of each dimension is discussed below.

Obligation (strength + scope)
Abbott et al. define obligation as meaning that 'states or other actors are bound by a rule or commitment or by a set of rules or commitments.' In their comparative assessment of various international legal instruments, they focus on the extent to which a legal obligation is unconditional. 15 In essence, this is determined by the strength of the language found in a particular obligation. For instance, commands such as 'shall incorporate' or 'will ensure' indicate hard legal obligations. In contrast, obligations that explicitly state they are to be non-legally binding or use loose language, such as recommendations or guidelines, will indicate a softer legal obligation. 14 We note that a comparable framework, albeit a non-legal one, is constructed in Sandra Lavenex While a focus on the language is instructive in assessing the strength of single international legal obligations, for external DI, it is not sufficient. In examining external DI, a crucial element is the scope of the acquis that is to be adopted or aligned with. As identified by Stubb, 'matter', defined as the relevant policy area of the acquis, is an important variable of differentiated integration. 16 To this extent, and in contrast to Abbott et al., the assessment of obligation can be expanded to include its 'scope'. For our purposes, scope is taken to refer to the extent to which a third country is required to adopt or align with the acquis. The maximum is adoption or alignment with the entirety of the acquis. At the other end of the spectrum, there is obligation to adopt or align with none of the acquis. The two variables and their accompanying indicators, strength and scope, are outlined displayed in Table 2 below. It is submitted that this tailored version of Abbott et al.'s dimension of obligation is better suited to the process of external DI.

Delegation (Implementation + enforcement)
The final element of legalization referred to by Abbott et al. is delegation. What is meant by delegation is the extent to which states and other actors delegate the implementation and interpretation of an agreement to third parties, namely: courts, arbitrators, and administrative organisations. 18 For the purpose of mapping the EU's international agreements, two dimensions of delegation are identified, as is done by Abbott et al., which are surveillance and enforcement. For our purposes, 'surveillance' refers to the extent to which an independent body is empowered to monitor a third country's alignment with the acquis. Various arrangements can be identified across the EU's international agreements. Enforcement relates to the extent that an independent body can issue binding interpretations of the subject matter of adoption or alignment. A particularly important aspect of this is whether CJEU is empowered to issue binding interpretations of EU law. The spectrum of indicators for both are displayed in Table 4 below.

Mapping external DI
The following section will examine five external DI mechanisms, namely: direct adoption, homogeneity, equivalence, approximation, cooperation and participation. 19 While each mechanism is described in greater detail below, it is necessary to define what is meant by 'mechanism'. In this context, the term refers to the legal process through which a third country is required to either adopt or align with an internal EU law. A focus on mechanisms, rather than individual EU international agreements, which has been the common in previous approaches to this topic, is preferred for a straightforward reason: the aim is to provide an overview of the legal framework of external DI. EU international agreements do not subscribe to a single mechanism, normally they mix and match depending on their aims. Thus, to provide an accurate overview, it is preferable to focus on the mechanisms of external DI. It is argued that the complexities of this process, found in the separate mechanisms, can be adequately captured by reference to the discussed threefold criteria of obligation, precision and delegation. Table 5 below outlines the agreements to which we apply our analytical framework. As outlined in the project proposal, we have included arrangements related to the UK's withdrawal from the EU (both the Withdrawal Agreement and Political Declaration) together with the EU's agreements with the Western Balkan states, among others. These are examined when relevant to one of the external DI mechanism identified above. While remaining committed to an examination focused on legal mechanisms, for the sake of completeness, in Annex I to this report, we have provided an additional table that demonstrates the primary mechanism associated with each category of agreement, the 'centre of gravity', listed in the table in the Annex below.

Direct adoption of secondary law
Direct adoption is the mechanism through which an internal EU law becomes legally valid in a third county's legal system. Consequently, EU law becomes both invocable before and enforceable by the domestic courts of a third country. While transposition of individual EU laws will often be required, the defining characteristic of this mechanism is that it requires the domestic incorporation of EU law into the legal order of a non-member states. 20  With reference to our terms of assessment, the Articles can be described as both unconditional and precise. Use of the wording 'shall be binding' together with 'be, or be made, part of their internal legal order' imposes a clear and unambiguous obligation on the contracting parties. Further, the internal legal effect of the relevant internal EU law is established clearly. 25 In this regard, the ECAA confirms that rights deriving from the ECAA are to be made domestically enforceable and that no domestic legislation is to be adopted unless in accordance with its provisions. 26 The breadth of the obligations with reference to the acquis does differ, which reflects the differing aims of the two agreements. The ECAA establishes a single market in aviation services and its scope extends only to air transport and to the associated matters mentioned in Annex I. 27 Annex I covers primarily aviation matters, such as air traffic management, aviation security and safety, and consumer protection. For each listed EU regulation and directive, Annex I specifies the provisions of EU law that are to be applied by non-EU contracting parties. In contrast, the EEA seeks to apply the four freedoms of the single market to the EFTA states. There are 49 Protocols and 22 Annexes to the EEA that outline what EU law is to be directly adopted by the EFTA states. 28 Its obligation is therefore of a much broader scope than that of the ECAA.
An alternative version of the obligation to directly adopt is found in the EU-UK WA, whose purpose is to provide for an orderly withdrawal of the UK from the EU. The Agreement establishes a 'transition period' (from 31 January 2020 to 31 December 2020) during which the permanent future EU-UK relationship is to be negotiated. 29 Throughout the transition period, Article 127(1) requires that 'Union law shall be applicable to and in the United Kingdom.' The effect of Union law in the UK is well-defined by paragraph (3) of Article 127: Robert Schuman Centre for Advanced Studies Working Papers During the transition period, the Union law applicable pursuant to paragraph 1 shall produce in respect of and in the United Kingdom the same legal effects as those which it produces within the Union and its Member States, and shall be interpreted and applied in accordance with the same methods and general principles as those applicable within the Union.
Article 4 contains the same obligation and adds: 1. …Accordingly, legal or natural persons shall in particular be able to rely directly on the provisions contained or referred to in this Agreement which meet the conditions for direct effect under Union law. 2. The United Kingdom shall ensure compliance with paragraph 1, including as regards the required powers of its judicial and administrative authorities to disapply inconsistent or incompatible domestic provisions, through domestic primary legislation.
Technically, the UK is not adopting EU law but rather maintaining, for the duration of transition period, the previous role played by EU law within its domestic legal system. Nevertheless, the situation also is an example of a third country agreeing, by way of an international agreement, to apply EU law domestically. As was the case with previous examples, the obligation's language is unconditional ('shall be applicable'/'shall produce'). The scope of UK's obligation is conditioned by the definition of 'Union law', which includes primary and secondary EU law as well as EU international agreements. 30 When compared to the ECAA and EEA obligation discussed above, this is clearly broader in terms of its coverage of the acquis. 31 The ECAA and EU-UK WA agreements each establish independent bodies (known as Joint Committees) to supervise compliance by the contracting parties with the incorporated EU law. 32 Decisions by either Joint Committee are legally-binding. 33 The Committees themselves are to be made up representatives of either Contracting Party, which renders both bodies diplomatic forums. 34 For the EU-UK WA, during the transition period, both the CJEU and EU Commission maintain their previous supervisory roles. 35 While the institutional setup of the EEA is slightly more complicated, which the following subsection discusses in greater detail, the supervision of the EFTA states adherence with their EEA obligations is undertaken by the EFTA Surveillance Authority ('ESA'). 36 The role of the ESA mirrors that of the European Commission and these two institutions collaborate closely in their tasks. ESA is able to impose enforceable pecuniary awards for breaches of the EEA. 37 Crucially, in contrast to the previously discussed Joint Committees, the ESA operates independently of the EFTA states and permanently based in Brussels with a staff of approximately 70. Consequently, it can be viewed as representing a stronger form of supervision as it is permanently established body as opposed to an adhoc Joint Committees.
The enforcement of directly adopted EU law varies considerably in each agreement. In the ECAA, the role of enforcement is delegated to the Joint Committee which has four months to resolve the matter. 38  to do so, the parties to the dispute can refer the matter to CJEU for resolution. For the EEA, again, things are slightly more complicated. Disputes concerning decisions of the ESA or conflicts between EFTA states are to be resolved by an independent court known as the EFTA Court. 39 However, disputes concerning the interpretation of the EEA itself are to be resolved by a Joint Committee, a diplomatic body, composed of representatives of the contracting parties. 40 In a similar fashion to the ECAA, contracting parties have the option, where the dispute concerns EEA provisions identical to EU law, to request a ruling from the CJEU. Under the EU-UK WA, the settlement of disputes during the transition period is responsibility of the CJEU. Thereafter, enforcement is delegated to an independent arbitration panel. 41 However, should a dispute concern the interpretation of Union law, as defined above, then the arbitration panel must request a ruling from the CJEU. 42 At first glance, it would appear that each agreement operates a similar form of enforcement in relation to the interpretation of EU law. A diplomatic body, the Joint Committees, attempts to resolve the disputes with option/obligation of referral to CJEU. There are, however, some important differences. First, the decision-making process of each of the Joint Committee differs; ECAA by unanimity (meaning all parties must agree), EEA by consensus (meaning a majority must agree) and EU-UK by mutual consent. Second, perhaps more importantly, is the effect of referral to the CJEU. Under the ECAA, decisions by the CJEU are to be final and binding. 43 This is not the case for the other two agreements. While the EEA makes clear that the Joint Committee should closely follow the case law of the CJEU, which the following section discusses further, it does not make such a ruling binding. 44 Moreover, under the EU-UK WA, referrals to the CJEU are to be remitted back to the established arbitration panel which then proceeds to resolve the dispute in light of the given judgment of the CJEU. 45 The analysis of this mechanism 1 (direct adoption) can be mapped as follows: There are three important themes to pull out of the above mapping: (1) direct adoption ranks as a strong obligation for third countries to adopt EU law which has been used to export parts of the acquis; (2) it is a precise mechanism, both in terms of the EU law to be adopted and the timeframe for adoption; and,

Homogeneity
Homogeneity, as a concept, implies a high degree of alignment. It is typically considered to refer to the uniform application and interpretation of the acquis. 46 Before examining the EU's various arrangements for homogeneity, it is worth distinguishing this mechanism from direct adoption as discussed in the previous section. Direct adoption can be described as a static mechanism; it refers to the process through which a third country immediately adopts EU law into its internal legal order. Direct adoption does not impose an obligation on a third country to ensure that the adopted law remains in alignment with future EU law. In contrast, homogeneity is a dynamic mechanism that facilitates continued alignment with EU law after its initial adoption. It is often paired with direct adoption meaning that the two work in tandem (as in the EEA example). As observed elsewhere, homogeneity may encompass two forms of obligation: (a) legislative -a third country is bound by EU new legislation falling within the scope of the relevant agreement; (b) interpretativea third country must ensure that 'imported' EU law continues to be interpreted in accordance with EU law as it develops. 47 The EU's use of each of these obligations is discussed below.

Legislative homogeneity
Various versions of the obligation to ensure continued alignment with EU law (legislative homogeneity) can be identified. A highly developed example, and perhaps the most famous, is found in the EEA that provides: The aim of this Agreement of association is to promote a continuous and balanced strengthening of trade and economic relations between the Contracting Parties…with a view to creating a homogeneous European Economic Area, hereinafter referred to as the EEA. 48 Thereafter, in a subsequent article, it states: In order to guarantee the legal security and the homogeneity of the EEA, the EEA Joint Committee shall take a decision concerning an amendment of an Annex to this Agreement as closely as possible to the adoption by the Community of the corresponding new Community legislation with a view to permitting a simultaneous application of the latter as well as of the amendments of the Annexes to the Agreement. 49 Two points should be highlighted here. The first is an expressed intention to create a homogenous economic area. This is found in the very first article of the EEA (as well as its preamble). Second, the responsibility for homogeneity falls to the EEA Joint Committee. This is achieved by the Joint Committee amending one of its Annexes to incorporate new Community legislation. As discussed above, once an EU act is listed in an Annex, the EFTA states have an unconditional legal obligation to incorporate it into their internal legal order. The obligation of legislative homogeneity extends across the EEA's 22 Annexes that relate to different parts of the internal market. The obligation imposed is precise as the relevant EU law will be specified in the Annexes, which should be updated shortly after internal EU law is adopted so as to achieve the goal of simultaneous application. In this regard, the EEA states that the 'Community shall, whenever adopting a legislative act on an issue which is governed by this Agreement, as soon as possible inform the other Contracting Parties in the EEA Joint Committee. There is a strong obligation of alignment. With regard to delegation, the comments of Section 3.1 above apply. 51 To date, the ECAA is the only other agreement to explicitly acknowledge homogeneity as a goal. 52 Like the EEA, its Joint Committee is empowered to amend Annex I as necessary which, in turn, places a binding legal obligation on its contracting partiers to incorporate the specified EU law domestically. 53 However, as noted above, the ECAA is of much narrower scope than the EEA. Its precision is of a high degree. That said, in contrast to the EEA, it does not specify how quickly alignment should be achieved with new EU law. Again, the previous findings of Section 3.1 in relation to delegation of the ECAA's obligation of Section 3.1 apply here.
While not explicitly referencing homogeneity, further examples can be identified. The EU-UK WA adopts both a general and specific approach. As explained in the previous section, the UK's obligations, now as a third country, are anchored around the definition of 'Union law'. Close examination of this term reveals it is of a dynamic, as opposed to static, nature, meaning it does not prevent post-signature legislation from falling within its scope. This is confirmed by Article 6(1) that reads 'Union law shall be understood as references to Union law, including as amended or replaced, as applicable on the last day of the transition period.' Separately, Article 36(1) empowers its Joint Committee to amend Part II of Annex I, 'to align it to any act amending or replacing Regulations (EC) No 883/2004 and (EC) No 987/2009 as soon as such act is adopted by the Union.' To this end, the EU is to inform the United Kingdom within the Joint Committee of any act amending or replacing those Regulations. Clearly, there is a degree of dynamism here.
In other areas, the EU-UK WA also provides for the continued alignment, albeit in highly specialised contexts. The first, certainly most controversial, example is the Northern Ireland Protocol. 54 The purpose of the Protocol is to prevent the creation of a hard border between Ireland and Northern Ireland. The effect of the Protocol is Northern Ireland will remain formally within the UK customs territory, while at the same time will be treated de facto as part of the Union's customs zone, and subject to a various portions of internal market regulation. 55 This includes technical regulations for goods, agricultural, environmental production and regulation, and state aid. 56 In these areas, Northern Ireland will need to keep pace with new EU law developments. The Protocol will come into effect at the end of the transition period, with the implementation and enforcement of remaining EU law being monitored by the Commission and CJEU. 57 The second example is citizens' rights. The EU-UK WA, for certain categories of Union and UK citizens, 58 provides a specified group of rights and protections, including the right to continue residing in the host State and the possibility of acquiring permanent residence based on continuous lawful residence. 59 The exercise of both rights is made subject to applicable EU law and therefore represents an example of continued alignment. Finally, the Political Declaration that accompanies the EU-UK WA contains a new innovation in the form of level-playing provisions. In theory, these commit the EU and 51 EEA, Article 93(1)-(2). 52  UK to 'uphold [their] common high standards…at the end of transition period in the areas of state aid, competition, social and employment standards, environment, climate change, and relevant tax matters.' 60 While tricky to place on the spectrum of external DI, the Declaration imposes a very weak obligation on the UK to remain aligned to the EU law.
Alternative external DI arrangements are found in EU-Swiss SA on Schengen and the Energy Community Treaty. The establishment of the former is for the purpose of implementing, applying and developing the Schengen acquis. 61 Annexes A and B of the agreement comprise the Schengen acquis. Article 2(3) provides for Switzerland's dynamic obligation: The acts and measures taken by the European Union and the European Community amending or building upon the provisions referred to in Annexes A and B, to which the procedures set out in this Agreement have been applied, shall also, without prejudice to Article 7, be accepted, implemented and applied by Switzerland. In this context, due account shall be taken of the period of time indicated by Switzerland in the Mixed Committee as being necessary to enable it to fulfil its constitutional requirements.
This obligation can be considered as imposing an unconditional obligation on Switzerland to ensure its continued alignment with EU law, as it develops. As evidenced in the text, it is conditional on the terms of Article 7. This provides that updates to either Annex 'shall enter into force simultaneously for the European Union, the European Community and its Member States concerned and for Switzerland, unless those acts or measures explicitly state otherwise.' 62 However, further subsections clarify that Switzerland retains the right to decide whether or not to implement measures into its internal legal order. 63 Further, the timeline for such a decision will depend on whether a referendum is required. In terms of surveillance and dispute settlement, the agreement's established Mixed Committee is responsible. 64 The latter agreement extends the EU acquis to its contracting parties in four policy areas. 65 Following its specification in each, the agreement provides: 24. For the implementation of this Title, the Energy Community shall adopt Measures adapting the acquis communautaire described in this Title, taking into account both the institutional framework of this Treaty and the specific situation of each of the Contracting Parties.
25. The Energy Community may take measures to implement amendments to the acquis communautaire described in this Title, in line with the evolution of European Community law.
To this end, the agreement established a Ministerial Council that is able to adopt legally binding decisions to adapt the transferred acquis. 66 In contrast to the examined obligation, the ECT's alignment requirement is much more conditional. While previous obligations have stressed that contracting parties shall adopt measures to facilitate alignment, the ECT states that measures may be taken for the purpose of remaining aligned. Surveillance of compliance with the ECT falls to its Secretariat, which is an independent body. 67 Alternatively, dispute settlement is the responsibility of the Ministerial Council, which is comprised of the representatives of contracting parties. 68 Comparatively, the various agreements discussed can be displayed as follows: Interpretative homogeneity The extent of interpretative homogeneity requirements across EU international agreements tend to mirror the extent of a particular agreement's legislative homogeneity requirements. Accordingly, it is the EEA and ECAA that maintain the most detailed rules related to interpretative homogeneity. The relevant EEA provisions are found in Articles 6 and 105. The former provides: Without prejudice to future developments of case-law, the provisions of this Agreement, in so far as they are identical in substance to corresponding rules of the Treaty establishing the European Economic Community and the Treaty establishing the European Coal and Steel Community and to acts adopted in application of these two Treaties, shall, in their implementation and application, be interpreted in conformity with the relevant rulings of the Court of Justice of the European Communities given prior to the date of signature of this Agreement.
The latter states: 1. In order to achieve the objective of the Contracting Parties to arrive at as uniform an interpretation as possible of the provisions of the Agreement and those provisions of Community legislation which are substantially reproduced in the Agreement, the EEA Joint Committee shall act in accordance with this Article. There are a few things to unpack here. Firstly, the articles clearly express the desire that EEA rules be interpreted homogenously (that is to say, in the same way) as internal EU rules. This is made explicit by their respective references to the goal of uniform interpretation or in conformity. Secondly, the scope of obligations are broad and immediate: they apply across the breadth of the agreement, are not limited to a particular set of rules or one policy area, and the rules substantially reproduce internal EU law. Moreover, they are to be effected instantaneously. Finally, and perhaps most importantly, the EEA establishes a detailed surveillance procedure to ensure that the jurisprudence of the EFTA Court and CJEU remains harmonious. The EEA Joint Committee is required to both keep under review the two Courts' jurisprudence and to establish procedures for the exchange of information between the two. Where there are differences between the Courts' jurisprudence, the EEA Joint Committee is to be responsible for its resolution, failing which recourse may be made to the EEA dispute settlement procedures. These procedures were discussed above in Section 3.
The ECAA follows a slightly different path. ECAA rules identical in substance to internal EU rules are to be interpreted in accordance with pre-signature CJEU case law. 69 In contrast, however, postsignature case law of the CJEU is to be communicated to the Contracting Parties and its implications are to be determined by the Joint Committee. 70 Notably, decisions by the EU Joint Committee under this procedure must be in conformity with the CJEU case law. Accordingly, the EU Joint Committee maintains a fundamental role in the surveillance of legislative homogeneity. The ECAA also establishes a procedure through which domestic courts of non-member states may submit questions concerning EU law to the CJEU. 71 Where a domestic court is unable make a request to the CJEU, contracting parties may raise issues of interpretation with the Joint Committee which may act to secure the homogenous interpretation of the agreement. 72 While not striving for homogeneity, the EU-Swiss Schengen Agreement establishes a comparable mechanism. To ensure 'the most uniform possible application and interpretation', the Mixed Committee is to set up a dialogue mechanism for the transmission of CJEU case law to the Swiss Courts. 73 Additionally, Swiss Courts have the right to submit a statement of case or written observations to the CJEU where members states have made a preliminary reference concerning the Schengen acquis. 74 Moreover, Switzerland is to report annually to the Mixed Committee on how its courts have applied the Schengen acquis and, in the event of substantial divergence from the case-law of the CJEU, it may initiate the agreement's dispute settlement procedures. 75 An exception to the above account are the agreements in air transport (where the European Commission and the CJEU have competences in surveillance and arbitration in specified areas); and the Schengen and Dublin association agreements, where new measures within the acquis are to be directly 69 ECAA, Article 16(1). 70  adopted subject to the approval of the Swiss legislature. 76 Consequently, a more 'dynamic' process of alignment is found in relation to Switzerland's alignment with the relevant acquis. 77 Less well-developed provisions are found in other EU agreements. For instance, Articles 94 of the ECT provides: The institutions shall interpret any term or other concept used in this Treaty that is derived from European Community law in conformity with the case law of the Court of Justice or the Court of First Instance of the European Communities. Where no interpretation from those Courts is available, the Ministerial Council shall give guidance in interpreting this Treaty.
Comparably, Article 4 of the EU-UK WA states: 4. The provisions of this Agreement referring to Union law or to concepts or provisions thereof shall in their implementation and application be interpreted in conformity with the relevant case law of the Court of Justice of the European Union handed down before the end of the transition period.
5. In the interpretation and application of this Agreement, the United Kingdom's judicial and administrative authorities shall have due regard to relevant case law of the Court of Justice of the European Union handed down after the end of the transition period. Similar provisions can be found across most categories of EU international agreements. 78 While each of these contains a broad commitment to remain aligned with the future developments in the interpretation of EU law as it develops, there are important differences in relation to the delegation that takes place within these agreements. Those of the EEA, ECAA and EU-Swiss Schengen are accompanied by an institutional setup that facilitates an ongoing dialogue between domestic courts and the CJEU together with surveillance by an independent third body. This is not the case for the latter agreement (i.e. EU-UK WA and ECT) which rely mainly on the work of their respective Joint Committees.

Equivalence
The mechanism of equivalence is mainly used in EU-Swiss sectoral agreements ('SAs'). 79 It is through the SAs that Switzerland's domestic law is aligned to EU law. 80 Through various annexes and appendices, EU law enters the Swiss domestic legal order by Switzerland adopting domestic law that is equivalent to EU law. The SAs are often described as 'static', meaning, in contrast to homogeneity, they do not envisage the adoption of 'post-signature' EU law. 81 Rather than being updated on a 'quasiautomatic' basis, they are renegotiated on a case-by-case basis. 82 The task of updating an agreement normally, although not always, falls to the relevant EC-Swiss Joint Committee. 83 Each SA has a Joint Committee that can make 'technical' adjustments to specified annexes or appendices of an SA, but not the main provisions of the SA. 84 In practice, Switzerland does not normally resist adjustment of the relevant EU acquis. 85 We examine alternative equivalence mechanisms found in two SAs, namely, those for the free movement persons and land transportation. 86 In each agreement, different variations of the equivalence obligation can be identified.
The purpose of the SA Free Movement is to guarantee freedom of establishment for EU citizens and Swiss nationals. It therefore prohibits discrimination against individuals on grounds of nationality and guarantees their right to entry and residence. The agreement, in terms of EU law, is centred around three Annexes. The right of establishment is to be created in accordance with Annex I, which refers to EU law in relation to the interpretation of the right to reside and the public order conditions under which rights of the SA may be restricted. 87 Annexes II and III, respectively, concern the coordination of social security rules and the recognition of professional qualifications. For such EU law references, Article 18 places Switzerland under a general obligation to take all measures necessary to ensure equivalent rights are available domestically: References to Community law 1. In order to attain the objectives pursued by this Agreement, the Contracting Parties shall take all measures necessary to ensure that rights and obligations equivalent to those contained in the legal acts of the European Community to which reference is made are applied in relations between them.
2. Insofar as the application of this Agreement involves concepts of Community law, account shall be taken of the relevant case-law of the Court of Justice of the European Communities prior to the date of its signature. Case-law after that date shall be brought to Switzerland's attention. In addition, Annexes II and III, require the Switzerland 'to apply [the listed EU law]…or rules equivalent to such acts' at the date of signature. 89 EU acts listed in Section B of either Annex are to taken into consideration, and those in Section C of Annex II are to be taken note of. Clearly, the later obligations are of a less binding nature. Switzerland's obligation to achieve equivalence is unconditional. This is indicated paragraph (1) of Article 18 that states 'shall take all measures necessary'. By virtue of nature of the SA, the scope of the obligation is limited to a single policy area. In terms of its precision, Annex I provides clear specification of the EU law for which there must be equivalent Swiss measures.
The equivalence obligation is also of an immediate nature. No transitional provisions are put in place and Switzerland is required to adopt equivalent measures promptly. Paragraph (2) of Article 18 also demonstrates a loose obligation to adhere to subsequent case-law of the CJEU. However, this is not a binding commitment. The agreement's Joint Committee is responsible for the SA's administration, which includes reviewing the impact of relevant Swiss domestic legislation and the settlement of disputes. There is no provision for third-party arbitration. 90 For Annexes II and III, but not Annex I, the Joint Committee is empowered to amend the listed law without having to go through the internal procedures of either contracting party, as is the case for amendments to Annex I. 91 Further, the Joint Committee is responsible for the SA's administration, which includes reviewing the impact of relevant domestic legislation and the settlement of disputes. 92 Importantly, in this latter regard, there is no option of referral to the CJEU for matters concerning the interpretation of EU law.
We now turn to the second version of equivalence found in the SA Land Transport Agreement. The agreement's aim is to develop a coordinated policy for the transportation of goods and passengers by road or rail. 93 Article 52(6) of the agreement establishes an obligation of equivalence for Switzerland: With a view to attaining the objectives set out in this Agreement, the Contracting Parties shall, in accordance with the timetable laid down in Article 49, take all necessary measures to ensure that the rights and obligations equivalent to those contained in the legal instruments of the Community, as listed in Annex 1, are actively applied in the course of their relations.
Annex I is split into five sections each of which lists the relevant provisions of the EU's transport acquis, which, 'in accordance with Article 52(6) of this Agreement, Switzerland shall apply legal provisions equivalent.' 94 Taken together, a clear and unconditional obligation, albeit of a limited scope in terms of the acquis, is imposed on Switzerland to establish equivalent domestic laws to the listed EU law. The strength of this obligation is emphasised by the fact that Switzerland must notify the adoption of domestic legislation in this area to the EU and seek its opinion on such legislation. 95 The precision of the obligation can be rated high: there is clear specification of the EU law for which there must be Swiss equivalents. The timeline for equivalence is not the same for all sections of Annex I. For EU measures contained Section 3, technical conditions related to road transport, Switzerland is not under an immediate obligation to adopt equivalent national laws. Rather, Switzerland has a transitional period of 88 95 SA Land Transport, Article 52(2)-(3). Robert Schuman Centre for Advanced Studies Working Papers two years in which to make its technical arrangements equivalent to those of the EU. 96 For other sections of Annex I, there is an immediate obligation for Switzerland to adopt equivalent national law. In contrast to the SA Free Movement, there is no obligation to stay dynamically aligned with evolving EU law.
Supervision of the agreement falls to its established Joint Committee, which is to be made up of EU and Swiss representatives. 97 The Committee is able to revise Annex I, at the request of a contracting party, 'so as to take account of developments in Community legislation in this area.' 98 This suggests Annex I will be updated to follow EU law and that there is an element of dynamism in the agreement. Decisions of the Joint Committee are not legally binding. Instead, the EU and Switzerland are to carry them out in accordance with their own rules. Leaving the effects of decisions to the contracting parties reflects the SA's overall bent toward reciprocity. Disputes on the interpretation of the SA Land Transport are to be resolved through the Joint Committee. 99 Importantly, there is no recourse to the CJEU for the interpretation of EU law. The consequence is a diplomatic model of dispute resolution.
In comparing the equivalence mechanisms of either SA, the following classifications can be made:

Approximation
Approximation requirements appear frequently in EU international agreements. The term can be compared to harmonisation, which does not regularly feature in EU agreements, in that it requires a third party to align its domestic law with EU law at some point in the future. 100 An important distinction can be drawn between binding and non-binding approximation. Binding approximation requirements are identifiable by the precision of their language and the fact that they delegate authority for their interpretation and implementation. 101 In EU international agreements, they characterised by expressions such as 'shall ensure', 'shall take necessary measures' or 'undertake to authorise'. 102 Non-binding 96 SA Land Transport, Article 7(2). 97 SA Land Transport, Article 51. 98 SA Land Transport, Article 55(2). 99 SA Land Transport, Article 54. 100 It should be noted that the distinction between approximation and harmonisation is quite difficult to decipher. Given the EU Treaties seem to use the term interchangeably (see TFEU, Article 114). What is important to emphasise is that approximation/harmonisation within the EU implies a mutuality or mutual adjustment, whereas approximation/harmonisation in external agreements works only in one direction (i.e. third country approximating to EU rules). 101 Shaffer and Pollack, above n. 13, 714-715. approximation refers to those rules that are not formally binding 103 , which may result from their vagueness or the fact they are not subject to third-party interpretation or implementation. 104 In EU international agreements, they are associated with terminology such as 'shall take the necessary measures in order to gradually achieve', 'shall seek to promote the use of Community regulations', 'will establish a plan' and 'will cooperate in order to align'. 105 Additionally, they do not impose deadlines for the completion of the action in question. The remainder of this section identifies and discusses examples of both forms of approximation.

Binding approximation
Earlier examples of binding approximation are found in Decision 1/95 of the EC-Turkey Association Council. 106 While the Ankara Agreement aimed to establish a customs union between Turkey and the EU 107 , with the possibility of eventual full Turkish membership, it was through subsequent decisions of its Association Council that the customs union materialised. 108 Early decisions of the Association Council covered administrative cooperation and rules of origin for agriculture, further substantive concessions for agriculture, and limited rights for Turkish workers. 109 It was, however, Decision 1/95 that completed the EU-Turkey customs union and, on the basis of which, Turkey must adhere to the Common Commercial Policy and apply the substantive customs acquis. 110 Further, it must 'align itself with the EC preferential customs regime' and apply 'substantially the same commercial policy as the Community' in the textile sector, including the agreements on trade in textiles and clothing. 111 Decision 1/95 of the Association Council imposes a number of binding approximation requirement on Turkey. For example, Article 8(1) provides: 1. Within five years from the date of entry into force of this Decision, Turkey shall incorporate into its internal legal order the Community instruments relating to the removal of technical barriers to trade. Subsequently, Decision 2/97 was adopted by the Association Council and specified the EU measures to be adopted by Turkey. 112 Article 2 clarified that EU regulations were to be made part of Turkey's internal legal order, while it would be free to determine how directives are to be implemented. Decision 2/97 came into effect on 1 January 1997. This example neatly captures how binding approximation requirements can effectively facilitate alignment with EU law. Decision 1/95 imposed an unconditional obligation on Turkey to align with EU law on a future date. In terms of its precision the obligation is lacking as it does not state what EU law is to be adopted, rather it refers broadly to 'technical barriers to trade'. However, it does set a deadline, five years, and delegates determination of the substance of the obligation to the Association Council, which itself has one year to come up with a list of EU law with which there is to be alignment.
A similar approach is found in the area of competition law. The 1970 Protocol established the basis for Turkey's future approximation with EU competition law. 113 However, it left the details to the Association Council. Binding harmonisation requirements are imposed by Article 39 of Decision 1/95, under which Turkey was required to establish domestic competition laws implementing Articles 101 and 102 TFEU and establish a competition authority to implement them. Further, it had one year to adopt EU block exemption regulations together with relevant EU secondary legislation and case law. The Association Council is responsible for monitoring Turkey's approximation and is able to settle disputes; failing such settlement, disputes may be referred to the CJEU or any other court or tribunal. 114 More recent examples of binding harmonisation requirements can be found in the AAs with Deep and Comprehensive Trade Areas ('DCFTAs'). Such agreements have been concluded with Georgia, Moldova and Ukraine. 115 The DCFTA of each agreement represents a standalone Title ('Trade and traderelated matters') with individualised rules. Its purpose is to provide access to the internal market in selected sectors and grant EU investors in those sectors the same regulatory environment in the associated country as in the EU. To this end, binding approximation requirements are established for a wide range of policy areas. 116 For our purposes, what is significant about such rules is their level of precision. For instance, if we take the policy area of public procurement, it can be seen that each agreement establishes highly detailed approximation requirements, composed of three parts, on their respective third countries. The first part requires the third country to submit to the relevant committee 'a comprehensive roadmap' for the approximation of the Union procurement acquis, as specified individual Annexes to each agreement. 117 The Annexes list the time period for alignment with relevant EU law (ranging from 6 month to 8 years) as well as the market access to be granted on completion of approximation. Following a favourable opinion of the relevant Committee, the roadmap will the serve as the reference document for  The final part is in relation to the opening of market access, which provides: 1. The Parties agree that the effective and reciprocal opening of their respective markets shall be attained gradually and simultaneously. During the process of approximation, the extent of the market access mutually granted shall be linked to the progress made in that process as stipulated in [relevant Annex].
2. The decision to proceed to a further phase of market opening shall be made on the basis of an assessment of the quality of the legislation adopted as well as its practical implementation. Such assessment shall be carried out regularly by [relevant committee]. 119 In comparison to the approach of Decision 1/95, it is evident that more recent approximation requirements have developed significantly. In terms of their precision, there is a much clearer expression of what each non-member state must adopt and when. Moreover, they draw an explicit link between the degree of approximation achieved and the degree of access to the internal market. The relevant Association Committees of each agreement are also given a clear mandate in conducting surveillance of approximation. An additional factor is that both the EU-Georgia and -Moldova agreements have standalone chapters for approximation. 120 These impose further obligations on non-member states, including: reporting requirements on efforts to approximate, repealing inconsistent domestic legislation, the terms according to which the EU will assess approximation, and duty to stay dynamically aligned with EU law. In terms of their enforcement, separate agreements also differ slightly. While each agreement establishes somewhat standard dispute settlement procedures that allows for referral to be made to the CJEU, the EU-Georgia and -Moldova agreements preclude their rules on approximation forming the subject of dispute settlement. 121 It is important to note that binding approximation requirements are also found in other types of agreements. For example, the Comprehensive and Enhanced Partnership Agreement with Armenia contains an obligation to approximate its legislation with EU law as specified in accompanying Annexes. 122 As was the case previously, the Annexes list various internal EU law measures that are accompanied by a timetable for implementation. The unconditional nature of this obligation is  123 Moreover, the approximation requirements are of a precise nature. Further, the surveillance authority established by the CEPA, its Partnership Council, is able to update the relevant Annexes to maintain their alignment with EU law and monitor Armenia's approximation progress. 124 Dispute settlement is purely diplomatic, through the Partnership Council, and CJEU referral is available only in selected areas, none of which are relevant to the binding approximation requirements. 125 The examined agreements rank accordingly:

Non-binding (general) approximation
Approximation of a less binding nature can be distinguished from the above on account of its conditional nature, lack of precision and unclear delegation. Some simple examples can be taken from the EU's association agreements with DCFTAs. The EU-Ukraine Agreement, Article 403, provides that '[p]arties shall cooperate to promote agricultural and rural development, in particular through gradual approximation of policies and legislation.' Alternative, the EU-Georgia, Article 47(4), states 'Georgia shall use best endeavours to ensure that its standards body: (a) progressively transposes the corpus of European standards (EN) as national standards.' A more detailed example of a non-binding approximation requirements is found in the EU's Stabilisation and Association Agreements ('SAAs'). The Stabilisation and Association Process is the EU's policy towards the Western Balkan states, and its aim is to stabilise the region and establish a free-trade area with the eventual goal of EU membership for the relevant states. 126 With this in mind, and in contrast to the agreements considered below, the SAAs operate against a backdrop of strict pre-accession conditionality meaning they are subject to stricter enforcement mechanisms outside the agreements themselves. Most SAAs contain the same 'Approximation of laws' obligation:   In most instances, this clause is supplemented by an EU commitment to provide technical assistance to the third country in question. In contrast to binding approximation clauses, what distinguishess these types of approximation clauses, as a mechanism of external DI, is the conditional nature of thee obligations they lay down. They do not impose binding obligations on the third country in question. Rather, they express an intention to move towards approximation at a later date. In this regard, some obligations are more precise than others. While the general clauses of the Euro-Med agreements do not establish a framework, those found in the SAA do contain a greater degree of precision and impose certain minor obligations.

Cooperation
As found in EU international agreements, cooperation clauses are typically non-binding commitments that represent a roadmap for future integration. They can be distinguished from non-binding approximation on the basis that there is not even a stated aspiration to approximate with EU law. On such a basis, one could legitimately question whether such clauses constitute external DI. Here, it is argued that they fall within the scope of external DI because they represent an initial step towards legal transfer for the acquis. However, this is not the case for all of the cooperation clauses surveyed. As will be seen below, there are cooperation clauses that do not offer any concrete prospect of legal transfer. Relevant clauses can be identified by phrases such as: 'shall take the necessary measures in order to gradually achieve', 'shall seek to promote', 'will establish a plan', 'will cooperate in order to'. This section identifies three versions of such clauses, namely: directed, sectoral and general. Additionally, reference is made to institutional cooperation that may be established by specific EU international agreements.

Directed cooperation
The EEA Agreement lists ten policy areas, ranging from social policy to civil protection, as potential avenues for future cooperation. 132 The EEA specifies the six forms that cooperation 'shall normally' take: participation by EFTA States in EU programmes; establishment of joint activities; formal and informal exchange or provision of information; common efforts to encourage certain activities; parallel legislation, where appropriate, of identical or similar content; coordination, where this is of mutual interest. 133 The penultimate form found on the list makes clear that the approximation of legislation in particular fields is a future possibility, although there is no formal obligation to approximate. Additionally, there is an individual EEA Protocol on cooperation in fields outside the four freedoms. This lists various policy areas, such as environment, education and social policy, in relation to which contracting parties have agreed to cooperate. 134 More recent agreements continue to contain detailed cooperation commitments. Association Agreements with DCFTAs provide for wide-ranging cooperation commitments in field such as tourism, civil protection, fisheries and maritime policy, regional development, cultural fields, agriculture and rural development, and industrial and enterprise policy. The commitments in each are structured in a similar way: (a) an expressed aim of cooperation in the policy area, (b) a statement of principles underpinning cooperation, (c) a list of the forms that cooperation may take, and (d) a commitment for the contracting parties to maintain a regular dialogue.

Sectoral cooperation
In certain fields, cooperation commitments will specifically reference EU law. For example, the EU-Georgia and EU-Ukraine agreements set out broad aims for cooperation in education and training: Georgia is required to 'conduct and develop policy consistent with the framework of EU policies and practices with reference to documents in Annex XXXII'; Ukraine 'shall cooperate taking into consideration the provisions of the recommendations listed in Annex XLII to this Agreement'. 135 Each Annex makes reference to recommendations of the European Parliament.
Weaker cooperation requirements are expressed in the SAAs with the Western Balkan states. For instance, in the fields of social policy and education and training, parties are required to cooperate together with a view to upgrading the state of play of their domestic legal systems. 136 In each sector, the requirements go no further than stating that cooperation 'shall take due account of priority areas related to the Community acquis in this field.' For the policy area of environment, loose cooperation requirements are also imposed. These require 'cooperation with the aim of strengthening administrative structures and procedures to ensure strategic planning of environment issues and coordination between relevant actors' that shall focus on the alignment of third-countries with the EU acquis. 137 General cooperation EU international agreements may contain general statements that Parties shall cooperate in specified fields. For example, the EU-UK Political Declaration states: 'Parties should engage in dialogue and exchanges in areas of shared interest, with the view to identifying opportunities to cooperate…such as culture, education science and innovation.' 138 Other examples are found in the Euro-Med agreements that contain individual provisions on cooperation. For social policy, each agreement aims to consolidate cooperation between parties in the field of social policy. 139 Each agreement sets a number of 'priority areas' that include, amongst others, (a) promoting the economic development of women; (b) improving social security, and (c) enhancing healthcare coverage. Again, the Association Council is made responsible for setting up a working party that is to be responsible for the continuous and regular evaluation of the implementation of cooperation efforts. 140  can be identified for both education and training and environment. 141 However, unlike those related to social policy, these are not supposed by the Association Council or a relevant working party. In a similar vein, the PCAs list a broad range of are areas for cooperation and provide for third countries to participate in Community programmes. 142 The EPCAs with Armenia and Kazakhstan upgrade this to 'with a view to promoting the modernisation of the education and training systems in the Republic of Kazakhstan and convergence with policies and practices of the European Union'. 143 However, overall, general cooperation clauses such as these offer only the remotest possibility for legal transfer for the acquis.

Institutional cooperation
An alternative form of cooperation is the EU-Swiss SA for cooperation on competition law. 144 The SA's purpose is to facilitate cooperation between EU and Swiss competition authorities and to prevent conflict between the two. 145 Comparable SAs exist for Eurojust and Europol. 146 The SA for competition law creates a process where the two authorities are able to share information relating to enforcement actions. 147 Pointedly, it rules out the potential for the SA to change the competition laws of either party. 148 Accordingly, the SA does not facilitate external DI. Further, there is no delegation to a supervisory authority or enforcement procedure, with direct consultations serving as the basis of dialogue between parties. 149

Participation
A final port of call is to examine how external DI can be facilitated through the participation of third countries in EU agencies or programmes. Most EU agencies allow for the full participation of third countries, under certain conditions and as provided for by the terms of their establishment. 150 For the majority, this requires the adoption of EU law within the field covered by the agency or of legislation recognised as equivalent to EU law. 151 This section examines the external DI obligations found in three types of participatory arrangements. As was the case for the previous section, it is arguable that that participatory arrangements do not constitute external DI. This is certainly the case for some of  arrangements surveyed and we highlight where this is the case. For other arrangements, however, such clauses can lay the foundation for future legal transfer.

Bilateral international agreements
Here, our focus is on two EU-Swiss SAs, which provide for the participation of Switzerland in the European Environment Agency ('EEA') and European Environment Information and Observation Network ('EEION') and, separately, Creative Europe, and two EU-Turkey bilateral agreements, which enable Turkish participation of the EEA and EEION and European Monitoring Centre for Drugs and Drug Addiction ('EMCDDA'). 152 Both EU-Swiss SAs require application of the relevant EU acquis as specified in their respective Annexes. 153 However, the obligations included in these SAs differ slightly. The SA Environment requires that 'Switzerland…shall apply the acts listed in Annex I'. 154 In return, Switzerland is able to fully participate, without the right to vote, in the Agency Management Board and shall be associated with the work of the Scientific Committee of the Agency. 155 Conversely, the SA Creative Europe provides 'Switzerland shall implement the measures described in Annex I, with a view to completing its legislative framework so as to ensure the required level of compatibility with the acquis communautaire.' 156 In contrast, neither of the EU-Turkey SA agreements noted above require application of the relevant EU law. Additionally, the two EU-Swiss SAs together with the EU-Turkey Environment obligate the establishment/designation of appropriate domestic institutions for participation purposes. 157 The language used in both of the EU-Swiss SAs indicates an unconditional obligation ('shall apply'/'shall implement'). This can be contrasted with non-binding obligations of the EU-Turkey agreements, which require relevant EU law to be taken into account or considered. 158 Clearly, the scope of either EU-Swiss SA is limited to the single pieces of EU law found in their respective Annexes. The separate obligations are therefore distinctly narrow. But within that scope there is a strong external DI obligation. In terms of their precision, the two SAs clearly specify what it is to be adopted by Switzerland. The primary obligation to adopt the listed EU legislation is of an immediate nature. The SAs differ, however, in terms of their precision as it relates to the establishment of domestic Swiss institutions. The SA Environment gives Switzerland six months from the entry into force of the agreement to make relevant designations or establishments. In contrast, the SA Creative Europe does not impose a deadline. Both SAs establish a Joint Committee for the purpose of administrating the agreement and amending any of its annexes. 159 Unlike its counterpart, the SA Creative Europe states explicitly that its Joint Committee is responsible for the settlement of disputes. Further, the Commission is given responsibility for monitoring and evaluating Switzerland's national measures adopted pursuant to the SA Creative Europe. 160 Alternatively, the EU-Turkey Environment leaves implementation to either contracting part to decided and the EU-Turkey EMCDDA remains silent on the issue. 161 Bilateral cooperation agreements That said, EU agencies can propose to third countries that they engage in bilateral cooperation and can establish institutional ties by signing formal working/strategic/technical/operational arrangements with a third country. 162 These forms of working arrangements are often restricted to technical collaboration underlining the relevant agencies' capacity building function. Cooperation can be proposed to a third country only if the relevant EU institutions has given its approval. Currently, Frontex has 20 'working arrangements' with non-EU countries and international organisations. 163 The intended purpose of such arrangements is to establish a structured dialogue between Frontex and the partner country to facilitate the exchange of information and technical expertise. 164 Frontex also has working arrangements with the Commonwealth of Independent States and MARRI Regional Centre. Europol has 17 'operational agreements' with non-EU states and one such agreement with an international organisation (Interpol). 165 Such arrangements exist for CEPOL, Eurojust, EMCDDA and ECDC.
We examine a selection of the such arrangements in relation to the identified three criteria. Working arrangements aim to facilitate cooperation, within the relevant fields, between EU agencies and third countries counterparts. For example, Frontex arrangements aim to facilitate cooperation between the agency and border authorities of third countries. 166 Cooperation agreements pursue similar aims in relation to the working arrangements of both Europol and Eurojust. 167 They do not explicitly obligate a third country to adopt EU law or to align its domestic law with EU law. In relation to our three criteria, working arrangements will score poorly across the board. While convergence may lead to the harmonisation of practices and policies 168 , the working arrangements themselves do not contain any obligation to adopt EU law or align with EU law. On this basis, such arrangements fall outside the scope 159

Participation in EU programmes
The EU has a number of internal programmes, established for the benefit of member states and in order to support internal policies, which can be opened up to third country participation. 172 While, in principle, such programmes are conceived exclusively for member states, the regulations establishing such programmes may allow for third country participation, provided that certain terms and conditions are met. Alternatively, the EU may agree with a third country that it may participate in its programmes. We see this in the EU-UK Political Declaration. Depending on their subject matter, some internal programmes have an external dimension. For example, Horizon 2020 funding (of which this report is a recipient) is open to the participation of specific third countries. 173 To this end, the EU has signed various several agreements that associate third countries to the Horizon 2020 Programme. It has also concluded general cooperation agreements related to the creation of a European Research Area.
This section examines three such Horizon 2020 agreements with Israel, the Faroe Islands and Switzerland. 174 It is necessary to emphasise that the purpose of each agreement is not to extend EU law to a third country. Rather, it is to open access to a specific EU programme for research entities situated in third countries. 175 That said, in the process of doing this, we can observe aspects of EU law touching upon third countries. For example, each agreement establishes procedures for the European Commission, in accordance with relevant EU law, to conduct audits and on-the-spot checks. 176 While the theme of dispute settlement is not addressed, each agreement establishes a committee to monitor their implementation. 177 As noted, the EU has concluded bilateral science and technology agreements with a number of individual countries. 178 These agreements constitute a framework and a privileged forum to identify common interests, priorities, policy dialogue, and the necessary tools for science and technology collaboration. That said, and similarly to bilateral cooperation agreements, such arrangements do not give rise external DI as defined in the introduction to this report. 169 Frontex-Albania, Article 7; Frontex-Belarus, Article 7; Frontex-Nigeria, Article 6.

The legal review of EU external action
The rule of law is a founding value of the EU. 179 Further, the Lisbon Treaty, in setting out the principles of EU external action, confirms its importance as a guiding value. 180 Law plays an important role in how the EU conducts itself on the international stage. This section focuses on the legal limits that may constrain external DI. It adopts a broad understanding of legal feasibility which is taken to mean compatibility with the law that constrains the EU in the exercise of its powers. Accordingly, the object of interest is twofold: (a) the legal limits to external DI that stem from EU law itself or other peripheral sources; and, (b) how such limits may be enforced. EU international agreements have been considered (for a long time) as an integral part of EU law. 181 Article 216(2) TFEU provides that international agreements concluded by the EU are binding on its institutions and Member States, which entails such agreements will have primacy over acts of secondary Community and national law. 182 Thus, when the EU concludes an international agreement, it not only binds itself as a matter of international law but also its Member States as a matter of EU law.
The legal review of EU external action has come primarily from its own Courts: the CJEU and General Court (formerly the Court of First Instance). The Courts can examine EU international agreements under various heads of jurisdiction, which, for our purposes, are threefold. First, before an international agreement is concluded, an agreement may be the subject of an Opinion from the CJEU regarding its compatibility with the Treaties. 183 Second, once an agreement has been concluded and has entered into force, the interpretation of its provisions may form the subject matter of a preliminary ruling. 184 Finally, again post-conclusion, an agreement may be subject to challenge through a direct action for annulment. 185 Other forms of legal review that are less relevant to external DI, which are not addressed here, include claims concerning the EU's non-contractual liability and enforcement actions against Member States. 186 The jurisdiction of EU Courts has impacted (and will undoubtedly continue to) the advancement of the external DI. This has primarily occurred in relation to principle of autonomy and the compatibility of external DI with the EU Charter of Fundamental Rights. Sections 4.2 and 4.3 address each of these limitations.
However, it is not only internal constraints (i.e. those stemming from EU law itself) that may constrain external DI. Rather, constraints may also stem from institutions that are external to the EU. For example, international legal obligations stemming from international treaties or customary international law (CIL) can result in the direct or indirect review of secondary law by EU courts. This form of constraint is more likely to limit unilateral external DI and is therefore examined in that context below. Of greater relevance for our purposes here are international organisations or agreements that establish independent adjudicators to settle disputes arising among Parties, which may include the EU 179  itself. Article 216(1) and 217 TFEU empower the EU to conclude agreements with international organisations that often maintain their own establish independent adjudicators to settle disputes. 187 A well-known example is the WTO's Dispute Settlement Body ('DSB'). The DSB has authority to establish dispute settlement panels and can authorise the suspension of concessions or obligations if its recommendations are not implemented. 188 The EU and its Member States are members of the WTO in their own right, although it is the European Commission speaks for both. GATT Article XXIV(4) authorises WTO members to establish customs unions and free trade agreements provided certain procedural and substantive requirements have been fulfilled. 189 As the EU is a WTO member, external DI agreements must also satisfy such criteria. Article XXIV gives rise to two forms of review of EU external DI agreements. The first is by the Committee on Regional Trade Agreements (CRTA), responsible for determining whether a notified FTA or customs union meets the substantive criteria of Article XXIV. Second, in instances of conflict, WTO Panels may review an agreement for compatibility. A high-profile example involves the EU's Association Agreement with Turkey. This legal limit is considered in Section 4.4.

The autonomy of EU law
The origins of autonomy can be traced to the development of the EU legal order itself. The internal face of autonomy describes the relationship between the EU and its Member States, and the degree to which the EU can exercise independent powers. Externally, autonomy refers to the integrity of EU law and its legal order vis-à-vis other international law and other legal orders. 190 For our purposes, it is the external face of autonomy that is relevant. In numerous Opinions, the CJEU has reviewed the compatibility of proposed EU international agreements with the autonomy of the internal EU legal order. 191 Here, autonomy is used by the CJEU as a mechanism to protect the internal legal order from destabilising effects that may stem from interactions with international law. How might external DI challenge the autonomy of EU law? As outlined, external DI agreements typically establish an external body to monitor their implementation: the Ankara Agreement establishes an Association Council 192 ; operation of the EEA Agreement is monitored by the EFTA Court 193 ; the various EU-Switzerland bilateral agreements create Joint Committees. 194 The operation of external bodies such as these are viewed by the 187 Additionally, Article 220(1) directs the Union to cooperate with specific international organisations: 'The Union shall establish all appropriate forms of cooperation with the organs of the United Nations and its specialised agencies, the Council of Europe, the Organisation for Security and Cooperation in Europe and the Organisation for Economic Cooperation and Development.' 188 The settlement of WTO disputes is governed by Annex 2: Understanding on rules and procedures governing the settlement of disputes. 189 While an equivalent provision is found Article of the GATS, we focus here on the Article XXIV of the GATT because of its longer history. Autonomy may be threatened if an external body can adjudicate on the division of competence, powers or allocation of responsibilities between EU and member states. The Court's concern is that external bodies will affect the balance of power between the EU and its member states. In Opinion 1/91, the CJEU concluded that the EEA Court's capacity to interpret the expression 'Contracting Party' could 'adversely…affect the allocation of responsibilities defined in the Treaties and, hence, the autonomy of the Community legal order.' 195 A further example can be found in the Draft Agreement on the accession of the European Union to the ECHR. Article 3 of the Draft Agreement established a co-respondent mechanism that allowed both the EU and a Member State to become parties to ECtHR proceedings. The mechanism was introduced to prevent the ECtHR from ruling on whether the EU or its Member States is the correct party to a case. 196 Nevertheless, the CJEU found that this mechanism is liable to adversely affect the autonomy of EU law: the ECtHR, in certain circumstances, would need to determine whether the conditions of co-respondency were fulfilled, which, according to the Court, would entail an indirect assessment of competence. 197 An external body may threaten autonomy if it disrupts the preliminary reference procedure between the CJEU and the domestic courts of member states. This criterion played an important role in Opinion 1/19, where the Court considered whether the proposed European and Community Patents Court, with jurisdiction to hear actions related to European and Community patents, was compatible with the EU Treaties. The proposed agreement created a separate preliminary ruling mechanism through which the Patent Court could refer questions of EU law to the CJEU. Crucially, this would prevent national courts from referring the same matter to the Court. In the CJEU's eyes, this threatened the autonomy of EU law as 'the tasks [read preliminary ruling procedure] attributed to the national courts and to the Court of Justice respectively are indispensable to the preservation of the very nature of the law established by the Treaties.' 198 In Opinion 2/13, the CJEU adopted a similar approach to Protocol No. 16 of ECHR. 199 This enables courts and tribunals of the Contracting Parties to request the ECtHR to give advisory opinions on questions of principle relating to the interpretation or application of the rights and freedoms defined in the ECHR. However, given the ECHR would form an integral part of EU law, the CJEU held that the mechanism of Protocol No. 16 could adversely affect the autonomy and effectiveness of the preliminary ruling procedure; specifically, where prior involvement of CJEU would be required, such as when rights guaranteed by the Charter correspond to rights secured by the ECHR. 200 Finally, an external body can threaten the autonomy of EU law through its interpretation of EU law. This is perhaps how autonomy is typically thought of; the safeguarding of the CJEU's role as the final arbiter of EU law. This is underpinned by Article 344 TFEU that requires Member States not to submit a dispute concerning the interpretation or application of EU law to any method of dispute settlement other than those provided for in the Treaties. On multiple occasions, the Court has considered the possibility of an external body pronouncing on EU law to threaten its exclusive jurisdiction. Autonomy was recently considered in Opinion 1/17, where the CJEU examined the compatibility of the Investor Court System (ICS) established by CETA. 202 Contrary to the expectations of most observers, the Court found that autonomy is not infringed. Key to this finding was the constrained jurisdiction given to investment tribunals established under the CETA; Article 8.31(2) provides that a tribunal will not have jurisdiction 'to determine the legality of a measure, alleged to constitute a breach of this Agreement, under the domestic law of a Party.' This was sufficient for the CJEU to conclude that the interpretive jurisdiction of a tribunal was limited to the CETA itself. 203 Coupled with the fact the ICS stood outside the EU judicial systems, a CETA tribunal would not have no jurisdiction to interpret or apply EU law. 204 This discussion is important for agreements facilitating external DI. Such agreements must make provision for external bodies to monitor the implementation and interpretation of the acquis communautaire, the design of which will need to accommodate the discussed elements of autonomy. The approach taken to ICS as found in the CETA suggests a ready-made model is available. That said, the CETA does not qualify as an external DI agreement. 205 For external DI agreements involved in the transportation of EU law, a higher standard for compatibility with autonomy is expected. In such agreements, it is the possibility of an external body interpreting EU law that is most likely to conflict with autonomy. A concerted effort has been made across different agreements to respect the CJEU's role as final arbiter of EU law. The EEA agreement establishes a system of dialogue with EU Courts and allows EFTA states to request an opinion, on a matter of EU law, from the CJEU. 206 The possibility of asking the CJEU for its opinion on the interpretation of EU law is found in more recent external DI agreements with Georgia and Ukraine. 207 From the perspective of legal review, it is clear the CJEU will typically be engaged by way of an Article 218(11) Opinion before an international agreement comes into effect. However, legal review based on autonomy has also occurred post-conclusion of the agreement by way of actions for annulment under Article 263 TFEU 208 or the preliminary reference procedure. 209 The consequences of a negative finding concerning autonomy are severe: an agreement will require renegotiation, as was the case with the original EEA Agreement, or will be abandoned, such as the Draft Accession of the EU to the ECHR.

Compatibility with Charter of Fundamental Rights
The Union is under an obligation to respect fundamental rights as guaranteed by the European Convention on Human Rights, and as they result from the 'constitutional traditions common to the Member States, as general principles of Community law', which extends to its external action. 210 Until relatively recently, the CJEU has appeared reluctant to examine the substantive compatibility of an EU international agreement with EU law. However, in two recent Opinions, given under Article 218(11) TFEU, the Court has expressed a willingness examine whether an agreement complies with the Charter. Robert Schuman Centre for Advanced Studies Working Papers In Opinion 1/15, it considered whether the Agreement on Passenger Name Records (PNR Agreement) complied with Articles 7 (Respect for private and family life) and 8 (Protection of personal data) of the Charter. 211 In Opinion 1/17, it assessed whether Article 47 (Right to an effective remedy and to a fair trial) was undermined by CETA's investor court system. Taken together, the two Opinions demonstrate the willingness of the Court to examine the substance of international agreements in light of the Charter. It is not immediately apparent whether either agreement constitutes an external DI agreement. Previously, we suggested that CETA is not an example of external DI. The same is not the case for the PNR Agreement, which can arguably be considered a weak form external DI agreement, as is explained below.
Directive 95/46/EC aims to protect in the fundamental rights and freedoms of natural persons, and in particular their right to privacy with respect to the processing of personal data. Article 25(6) of the Directive allows the Commission to recognise the level of data protection of a third country as adequate. Previously, it had negotiated with Canada an agreement on the processing of PNR data, which expired in 2009. The PNR Agreement was subsequently negotiated. The Agreement's aim is to limit the processing of PNR data by Canadian authorities so that it accords with Directive 95/46/EC. Notably, the Preamble of the PNR Agreement explicitly references the Charter. While not requiring relevant authorities to align with the EU standards, the PNR Agreement can be viewed as an example of the EU lifting third-country standards to ensure equivalence with its own requirements. Viewed from this perspective, the PNR agreement can be seen as a loose form of external DI. Opinion 1/15 therefore makes clear that the substance of an agreement should be Charter compliant. In terms of legal feasibility, a finding of incompatibility are stark: in Opinion 1/15, the Court held that the PNR Agreement could not be concluded because it was not compliant with the Charter. 212

Compatibility with GATT Article XXIV(4)
The importance of international law is expressed in Article 3(5) TEU: 'In its relations with the wider world, the Union shall uphold and promote its values and interests and contribute to…the strict observance and the development of international law, including respect for the principles of the United Nations Charter.' Subsequent articles of the TEU 213 together with judgments of the CJEU to the effect that international law is an 'integral part of EU law', emphasise the EU's commitment to international law. 214 International legal obligations of the EU may arise from various sources, in particular customary international law and treaty law. As explained above, we consider the former to be of greater relevance to unilateral external DI. Consequently, the focus of this section is how obligations stemming from the EU's membership of the WTO may restrict its capacity to conclude external DI agreements.
GATT Article XXIV provides an exception to the WTO obligation of most-favoured-nation for the formation customs unions and FTAs. 215 The Article contains two obligations for proposed PTAs of WTO members: (a) an obligation not to raise the overall level of protection and make access for products and services from non-participating WTO members more difficult; and (b) an obligation to liberalise substantially all trade among members of the PTA. As noted, both the EU and its member states are party to the WTO. Accordingly, it is against these substantive criteria that EU agreements facilitating

External Differentiated Integration: Legal Feasibility and Constitutional Acceptability
European University Institute 39 external DI may be subject to legal review. Rather than address each of the two substantive criteria, we instead focus on the two forms of legal review to which such agreements may be subject. 216 The first form of review is by the WTO's CRTA. Pre-formation of a customs union or FTA, WTO members must notify the relevant committee. For agreements in goods, members must notify the Committee for Trade in Goods, and, for agreements in services, they must notify the Council on Trade in Services. The EU has notified 43 agreements to WTO, several of which are external DI agreements, including the EEA Agreement and EU-Turkey Association Agreement. 217 On notification, the agreement in question is referred to the CRTA for examination with the substantive requirements of Article XXIV. 218 The CRTA has a broad mandate to 'carry out the examination of [an FTA or customs union]…and thereafter present [a] report to the relevant body for appropriate action' and direct the Committee to 'consider the systemic implications for such [PTAs] and regional initiatives for the multilateral trading system.' 219 While Article XXIV:7(a) appears to require notification (and review) before formation of an agreement, the reality is that most notifications take place once an agreement is in place. Accordingly, review by the CRTA tends to take place post-completion of an agreement. It is open to the CRTA to conclude that an agreement is not compatible with Article XXIV and recommend its modification. 220 Additionally, Article XXIV:7(b) states 'parties shall not maintain or put into force, as the case may be, such agreement if they are not prepared to modify it in accordance with these recommendations.' However, given the CRTA operates according to consensus, a negative determination has not been given.
The second form of review is by WTO panels, subject to appeal on points of law to the Appellate Body. This form of review differs as it can take place only where the formation of an FTA or customs union is used to defend an infringing measure. In those circumstances, the Appellate Body has confirmed that WTO panels will have jurisdiction to assess the overall compatibility of PTAs with the substantive requirements of Article XXIV. 221 The Understanding on Article XXIV confirms the jurisdiction of WTO panels and appellate bodies to review claims involving the formation of FTAs or customs unions. 222 A relevant example involves the EU-Turkey Association Agreement. By virtue of Decision 1/95, Turkey committed to 'align itself with the EC preferential customs regime' and apply 'substantially the same commercial policy at the Community' in the textile sector. 223 Pursuant to these commitments, Turkey imposed quantitative restrictions on imports of Indian textiles. When challenged at the WTO, Turkey sought to justify the quantitative restrictions by way of GATT using Article XXIV. The Appellate Body accepted the Article could justify an inconsistent measure provided Turkey prove (1) the existence of a customs union and (2) the formation of a customs union would be prevented if it were not able to  224 The Appellate Body rejected this argument and Turkey was required to revise its textile arrangements in line with WTO law. 225 The above overview shows there external DI agreements may be subject to two forms of WTO legal review. This review will assess the compatibility of those agreements with the substantive criteria of Article XXIV. This is relevant for those agreements qualifying as either as a customs union or an FTA. Other external DI agreements, such MRAs, will not be subject to such review. The first form, review by the CRTA, will almost certainly be undertaken. However, as discussed above, it would seem unlikely that the CRTA would conclude an external DI agreement to be incompatible with Article XXIV. Given the consensus requirement, the EU would have to vote against the compatibility of its own external DI agreement which is an unlikely scenario. Review by a WTO panel will only arise when another WTO member challenges an external DI agreement. The case of Turkey-Textiles serves as a relevant example. The case demonstrates how the conclusion of an external DI agreement may be affected by a negative WTO decision.

Conclusions
At the outset, Part II of this report had two aims. First, to construct an analytical framework for the legal examination of external DI and thereafter map its ranging landscape. Sections 2 and 3 addressed this aim. Refashioning the work of Abbott et al., Section 2 established a framework situated around three dimensions of (a) obligation, (b) precision and (c) delegation. Each dimension was tailored for the examination of external DI. Section 3 then used this framework to examine and compare the various legal used by the EU in its international agreement to facilitate external DI. The results of the mapping clearly that demonstrated that the EU facilitates external DI through a wide variety of mechanisms that are both fact and scenario dependent. The second aim of Part II was to identify the legal constraints that may restrict the advancement of external DI. Section 4 identified various legal means through which external DI may be restricted. The most important mode of review comes via the EU Courts. As demonstrated, there are various means through which such review can take place, namely: advisory Opinions under Article 218(11) TFEU, direct actions for annulment, and judgments pursuant to preliminary references. The section highlighted various grounds on which these forms of review may be based. Of those highlighted, the principle of autonomy looms large having previously scuppered various EU international agreements. Aside review by EU Courts, the discussion identified the important role of GATT Article XXIV that lays the basis for review by the WTO's CRTA and, when necessary, panels.

Part III: Unilateral external differentiation
As was noted in the introduction to this report, we distinguish here between 'legal alignment' and 'conduct-related' unilateral external differentiation. In this part of the report, we will describe and exemplify these two forms of unilateral external differentiation, before analyzing their legal feasibility and constitutional acceptability. As we also stated in the introduction, external differentiation will be considered to be unilateral when it results from the adoption of an internal EU act such as a regulation or directive; again bearing in mind that unilateralism is a spectrum concept rather than a fixed point on a scale. 226 In order to situate our discussion of unilateral external differentiation in the existing legal literature, it may be helpful to observe that this phenomenon arises when the EU adopts legal acts which are extraterritorial in nature or which give rise to 'territorial extension'. 227

Unilateral external differentiation: legal alignment
Legal alignment unilateral external differentiation arises when EU law establishes a framework for assessing whether third country law is sufficiently aligned with that of the EU. This practice is relatively common in EU law. Examples are found in diverse areas such as banking and financial regulation, 228 data protection, 229 food safety, 230 aviation security, 231 and climate change. 232 As noted above, unilateral measures of this kind promote alignment between EU and third country law. As is sometimes the case with the international agreements discussed above, 233 the concept of 'equivalence' typically provides the benchmark in unilateral acts for assessing whether an appropriate degree of alignment has been achieved. While the precise wording of equivalence clauses included in EU legislation varies, equivalence is normally judged by reference to the capacity of third country law to satisfy particular requirements laid down in EU legislation, or to achieve a level of protection equivalent to that of the EU in relation to a specified objective. Occasionally, EU legislation references a standard other than equivalence to guide the evaluation of third country law. As we will see below, the General Data Protection Regulation (GDPR) permits data transfers to third countries which have been recognized as ensuring an 'adequate level of protection'. 234 226 Article 288 TFEU. Bodansky, above n. 4. We noted there that the degree of unilateralism inherent in a measure does not only depend on its legal form, but also on whether the objective(s) pursued by it, and the standards and enforcement mechanisms included within it, have been drawn up unilaterally or find support in instruments that have been multilaterally agreed. 227 Scott n. 2 above. Legal-alignment unilateral external DI corresponds to the concept of country-level territorial extension, while conduct-related DI corresponds with the concept of transaction-level territorial extension. 228  The importance of a third country demonstrating alignment between EU law and third country law varies between different measures. In its strongest form, measures of this kind make access to the EU market conditional on a third country demonstrating that it has attained the requisite degree of legal alignment. For example, for food products of animal origin to be imported into the EU, a third country must be included on an EU list of authorized third countries. 235 In order to be authorized, a third country must demonstrate that it complies with the requirements laid down in EU law or offers an equivalent animal health guarantee. 236 In making this assessment, account will be taken of the legislation of the third country, as well as the means at the disposal of its services to apply this legislation effectively. 237 More commonly, however, EU legislation will set out different pathways to achieve compliance with EU law, with legal alignment constituting one among several compliance pathways. An example will help to clarify this point. The EU's General Data Protection Regulation (GDPR) permits the transfer of personal data to a third country for processing either where the third country in question has an adequate level of data protection in place (legal alignment); or where the data controller can show in a different way that personal data will be adequately protected by the recipient. In the absence of legal alignment, an adequate level of data protection can achieved by 'appropriate safeguards' or 'binding corporate rules'. 238 However, where alternatives to legal alignment exist within a single instrument, they are frequently burdensome for third country entities seeking access to the EU market. For example, by contrast to an adequacy determination, reliance on certain types of 'appropriate safeguards' in the context of data protection will require authorization of specific data transfers by the competent supervisory authority. 239 Similarly, in banking and financial regulation, EU recognition of the equivalence of EU and third country law will often serve to mitigate the burden on financial institutions of being required to comply with overlapping or conflicting norms. 240 Legal alignment is often recognized by way of an equivalence decision. Such decisions normally take the form of implementing or delegated acts and they are unilateral decisions adopted by the Commission. 241 In adopting an equivalence decision, the Commission is required to follow the procedure laid down in the basic act. However, as we will discuss below in examining the enforcement of EU legislation giving rise to unilateral external differentiation, in practice, such decisions require intensive and continuing cooperation with third country regulatory authorities.

Unilateral external differentiation: conduct-related
Counter-intuitive though it may initially appear, it is relatively common for EU legislation to be drafted or interpreted in a way that renders it applicable to conduct that takes place outside of the EU. We labelled this conduct-related unilateral external differentiation in the introduction to this report. As was noted there, we consider that conduct-related unilateral external differentiation arises when the EU

External Differentiated Integration: Legal Feasibility and Constitutional Acceptability
European University Institute 43 regulator is required, as a matter of law, to take foreign conduct into account when assessing whether a natural or legal person has complied with obligations laid down in EU law. We offer two examples of this kind of unilateral external differentiation below.
First, in 2009 the EU enacted a regulation laying down rules for the killing of animals that are bred for the production of food, wool, skin, fur or other products. 242 This provides that '[a]nimals shall be spared any avoidable pain, distress or suffering during their killing and related operations'. 243 They 'shall only be killed after stunning and the loss of consciousness and sensibility shall be maintained until the death of the animal'. 244 As a consequence, the regulation applies directly to the killing of animals outside of the EU in so far as the animal products are imported into the EU. For an EU competent authority to make an assessment of whether imported products comply with the regulation, it is required, as a matter of law, to ascertain whether foreign conductnamely the manner in which animals have been killed in a third countryis compliant with EU law. Where animals have been killed in a manner that is not compliant, animal products may not lawfully be imported into the EU.
Second, in 2008 the EU enacted a directive including international aviation within the scope of the EU's greenhouse gas emissions (GHG) trading scheme (GHG aviation directive). This required aircraft operators to surrender greenhouse gas (GHG) emission allowances to cover each tonne of CO2 emitted during the course of flights taking off from, or landing at, an EU airport. 245 This requirement was subject to a possible exemption for flights arriving from third countries which had been recognized as having adopted EU-equivalent measures to reduce the climate change impact of flights to the EU. 246 As in the data protection example above, 247 this directive therefore combined two forms of unilateral external differentiation; conduct-related (applying to aircraft operators) and legal alignment (assessing the equivalence of third country law).
On a flight from San Francisco to London, only 9% of the GHGs emitted occur within the territory or airspace of an EU Member State. 248 The rest are generated in the territory or airspace of a third country, or in areas beyond national jurisdiction such as the airspace above the High Seas. While the economic instrument relied upon in this directive does not directly 'regulate' foreign conduct in the same way as the EU regulation on animal killing, 249 it nonetheless falls within the scope of our definition of conduct-related unilateral external differentiation. This is because, subject to the granting of a legal alignment exemption, it would not be possible for an EU competent authority to verify that an aircraft operator has surrendered sufficient GHG emission allowances to comply with the directive without taking foreign conduct in the form of extraterritorial GHG emissions into account. This example has proved to be controversial and the directive in question has formed the subject matter of a validity challenge heard by the European Court. 250 We will return to this example below when addressing issues of legal feasibility and constitutional acceptability. 242 Regulation 1099/2009, above n. 235. 243 Ibid, Article 3(1). 244  There are many examples of conduct-related external differentiation in EU law and they are found across a wide range of different policy areas. 251 We see prominent examples in criminal law, 252 transport, 253 climate change, 254 environmental protection, 255 banking and financial services 256 and competition law. 257 It is almost invariably the case that the EU uses the 'carrot' of market access to encourage actors to demonstrate that foreign conduct is in conformity with EU law.

Legal feasibility
As was noted in Part II of this report, we adopt a broad understanding of legal feasibility to mean compatibility with legal requirements that constrain the EU in the exercise of its power. We are thus interested in (a) the legal limits to external DI that stem from EU law itself, including public international law to the extent that this is binding on the EU and, (b) how such limits may be enforced. As in relation to international agreements, the legal review of EU acts giving rise to unilateral external differentiation has come primarily from the EU Courts, namely the CJEU and the General Court.
In assessing the legal feasibility of unilateral external differentiation, we will examine the relevant case law of the CJEU. This includes preliminary rulings relating to the interpretation or validity of an act of the EU institutions as well as direct actions for annulment (i). We will also explore the possible tension between unilateral external differentiation and enhanced cooperation, as exemplified by the United Kingdom's legal challenge to the Council Decision authorizing enhanced cooperation relation to the financial transaction tax (ii). Finally, we consider possible constraints arising under WTO law (iii) and discuss issues relating to the feasibility of enforcing measures giving rise to unilateral external differentiation (iv). As we noted in the introduction to this report, in so far as the norms constraining unilateral external differentiation are considered to be constitutional in nature, there is an overlap between considerations of legal feasibility and constitutional acceptability.

Unilateral External Differentiation and the CJEU
The CJEU has been willing to review the validity of EU legislation in light of its compatibility with customary international law, including customary international law principles relating to the exercise of legislative competence known as prescriptive jurisdiction. 258  international law requires the exercise of prescriptive jurisdiction to be based on the existence of a 'genuine connection between the subject matter of the jurisdiction and the territorial base or reasonable interests of the state in question [or the EU]'. 259 A genuine connection can be demonstrated by relying on one of the heads of jurisdiction recognized by customary international law. This includes the territorial principle, the personality (nationality) principle, the protective principle and, in narrowly constrained circumstances, universal jurisdiction'. 260 As we will see below, the 'effects doctrine' is also increasingly viewed as a valid basis for exercising prescriptive jurisdiction.
The EU has drawn on a wide range of connecting factors in adopting legislation giving rise to unilateral external differentiation, including territory, nationality and effects. 261 Whenever the Commission adopts a legislative proposal, it must select jurisdictional 'triggers' to define the scope of application of the law. For example, if it selects nationality as a jurisdictional trigger, the law will apply to all EU citizens. If it selects habitual residence as a jurisdictional trigger, the law will apply to a differently circumscribed group, namely all persons who are habitually resident in the EU. When reading EU legislation, it is crucial to pay careful attention to the precise nature and contours of these jurisdictional triggers because they play a key role in distributing the burden of complying with EU law and in allocating rights and other benefits.
Some of the jurisdictional triggers included in EU legislation are well established and uncontroversial. This is the case, for example, when the EU relies on a nationality trigger which serves to bring EU citizens or EU-established corporations within the scope of application of EU law. Likewise, when it relies on a territorial trigger which serves to bring conduct within the EU, or persons present within the EU, within the scope of EU law. Other jurisdictional triggers are more novel and/or controversial and sometimes it may even be difficult to identify which head of jurisdiction they fall under. For example, the EU sometimes includes an 'anti-evasion' trigger which renders EU law applicable where this is necessary or appropriate to prevent evasion of the relevant law; 262 or a counterparty principle which makes EU law applicable to foreign nationals which enter into a contract with an entity that is established within the EU. 263 So far, the CJEU has adopted a permissive stance when confronting the phenomenon of unilateral external differentiation (although the Court has not labelled the phenomenon as such). We will explore the Court's approach below as well as highlighting some areas of continuing uncertainty. Before doing so, however, it is pertinent to observe that notwithstanding its permissive stance, the CJEU has where possible construed measures giving rise to unilateral external differentiation in a manner that is intended to facilitate the peaceful co-existence of EU and third country law. In particular, it has been attentive to the importance of avoiding situations in which natural or legal persons may be faced with having to comply with conflicting legal requirements emanating from two or more legal orders.
For example, notwithstanding the EU's ban on the testing of cosmetics products on animals, products that have been tested on animals in a third country in order to satisfy regulatory requirements imposed by that third country, can still be lawfully imported into the EU so long as the data generated by this animal testing has not been used to demonstrate the safety of the cosmetics products under EU law. In this way, a manufacturer of cosmetics products is not obliged to choose between producing for the EU market or for a third country market.
The CJEU's permissive stance to unilateral external differentiation may be illustrated first by examining case law concerning measures which rely on a territorial connection. In this respect, the CJEU's most famous judgment is in the Air Transportation of America case. 265 Here, in assessing the validity of the EU directive including extraterritorial GHG emissions within the scope of the EU's emissions trading scheme (GHG aviation directive), the CJEU emphasized that the directive only applies to aircraft operators performing a flight that arrives at, or departs from, an EU airport. Therefore, the aircraft in question 'are physically in the territory of one of the Member States of the European Union...[they] are thus subject on that basis to the unlimited jurisdiction of the European Union'. 266 When the CJEU considered that aircraft operators could be required to surrender emissions allowances to cover extraterritorial GHG emissions, it insisted that it is permissible for the EU to make the conduct of a commercial activity within its territory conditional upon operators complying with EU law. It states that this is 'in particular' the case when the objectives pursued by EU law 'follow on from an international agreement to which the European Union is a signatory such as the [UNFCCC] and the Kyoto Protocol'. 267 Further, in justifying the 'full applicability' of EU law within the territory of the EU Member States, even as regards conduct that occurs partly outside the EU, the CJEU observed that the relevant foreign conduct is capable of causing pollution within the EU. 268 There is, therefore, a degree of ambiguity in the CJEU's judgment. Would the existence of a strong territorial connection (planes landing at and taking off from EU airports) have been sufficient to justify the inclusion of foreign conduct within the scope of EU law even in the absence of evidence that the relevant foreign conduct could generate negative effects within the EU? In a recent case, however, the CJEU exhibited a willingness to interpret EU legislation as including foreign conduct within its scope, even in the absence of any suggestion that the conduct in question would cause negative effects within the EU. In the Zuchtvieh case, the CJEU accepted that animal welfare requirements included in EU legislation could apply to animal transportation operations outside the EU, in so far as the journey had commenced within the EU. 269 While the CJEU has adopted a broad interpretation of the territorial principle, it has also more recently recognized the compatibility of the effects doctrine with public international law. While the 4 th Restatement of US Foreign Relations Law states that '[i]nternational law recognizes a state's jurisdiction to prescribe law with respect to conduct that has a substantial effect within its territory', 270 other commentators adopt a position that is more circumspect. Crawford observes, for example, that this doctrine is controversial but 'not objectionable in all cases'. 271 He argues that the validity of the effects doctrine remains uncertain except in relation to certain criminal offences and that this uncertainty is present also in the area of competition law. 272 law], Member State competent authorities may 'accept realistic planning for transport' which 'indicates that the planned transport will safeguard the welfare of the animals at a level equivalent' to EU law. 265 ATAA, above n. 250. 266 Ibid, para. 125. 267 Ibid & para. 128. 268 Ibid, para. 129. 269 Zuchtvieh, above n. 264. 270 4 th Restatement, above n. 260, §409. It notes that the doctrine has been controversial in the past but that it has achieved a wider degree of acceptance over time. It recognizes that some states continue to object to it in particular cases (409(b)). 271 Crawford, above n. 259, p. 447.

European University Institute 47
Be this as it may, the CJEU has endorsed a 'qualified effects test' in the area of EU competition law, 273 upholding the judgment of the General Court that 'the qualified effects test allows the extraterritorial application of EU competition law where it is foreseeable that the [foreign] conduct in question will have an immediate and substantial effect in the European Union'. 274 The European Court had previously favoured an 'implementation test' which permitted the exercise of jurisdiction over anticompetitive agreements that were concluded abroad but implemented within the EU. 275 In endorsing the qualified effects test, the CJEU noted that it is intended to serve the same objective as the implementation test, namely to prevent 'conduct which, while not adopted within the EU, has anticompetitive effects liable to have an impact on the EU market'. 276 Although the CJEU has not yet ruled on the compatibility of the effects doctrine with customary international law in other policy domains, including banking and financial services regulation, it seems highly likely that it would adopt a similarly permissive approach.
Despite these developments, uncertainties remain in defining the outer limits of the EU's competence in adopting measures giving rise to unilateral external differentiation.
First, the CJEU has not yet had an opportunity to rule on the lawfulness of legal alignment unilateral external differentiation, where compliance with EU law depends on the content of third country law. It has nonetheless interpreted legislation containing a legal alignment obligation in a robust and farreaching way. In Schrems, in interpreting the 'adequacy' requirement in the GDPR, the Court stated that while a third country cannot be required to ensure a level of protection which is identical to the EU, it must ensure 'a level of protection of fundamental rights and freedoms that is essentially equivalent to that guaranteed within the European Union'. 277 While 'the means' used by the third country in question to ensure this level of protection 'may differ' from those used by the EU, 'those means must nevertheless prove, in practice, effective in order to ensure protection essentially equivalent to that guaranteed within [the EU]'. 278 Applying this standard, the European Court has annulled both the Commission's 'Safe Harbor' and 'Privacy Shield' Decisions for wrongly concluding that the United States ensures an adequate level of data protection for this purpose. 279 Second, there have been occasions when the EU appears to push at the outer boundaries of established heads of jurisdiction. We see an example of this in the EU's regulation of bankers' bonuses after the financial crisis. 280 Here, the EU sought to limit the bonuses payable to certain categories of staff employed by banks and investment firms. The limitations introduced by the EU were applicable at 'group, parent company and subsidiary level', including in relation to those employed in third country subsidiaries of entities established in the EU. While the personality (nationality) principle permits a state to exercise jurisdiction with respect to 'the conduct, interests, status and relations of its national abroad, which include natural persons and corporations', 281 it is highly unlikely that this extends to the foreign Robert Schuman Centre for Advanced Studies Working Papers incorporated subsidiaries of EU firms. 282 Although the United Kingdom launched an action for annulment in relation to the bankers' bonuses cap, 283 it withdrew its legal challenge following the release of an unsympathetic Advocate General's Opinion. 284 While the Advocate General acknowledged the existence of the personality (nationality) principle in customary international law, he did not explore its scope. 285 Third, doubts have been expressed about the compatibility with customary international law of what may be considered to be novel jurisdictional triggers. We saw a stark illustration of this in relation to the Commission's (unadopted) proposal for a financial transaction tax. 286 Here, the United Kingdom objected to the inclusion of the so-called counterparty principle and issuance principle as heads of jurisdiction for levying the tax 287 . The counterparty principle would have rendered a third country entity entering into a financial transaction with an EU-established entity liable for the tax. 288 The issuance principle would have rendered a third country entity dealing in financial instruments issued within the EU liable for the tax. 289 For example, if a U.S. and Hong Kong bank were to enter into a transaction involving Volkswagen shares, each of these banks would incur liability for the tax. This would remain the case regardless of where in the world the transaction took place.
Although the CJEU has not ruled on the compatibility of these novel jurisdictional triggers with customary international law, 290 , the Commission appears to be walking on shaky ground in its proposal for a transaction tax. The Commission's claim that '[a]s in existing EU legislation in the area of indirect taxes, territoriality principles are fully respected' is not self-evidently true. 291 When a third country entity enters into a transaction outside of the EU with a counterparty established in the EU, or involving a financial instrument issued in the EU, that third country entity is entering into a transaction with a person or property with a territorial connection to the EU. However, the third country entity itself can only be considered to have an indirect territorial connection with the EU.
In relation to its proposal for a financial transaction tax, the Commission laid emphasis on two other elements within it to support its claim that 'territoriality principles' were fully respected. First, it explained that the issuance principle should be seen as a principle of last resort which is intended 'to improve the resilience of the system against relocation'. 292 The Commission noted that this principle would make it 'less advantageous to relocate activities and establishments outside the  290 For reasons we will discuss below, the CJEU did not rule on the lawfulness of relying on these novel jurisdictional triggers. jurisdictions'. 293 That may be so, but the Commission did not explain how or on what basis this should be viewed as somehow strengthening the territorial connection upon which the issuance principle is said to based. Second, the Commission pointed to the fact that the proposal contained a 'general rule' which would have exempted entities from liability under the tax where there is 'no link between the economic substance of the transaction and the territory of any participating Member State'. 294 Not only does this general rule place the burden of proof on the entity seeking exemption, but no further guidance was offered to indicate the kinds of situations in which this exemption might apply None of this is to suggest that there may not be circumstances in which the EU is justified in extending its regulatory or tax jurisdiction to entities and activities outside of the EU. However, when the EU relies on novel jurisdictional triggers, it is incumbent upon it to explain more clearly why it has a legitimate interest in extending its jurisdiction in this way. We saw a positive example of this when the EU decided to extend its clearing obligation to third country entities trading in derivatives. Here, the European Securities and Markets Authority (ESMA) both explained when and why this is necessary and engaged in lengthy and responsive consultations with affected stakeholders. 295 While the EU has not yet adopted the Commission's proposal for a financial transaction tax, 296 the proposal is also pertinent to this report for a second reason. It illustrates a possible tension between unilateral external differentiation and the enhanced cooperation procedure laid down in the TEU. 297 Given the importance of the enhanced cooperation procedure for the InDivEU project as a while, we will exemplify this tension by reference to the proposal for a financial transaction tax in the section below.

Unilateral External Differentiation and Enhanced Cooperation
In areas that do not fall within the EU's exclusive competences, it is open to a group of at least nine Member States to pursue enhanced cooperation. This is subject to authorization by the Council and to the conditions laid down in the Treaties 298 A decision authorizing enhanced cooperation shall only be adopted by Council as a last resort when it has been established that the objectives of such cooperation cannot be achieved within a reasonable period of time by the Union as a whole. 299 Acts adopted in the framework of the enhanced cooperation procedure are only binding on the participating Member States, 300  The Council adopted a decision authorizing enhanced cooperation in the area of financial transaction tax in 2013. 303 This authorized eleven Member States (EU-11) to move forward with the adoption of a financial transaction tax. The United Kingdom, being among the non-participating Member States, challenged the legality of the Council's decision in a direct action before the CJEU. 304 Alongside arguments relating customary international law limits on the exercise of prescriptive jurisdiction, the United Kingdom raised two points concerning the compatibility of the Decision with EU Treaty provisions on enhanced cooperation. First, the UK argued that due to the inclusion of the counterparty and issuance principles, discussed above, the financial transaction tax would apply to institutions, persons and transactions taking place in the territory of non-participating Member States; and consequently adversely affect the competences and rights of non-participating states in a manner that is contrary to Article 327 TFEU. 305 Second, the UK argued that the decision was contrary to Article 332 TFEU because implementation of the tax would entail costs for the non-participating Member States' 306 The European Court dismissed the action on the basis that neither the counterparty principle nor the issuance principle were 'constituent elements' of the Council decision authorizing enhanced cooperation which the United Kingdom had challenged before the Court. 307 Rather the emergence of these effects would be contingent on the adoption by the EU-11 of the Commission's proposal for a financial transaction tax. Although the CJEU did not address the substantive arguments in the case, these arguments raise issues of importance for our discussion of unilateral external differentiation and we will therefore examine them below.
Turning first to the argument that, if adopted, the directive establishing a financial transaction tax would affect the competences, rights or obligations of non-participating Member States: While it is clear that the tax would impose a fiscal burden on financial institutions in non-participating Member States, due to the operation of the counterparty and issuance principles, it is less certain that the directive would affect the competences, rights or obligations of non-participating Member States. After all, the directive establishing a financial transaction tax would neither oblige a non-participating Member State to exercise its fiscal competence by introducing a financial transaction tax, nor would it prevent it from doing so. In formal terms then, the competence of non-participating Member States would be unaffected.
However, if a non-participating Member State were to exercise its fiscal competence to introduce a financial transaction tax, in practice, it is likely that the design of this instrument would be influenced by the EU-11 tax financial transaction tax (if adopted). In particular, the Commission proposal for a financial transaction tax does not provide for double tax relief when the tax is levied on a financial institution in a non-participating Member State. Thus, if a non-participating Member State were to adopt a domestic tax, it would have a strong incentive to include provisions to prevent its financial institutions from incurring double taxation. Given the absence of double tax relief in the Commission's proposal for a financial transaction tax, it would be the non-participating Member State that would suffer a loss of revenue as a result of such provisions. Awareness of this could even influence a non-participating Member State's decision about whether to move forward with adopting a financial transaction tax. All in all, in this example, it is possible to argue that recourse to enhanced cooperation would be liable to affect the competence of a non-participating Member State, if by 'competence' we mean the mode of exercising legal authority rather than its mere existence. This is due to the manner in which the instrument based on enhanced cooperation reaches out beyond the territory of the participating Member States. 303  Moving on to the United Kingdom's second argument which related to the imposition of costs on non-participating Member States, in a manner said to be contrary to Article 332 TFEU: In essence, the argument here was that UK tax authorities could be obliged to collect the tax due by a financial institution established within it, and to transfer the monies to the tax authorities of a participating Member State. 308 This would generate additional costs for the tax authorities of a non-participating Member State. 309 While this point was conceded by then Commission Vice-President Maroš Šefčovič when giving evidence to the House of Lords European Union Committee, he argued that it was a 'minor consideration' and that this expenditure should be considered as primarily a consequence of the mutual recovery regime rather than the proposal for a financial transaction tax. 310 This seems unconvincing given that tax authorities would incur additional expenditure due specifically to the adoption of a measure based on enhanced cooperation. It is not far-fetched to view this additional expenditure as 'resulting from implementation of enhanced cooperation' within the meaning of Article 332 TFEU.
While this second argument is more specific to the tax context of the Commission's proposal, it does nonetheless flow from the fact that the proposal gives rise to unilateral external differentiation and imposes tax liability on institutions based in non-participating Member States. It is only due to this that a non-participating Member State could incur an obligation under existing EU legislation to collect the tax due within its territory and to transfer the proceeds to a participating Member State. This second argument also demonstrates the importance of examining the interaction between the extraterritorial effects of measures based on enhanced cooperation and existing arrangement for administrative cooperation between Member States.

Legal Feasibility: WTO Law
Having examined the jurisdictional constraints that may limit the legal feasibility of unilateral external differentiation, it is necessary to consider whether WTO law imposes legal limits on measures of this kind. The EU and its Member States are Parties to the WTO Agreement which is therefore binding upon them.
Although the 'case law' of the WTO Appellate Body is ambiguous in some important respects, this body has on the whole been willing to accept that it may, in principle, be lawful for Member States to restrict imports due to the manner in which the goods in question have been harvested or produced. 311 This is clear from the famous Shrimp-Turtle case. Here, India, Malaysia and Thailand challenged the WTO-compatibility of a U.S. measure that restricted the importation of shrimp which had been harvested in a manner that posed a threat to endangered species of sea turtles. More particularly, the U.S. measure prohibited the importation into the U.S. of shrimp products that had been harvested in a country that did not meet U.S. standards relating to the use of turtle exclusion devices. These devices are designed to prevent turtles being killed as a 'by-catch' of fishing for shrimp. In this example, had the contested measure been adopted by the EU, we would have considered it as giving rise to legal alignment unilateral external differentiation. It is for this reason that the Appellate Body's report in this case is informative in assessing the WTO-compatibility of measures of this kind. 308 House of Lords, European Union Committee, 'Financial Transaction Tax Alive and Deadly' (7 th Report of Session 2013-2014), para. 18. Due to the principle of joint and several liability established by the proposal, financial institutions within a non-participating Member State could ultimately also be liable for unpaid taxes due by third country counterparties to a relevant transaction. 309 Ibid, para. 37. 310 Ibid. In the Shrimp-Turtle case, the Appellate Body adopted a tolerant stance, accepting that the contested measure was capable in principle of falling within the general exception laid down in GATT, Article XX. It is unfortunate that the clarity of the Appellate Body's report in Shrimp-Turtle is somewhat undermined by its refusal to rule on the issue of whether it is lawful for a WTO Member to enact and apply measures of this kind where they are intended to protect the environment or natural resources outside of the territory of the regulating state. The Appellate Body dodged this question by stating that in the circumstances prevailing in relation to the Shrimp-Turtle case, there was a 'sufficient nexus' between the U.S. and the endangered species of sea turtles; because these species were known to occur in waters over which the U.S. exercises (territorial) jurisdiction. 312 It went on to add that 'it is not claimed that all populations of these species migrate to, or traverse, at one time or another, waters subject to United States jurisdiction '. 313 In the specific circumstances of this case, however, the U.S. measure was found to give rise to unjustifiable and arbitrary discrimination and hence to breach the 'chapeau' of the 'limited and conditional exception' included in GATT, Article XX. 314 For one, the U.S. had failed to enter into serious, across-the-board negotiations with Members exporting shrimp to the United States with a view to entering into a bilateral or multilateral agreement for the protection of sea turtles, prior to imposing a unilateral import ban. 315 In addition, the certification process followed by the U.S. for recognizing that a third country did in fact meet U.S. standards for the protection of sea turtles was described by the Appellate Body as 'singularly informal and casual' and was 'conducted in a manner such that these processes could result in the negation of rights of Members '. 316 Further, in its mode of application, the U.S. measure was not imbued with sufficient flexibility to permit the conditions prevailing in exporting countries to be taken into account, including the adoption by those countries of comparable or equivalent policies or measures for the protection of sea turtles. 317 This suggests that measures giving rise to what we have called legal alignment unilateral external differentiation must be sufficiently flexible to allow for equivalent third country measures to be taken into account, with equivalence being assessed by reference to the effectiveness of the measures in achieving its stated objective. 318 The emphasis the EU places on the concept of equivalence in designing and applying regulatory measures of this kind is therefore not misplaced.
Two more recent reports issued by the Appellate Body lend weight to the contention that it may, in principle, be lawful for WTO Members to enact trade-related measures that make importation conditional on third country conduct. First, in the Tuna-Dolphin labelling case, the Appellate Body examined the lawfulness of a U.S. measure laying conditions for using the 'dolphin-safe' label. 319 One objective of this measure was to contribute to the protection of dolphins, 'by ensuring that the US market is not used to encourage fishing fleets to catch tuna in a manner that adversely affects dolphins'. 320 While the Appellate Body carefully scrutinized the arguments put forward by Mexico claiming that the 312 Report of the WTO Appellate Body, United States -Import Prohibition of Certain Shrimp and Shrimp Products, WT/DS58/AB/R, 12 October 1998, para. 133. 313 Ibid. 314 Ibid, para. 186. 315 Ibid, paras. 166-172 (unjustifiable discrimination) and para. 177 (arbitrary discrimination). 316 Ibid, paras. 181-184. The Appellate Body also found that different phase-in periods for different countries and differential efforts in providing technology transfer gave rise to unjustifiable discrimination between countries, contrary to the Article XX chapeau (see ibid, paras. 174-175). 317  measure was discriminatory and more trade restrictive than necessary to achieve its objectives, 321 at no point did it suggest that it was not open to the U.S. to make market access conditional on whether fishing practices in third countries were sufficiently protective of dolphins swimming in waters outside U.S. control. Indeed, the Appellate Body implicitly rejected Mexico's argument that the measure should be condemned on the basis that it pursued a 'coercive objective' because it sought to 'coerce' Mexico 'to change its practices to comply with the unilateral policy of the United States '. 322 Second, in EC-Seals, the Appellate Body evaluated the compatibility of the EU's seals regulation with the GATT. 323 Subject to a number of exceptions, the contested EU measures prohibited the placing of seal products on the EU market. As in the Shrimp-Turtle case, the AB decided to avoid addressing the question of whether there is an implied jurisdictional (territorial) limitation in the GATT's general exception, and whether it may ever be lawful for a Member to take steps in order to protect the environmental resources of other countries. In EC -Seals, the AB was able to avoid addressing this issue 'of systemic importance' by concluding that '[t]he principal objective of the EU Seal Regime is to address EU public moral concerns regarding seal welfare, while accommodating IC [indigenous communities] and other interests so as to mitigate the impact of the measure on those interests'. 324 The Appellate Body therefore 'territorialized' the EU's objective by focusing on the moral objections of EU citizens and consumers regarding the consumption of seal products rather than on the welfare of the seals themselves. Moreover, the AB adopted a generous reading of the public morals exception included in the GATT's general exception in Article XX(a). For example, it not only recognized that Members 'should be given some scope to define and apply for themselves the concept of public morals according to their own systems and scales of values', 325 but also that there is no obligation for Members to treat similar moral concerns in a similar way. 326 In concluding this discussion about WTO law, it appears that WTO law is capable of accommodating measures giving rise to unilateral external differentiation. 327 However, such measures must be designed and applied with great care to ensure that that they do not discriminate between countries, are not more trade restrictive than is necessary to achieve their objective(s), and that relevant due process conditions have been met.

Legal Feasibility: Enforcing Unilateral External Differentiation
The legal feasibility of unilateral external differentiation depends upon it being possible to enforce EU legislation which makes compliance conditional on third country conduct or the content of third country law. This is at least the case if legal feasibility is understood to imply the enactment of legislation with 'bite' as well as 'bark'. 328 For the purpose of this discussion, the concept of enforcement will be taken to include the authorization of activities taking place within the EU, the monitoring of compliance with EU law, the investigation of possible violations, and where necessary the imposition of penalties to sanction a breach of EU law. 321  A wide range of different actors, operating at different levels of governance, including also private actors, are involved in these different dimensions of enforcing EU law. The identity, role and configuration of these actors in relation to one another varies enormously across different policy domains. Traditionally, at least outside of the area of competition law, responsibility for enforcing EU law has rested principally with the EU Member States, albeit that they have always been subject to oversight by EU authorities including prominently the European Commission. However, today 'European enforcement authorities' play an increasingly important role. 329 This is the case in a number of policy domains in which prominent examples of unilateral external differentiation are found. This includes, for example, banking and financial services regulation, 330 food safety, 331 aviation 332 and maritime safety, 333 pharmaceuticals, 334 and fisheries. 335 Given the highly complex and variegated enforcement landscape in EU law, it is impossible to offer anything approaching a comprehensive overview here. Rather, this section presents four key observations that should be borne in mind when assessing the legal feasibility of enforcing EU measures giving rise to unilateral external differentiation.
First, customary international law prohibits states from taking steps to enforce their laws in the territory of a third country, unless the third country in question has given its consent, whether on an ad hoc basis or as a result of the terms of a treaty. In the absence of consent, enforcement jurisdiction is strictly territorial although '[t]he principle of territoriality is not infringed just because a state takes action within its own borders with respect to acts carried out in another state '. 336 Second, in keeping with this, it is significant that many of the tasks associated with enforcing EU measures giving rise to unilateral external differentiation can be undertaken domestically within the territory of EU Member States. For example, the EU often makes access to its market conditional upon a third country entity being authorized by a Member State or the EU. 337 An entity seeking authorisation may be required to submit information and documentation that can be evaluated 'at home' by regulatory authorities within the EU. Where deficiencies are uncovered, an authorization may be refused, suspended or revoked. Moreover, an entity engaging in commercial activities within the EU will https://www.ema.europa.eu/en/documents/regulatory-procedural-guideline/reflection-paper-ethical-good-clinicalpractice-aspects-clinical-trials-medicinal-products-human-use/european-economic-area-submitted-marketing-autho_en.pdf. 335  Third, even 'at home' enforcement will often necessitate cooperation with third countries. To take an obvious example: While EU legislation sometimes requires or permits Member States to establish criminal jurisdiction over offenses that are committed in a third country, 339 it is not open to them to pursue or arrest the accused in a third country in the absence of that country's consent. Where a third country does not agree to extradite (or expel or deport) an alleged offender to the relevant EU Member State, 340 the Member State's only option is to engage in the 'unsatisfactory procedure' of trial in absentia. 341 It would likewise be impossible for the EU to evaluate whether the law of a third country's regulatory, supervisory and enforcement regime is equivalent to the corresponding EU framework without intensive and time-consuming cooperation with the third country in question. 342 This is true both for the adoption of an initial equivalence decision by the European Commission and in relation to the process of monitoring regulatory and supervisory developments in third countries. Cooperation is based on administrative arrangements that allow the EU authorities to obtain relevant information for the purpose of evaluating and monitoring equivalence.
Fourth, there are, however, clear limits to what 'in territory' enforcement can achieve in relation to unilateral external differentiation. On-site visits in third countries will sometimes be necessary to monitor compliance with EU law and to investigate possible violations.
Such visits may be carried out by Member State or EU regulatory authorities. For example, the European Aviation Safety Agency may decide to conduct a safety audit at a third country operator's facilities where the operator is subject to an operating ban. 343 This is subject to the consent of the operator and the audit may also include an assessment of the oversight of the third country where there is evidence of major deficiencies in the oversight of the applicant. 344 Similarly, National Data Protection Authorities are reported to have conducted on-site audits of data processing facilities in third countries in order to verify compliance with EU data protection law. 345 This includes, for example, a visit by the Italian Data Protection Authority to Google's premises in California. On the other hand, such visits may be carried out by the regulatory authorities of a third country or by private bodies, subject to oversight by Member State or EU regulatory authorities. In this case, the EU performs a meta-regulatory function, laying down standards with which third country authorities or private bodies must conform. 346 Robert Schuman Centre for Advanced Studies Working Papers In her PhD thesis, Jaruprapa Rakpong provides a compelling illustration of the role played by national authorities, serving as EU competent authorities, in ensuring respect for food safety standards in Thailand. 347 These national authorities operated under the close supervision of the EU Food and Veterinary Office (FVO) which carries out inspections in third countries. Rakpong shows how the recommendations put forward by the FVO led to substantial alterations in the institutional infrastructure for food safety in Thailand.
Private actors also play an important role in verifying respect for EU standards in third countries. For example, the role of private certification bodies in ensuring respect for EU sustainability standards in the production of biofuels in third countries is well known. 348 Here, the EU has laid down conditions that 'voluntary schemes' must meet in order to obtain EU approval. 349 A less well known example relates to air carriers flying cargo or mail from a third country into the EU which are required to comply with the 'EU ACC3' programme. 350 These air carriers must ensure that all cargo or mail is physically screened according to EU standards or comes from an EU aviation security validated secure supply chain. EU aviation security validation is carried out by an individual who has received approval from the EU as an EU aviation security validator by the competent authority of the relevant Member State. The EU has laid down conditions that individuals seeking such approval must meet. 351 As these and other examples make clear, that the challenge of enforcing EU law in third countries has spawned a multitude of different institutions and mechanisms. These rely heavily on achieving effective cooperation between EU and third country entities and regulatory authorities. There is a strong incentive within third countries to cooperate where access to the EU market is rendered conditional on demonstrating that third country conduct or third country law meets the standards laid down in EU law.

Constitutional Acceptability
We will turn finally to the question of the constitutional acceptability of unilateral external differentiation, having specific regard to the constitutionalisation of EU external relations under the TEU.
The constitutional framework for EU external relations is constituted in significant part by Articles 3(5) and 21 TEU. 352 In evaluating the constitutional acceptability of unilateral external differentiation, it is important to be aware that this framework is both complex and vague. In light of this, it will be argued that this framework should be viewed first and foremost as providing a set of normative benchmarks to guide the legislature in enacting and justifying measures of this kind. However, there are certain norms included in this framework that may be considered to be sufficiently concrete to serve as benchmarks for substantive judicial review (as opposed to reason giving judicial review).
The constitutional framework constituted by Articles 3(5) and 21 TEU is complex. This is true, in the first instance, in relation to its scope of application. Article 3(5) applies to the EU 'in its relations with the wider world'. While it is not self-evident that this would extend to the adoption of measures giving rise to unilateral external differentiation, the CJEU had no difficulty in invoking this article when reviewing the validity of the EU's directive including extraterritorial GHG emissions within the scope of the emissions trading scheme. 353 Article 21, by contrast, applies to EU external action, such as Common Commercial Policy and Common Foreign and Security Policy, 354 as well as to 'the external aspects of [the EU's] other policies'. Due to this latter phrase, it is clear that it is relevant in assessing the constitutional acceptability of unilateral external differentiation.
This constitutional framework is rendered more complex due to the variety of sources and standards of normativity that are enunciated within these articles. In terms of sources, it includes values, interests, principles and objectives. In terms of standards, the EU's obligations in relation to these different sources varies considerably. The different standards of normativity are summarized in the table below. While recourse to the term 'shall' implies a legal obligation, the nature of the obligation is highly variable. Some obligations clearly take the form of obligations of conduct, 355 particularly the obligation for the EU to promote its values and interests in Article 3(5) and to pursue the objectives in Article 21(5). Other obligations may be considered to be obligations of result, particularly the obligation to uphold its values and interests and to respect the principles in Article 21(1). The EU's obligation to contribute to the protection of its citizens, while an obligation of result, is extremely vague as there is no specification as to the degree of contribution that the EU must achieve. 353  The EU shall pursue the objectives in Article 21(2) As this discussion already suggests, the constitutional framework governing EU external relations is not only complex but also vague. This is true in relation to the substantive content of the EU's values set out in Article 2, and also in relation to both the principles in Article 21(1) and the objectives set out in Article 3(5) and 21(2). There is no guidance about how the EU should prioritize between the different values that it is required to uphold, or the different principles that it is required to respect; or how tradeoffs between values and/or principles, and between the multitude of different objectives, are to be achieved. As Cremona has pointed out in relation to the EU's external policy objectives, these are 'nonteleological, non-prioritsed, open-ended and concerned more with policy orientation than goal setting '. 356 In view of the complexity and vagueness of this constitutional framework, how then should the question of constitutional acceptability be approached in relation to unilateral external differentiation? We suggest that first and foremost this constitutional framework should be regarded as providing reference points according to which EU reason giving, or justification, of unilateral external differentiation should proceed. There are many elements of this constitutional framework that can be used to support or arguably even require interventions of this kind. As many of the examples above make clear, such measures may promote EU values and interests and contribute to protecting EU citizens as required by Article 3(5). 357 They may improve the quality of the environment or consolidate and support the protection of human rights. 358 However, it will almost invariably be the case that in adopting such measures, the EU will have to identify priorities and endorse trade-offs between the many different sources of normativity that have been identified. There is some evidence to suggest that the EU is not always sufficiently attentive to the negative consequences of unilateral external differentiation in third countries; 359 for example as regards the impact of such measures on the eradication of poverty in third countries. 360 Of particular importance in relation to unilateral external differentiation, is the compatibility of unilateral external differentiation with the EU's clearly spelt out obligation to promote an international system based on stronger multilateral cooperation. 361 Here, in terms of reason giving, the EU should be required to demonstrate that it is acting as a 'contingent' unilateralist. 362 As such, the EU should provide evidence that it has pursued good faith efforts at multilateral cooperation prior to the adoption of a unilateral measure; and that the EU's unilateral intervention is provisional pending the emergence of a satisfactory multilateral solution to the problem at hand. Acting as a contingent unilateralist, the EU can quite reasonably claim that there are circumstances in which unilateral action can contribute to the development of international law and can promote an international system based on stronger multilateral cooperation. 363 We see clear examples of contingent unilateralism of this kind in relation to the reduction of GHG emissions from both international aviation and shipping. 364 This does not negate the fact that the EU sometimes faces difficult questions about when the results of multilateral cooperation should be considered sufficiently ambitious to permit the EU to step back from its unilateral intervention. In relation to climate change and international aviation and shipping, the steps taken by the relevant international organization fall well short in terms of their level of ambition by comparison with the 357 For example, the effects doctrine permits the adoption of measures which regulate extraterritorial conduct that has negative effects on the EU's interests and/or protect EU citizens. Examples of unilateral external differentiation which promote protection of human rights in third countries offer examples of measures that promote EU values. See e.g. Regulation 2019/125 concerning trade in certain goods which could be used for capital punishment, torture or other cruel, inhuman or degrading treatment or punishment, http://data.europa.eu/eli/reg/2019/125/oj (torture equipment regulation). 358 E.g ibid torture equipment regulation in relation to human rights and the EU's directive including international aviation in the ETS (above n. 232 ) in relation to climate change/sustainable development. 359  measures adopted by the EU. 365 While the EU continues to apply its Monitoring, Reporting and Verification (MRV) Regulation in relation to shipping, international flights starting or ending their journey outside of the EEA are still not included in the EU-ETS pending a further review of international developments in this area. 366 Although we consider that the constitutional framework established by Articles 3(5) and 21 TEU should be viewed primarily as providing reference points for the EU legislature to justify the adoption of legislation giving rise to unilateral external differentiation, we are also clear that there are circumstances in which it may be possible to point to a substantive breach of this constitutional framework. Such a breach could, in principle, form the basis for judicial review before the CJEU. In view of the complexity and vagueness of the different sources of normativity identified above, this is most likely to be the case when it can be demonstrated that there is an incompatibility between EU law and public international law. This is because the EU is required to contribute to the strict observance of international law, 367 and to respect the principles of public international law. 368 This includes international human rights norms in both customary international law and treaties that are binding on the EU. 369 Consistent with this, as was already seen above in relation to the exercise of prescriptive jurisdiction, the CJEU has been willing to scrutinize the compatibility of measures giving rise to unilateral external differentiation with customary international law. '[T]he European Union is to contribute to the strict observance and development of international law. Consequently, when it adopts an act, it is bound to observe international law in its entirety, including customary international law, which is binding upon the institutions of the European Union'. 370 Moreover, the priority of international agreements over EU secondary law has been recognized both by the TFEU, 371 and by the European Court. 372 However, other than in relation to bilateral trade and Association Agreements, the CJEU has adopted an extremely restrictive attitude to the invocability of multilateral agreements that are binding on the EU as a basis to challenge the validity of an EU act. 373 It has denied this on multiple occasions, either on the basis that the agreement in question is not unconditional or sufficiently clear and precise; or on the basis that the invocability of an agreement is precluded due to its 'nature and broad logic'. 374 In assessing the latter, the CJEU considers whether an agreement establishes 'rules intended to apply directly and immediately to individuals and to confer upon them rights or freedoms capable of being relied upon against States'. 375 In applying these conditions, the CJEU has denied direct effect to many 365 The environmental NGO, Transport & Environment has produced comparisons of the EU measures and the ICAO/IMO important agreements, among them the WTO Agreement, 376 the UN Convention on the Protection of Persons with Disabilities, 377 the Kyoto Protocol 378 and the United Nations Convention on the Law of the Sea. 379 The CJEU's restrictive attitude to the direct effect of international agreements has important implications in assessing the constitutional acceptability of measures giving rise to unilateral external differentiation. This is because in the absence of direct effect such agreements cannot be invoked before the CJEU as a basis for challenging the validity or legality of an EU act giving rise to unilateral external differentiation. We find evidence of this in both the Air Transportation of America case in relation to the Kyoto Protocol and in Intertanko in relation to the UN Convention on the Law of the Sea. In view of the Court's restrictive attitude, it is clear that measures giving rise to unilateral external differentiation can remain valid law even in circumstances in which there may be a strong argument to suggest that they are incompatible with an international agreement that is binding on the EU.
The unwillingness of the CJEU to treat most international agreements as directly effective has been matched over recent years by a similar reticence on the part of the EU legislature. 380 It is increasingly common for international agreements entered into by the EU to expressly exclude their invocability before courts. All in all, these developments do call into question the constitutional acceptability of unilateral external differentiation because it is excessively difficult to enforce international agreements in the EU legal order, thereby making it impossible to ensure that such measures contribute to the strict observance of international law.
The CJEU's approach to the direct effect of international agreements forms part of a broader story about the role of the CJEU as a 'reticent' or 'non-interventionist' court when it comes constraining the policy choices of the EU's political institutions in external relations. 381 More particularly, as Cremona has pointed out, '[t]he Court has not sought to establish priorities among the external objectives expressed in the Treaties....neither has it looked for opportunities to establish fundamental principles to guide external action...[and the Court] places emphasis on the need to preserve room for the exercise of political discretion' on the part of the EU legislature and those negotiating agreements on behalf of the EU. 382 To conclude this discussion: It is our view that the constitutional acceptability of unilateral external differentiation depends in significant part on the adequacy of the reasons put forward by the EU legislature to justify the adoption of measures of this kind. These reasons should be framed according to the different sources of normativity included in the EU's constitutional framework for the conduct of its external relations broadly defined. These sources of normativity are varied and vague and the relationship between the different elements is not clearly defined. The EU legislature is therefore required to play an important role in elaborating and applying this framework. The choices that it makes in doing so should be clearly explained and justified. The adequacy of the reasons given by the EU political institutions should be assessed by the CJEU having regard to the values, principles and objectives laid down in Articles 3(5) and 21 TEU. We would also argue that the constitutional acceptability of unilateral external differentiation would be greatly enhanced if the CJEU were to adopt a less restrictive stance to reviewing the compatibility of EU measures with international agreements. In view of the Court's current stance, there can be little confidence that such measures always contribute to the strict observance of international law. It is unfortunate to conclude in light of this that it is the case law of the CJEU concerning the direct effect of international agreements that poses a fundamental challenge to the constitutional acceptability of unilateral external differentiation.

Part IV: Overall conclusion
Taking our lead from the InDivEU project document, we have adopted an understanding of external differentiation as a legal phenomenon in the sense that it is embedded in the text of EU law. External differentiation arises on the one hand when an international agreement entered into by the EU requires one or more third countries to adopt EU law or to achieve a specified degree of alignment between third country and EU law. It also arises when a unilateral EU act makes the granting of an advantage conditional on legal alignment and/or on 'foreign conduct' complying with EU law. It is a legal phenomenon in the sense that it arises when EU law is written in such a way that it expressly reaches out to include foreign conduct within its scope or to shape the content of third country law. It is because of this that when external differentiation occurs, we consider that EU law may be viewed as applicable in some form within a third country.
Although we have treated external differentiation as a legal phenomenon and evaluated its legal feasibility and constitutional acceptability in light of this, we do not mean to suggest that legal instruments giving rise to external differentiation do not depend on non-legal sources of EU influence or power. For example, when third countries decide to enter into an international agreement with the EU which requires them to align their domestic law with that of the EU, they may be motivated by a whole range of different types of considerations. For example, Lavenex and Schimmelfennig have distinguished between institutionalist, power-based and domestic structure-based explanations of what they term external governance. 383 Gilardi, by contrast, has distinguished between different drivers of norm diffusion: coercion, competition, learning, and emulation. 384 While our study of external differentiation as a legal phenomenon recognizes the existence of a legally distinctive form of external governance which we call external differentiation, it does not seek to address or answer the question of why third countries decide to align their domestic law with EU law, or why actors within third countries are willing to tailor their conduct to ensure compliance with EU law. The question of 'why' external differentiation occurs will be addressed in WP6 ('DI beyond the Member States: External and Regional Differentiation') which forms part of the broader InDivEU project.
Our concept of external differentiation as a legal phenomenon overlaps with closely related concepts which have been developed in academic literature in law and political science. This includes external governance, 385 extended experimentalist governance 386 the Brussels Effect 387 and Market Power Europe. 388 It should not, however, be regarded as coterminous with any of these other concepts. At the risk of repetition, this is because of the emphasis we place on the existence of a legal obligation which is embedded in EU law requiring legal or behavioural change within a third country to ensure conformity with EU law.
Many of our examples of external differentiation can also readily be viewed as examples of the related concepts identified above. However, these other concepts are more broadly drawn than our concept of external differentiation, in that they include instances of third country alignment with EU law which arise even in circumstances when this alignment is not legally required by EU law. For example, the de facto Brussels Effect arises when third country corporations decide to over-comply with 383 Sandra Lavenex & Frank Schimmelfennig, 'EU Rules Beyond EU Borders: Theorising External Governance in EU Robert Schuman Centre for Advanced Studies Working Papers EU law, aiming to achieve conformity with its standards even in circumstances when compliance is not required by EU law. 389 Similarly, extended experimentalist governance may be deemed to arise even when the EU acts as a 'a de facto [as opposed to de jure] global standard setter, peer reviewer, and capacity builder operating along experimentalist lines'. 390 While some examples of extended experimentalist governance are undeniably built on measures that give rise to external differentiation, 391 this is not invariably the case. For instance, although the EU's REACH Regulation concerning chemicals has 'migrated' outside of the EU, and has been viewed as an example of extended experimentalist governance, this measure does not require alignment between EU and third country law as a condition of market access, or explicitly regulate the process of manufacturing chemicals in third countries. 392 Consequently, on our terms, REACH does not give rise to external DI.
There is, however, a very close relationship between our understanding of external differentiation and the concepts of extraterritoriality and territorial extension in EU law. 393 This is because these latter phenomena are also legal in nature and depend for their existence on the content of EU law. More specifically, they arise when EU legislation requires third country law or 'foreign conduct' to be in accordance with EU law. However, the conception of unilateral external differentiation relied on in this report is broader than the concepts of extraterritoriality and territorial extension taken together. This is because these later concepts only include unilateral EU acts (unilateral external DI) and therefore do not include external DI that is achieved through international agreements.
We took our lead in defining external differentiation as a legal phenomenon from the 'Concept and Methodology' section in the InDivEU project document. 394 As noted in the introduction, this considers external differentiation to arise 'when EU rules are legally valid in at least one non-member country for some time'. 395 Due to this emphasis on the concept of the legal validity of EU rules in third countries, we decided to centre our study on circumstances in which EU law explicitly requires legal or behavioural change within a third country as a condition for achieving compliance with EU law. There is, moreover, a sensible fit between this understanding of external differentiation as a legal phenomenon and the research question that we were asked to address; namely, does external differentiation generate concerns relating to legal feasibility or constitutional acceptability? This is because distinct issues of this kind arise when the phenomenon of external differentiation is somehow enshrined in the very fabric of EU law.
In defining the scope of our study, we were however cognizant of the fact that a different understanding of external differentiation is laid down in WP6 which is concerned with external and regional differentiation. 396 As previously noted, external differentiation is defined here as involving he participation of non-EU Member States in EU policies, institutions and programmes. This broader definition was one factor leading us to include 'cooperation' and 'participation' within the scope of our 389 Similarly, external governance includes three different modes of external governance, namely hierarchy, network and market. While there is overlap between both the hierarchy and market modes and our notion of external differentiation, external governance also includes examples that do not involve binding obligations in EU law, requiring legal alignment between third country and EU law and or EU law-compliant foreign conduct. 390  study. It is also notable from the point of view of WP6 that our investigation into external differentiation as a legal phenomenon threw up some interesting examples of situations in which the EU is willing to treat sub-national units within third countries differently. For example, it is open to the EU to recognise that a 'third country, a territory or one or more specified sectors within that third country' ensures an adequate level of data protection, such that transfers of personal data from the EU may take place. 397 While the concept of a 'territory' is not defined, it has been recognised by leading commentators that this is capable of including a sub-national unit within a state. 398 These examples of what may be considered to be territorially differentiated unilateral external differentiation are relevant for the discussion of Brexit in WP6.
In order to facilitate linkages between WP2 ('The legal feasibility and constitutional acceptability of differentiated integration') and WP6 ('DI beyond the Member States: External and Regional Differentiation'), we include a list of examples of territorially differentiated external differentiation in Annex II. We hope that, as the InDivEU project proceeds, additional examples may be identified for inclusion in this list. 397  The European Commission supports the EUI through the European Union budget. This publication reflects the views only of the author(s), and the Commission cannot be held responsible for any use which may be made of the information contained therein.